Video transcript:

Hey, it’s Mark with a super quick video just to let you know that my game Word Play is now available on iPhone and iPad over on the App Store.

So it’s available as a free trial, but if you like what you play, you can upgrade to the full version with a single $5 in-app purchase.

That’s it. No ads, no other IAPs. Just a single “Try Before you Buy” model.

So if you don’t know what Word Play is, it’s my vocabulary roguelike where you try to spell the best word from these tiles, kind of like Scrabble or Boggle.

But when you score enough points, you’ll get to the end of the round, at which point you can pick one of these three perks which have all sorts of crazy, game-changing, rule-bending modifiers to the gameplay, like the ability to add more tiles to the board or upgrade your tiles to golden or emerald tiles. You can change how the scoring works so you get more points if your word starts with a vowel, stuff like that.

I think it’s super fun, it’s really addictive. I still play it quite often when I’m like on the train and whatnot, and so it’s so much better to have it available now on mobile.

Now I’m not completely ruling out an Android version of the game. It’s not too hard to do with Unity. I just need to get an Android device to be able to test it. Let me know if you’d like one in the comments and I’ll see what I can do.

Anyway, that’s it. Don’t want to take up loads of your time. If you like the look of the game, there’s a link to the App Store in the description and a pinned comment.

So, please go check it out. Like I say, it’s free to try, so why not? Word Play is also still available on Steam, and it works really nicely on the Steam Deck if you prefer using one of those.

That’s all I have to say, so thanks for watching and see you soon. Bye!

Every year, thousands of people sign up to the GMTK Game Jam and, somehow, manage to make an entire ass video game in just a few days.

Now - for most developers, that’s that. The jam entry has served its purpose. Perhaps it was simply a chance to practice a new genre or engine or art style. Or it was a welcome break from a longer project. Or maybe the game is a perfectly formed five minute miracle that does everything it needs to do.

But every now and again, the developer thinks - hey, this game is pretty good. This thing has the sauce. And so they develop it further. Add more features. Add more levels. Replace the programmer art with professional assets. And release it as an actual game on Steam.

And you know what? It’s happened a bunch of times now. Like... 100 times. That’s right, there are more than 100 games on Steam that started life as GMTK Game Jam entries. And I want to celebrate them all.

So in this video, let’s look at 100 games that went from GMTK Game Jam entries to full on Steam releases. Ready? Let’s go.

100 Keys to Your Heart is a puzzle platformer where you collect keys to open doors - but the keys are heavy, so you’ve got to leave keys behind on the rack, or pick up keys in the correct order. I’ve already said the word “keys” too much. One thing I can tell you for sure is that it’s 809% bigger than the original jam version, 11 Keys to Your Heart. That’s just maths.

A Little to the Left is a bonafide cozy game classic. I mean, it’s among the top rated games for the tag Cozy, on Steam. That’s amazing! Anyway - in this one, each screen is a miniature messy vignette where you have to intuit how to make the scene tidy again - perhaps sorting by size or colour or shape. It’s chill as hell, it’s the only game in this video with a physical release, and it’s the only game in this video that my wife has completed.

The Baby in Yellow is a Lovecraftian horror game about a nightmare baby. It was a viral smash hit on mobile, so it’s no surprise that the developers are trying their luck on Steam. I would play it - but as a brand new dad, I think a game about a crying, pooping baby who doesn’t sleep... yeah, that’s a bit too close to home.

Blood Typers is a survival horror game where you type words to kill zombies. So, yeah, I know - it’s totally Dreamcast cult classic Typing of the Dead. But actually this game goes much further as you must type to move around the environment, open doors, scavenge and use items, and interact with puzzles. It’s also got four player co-op, if you’ve got some buddies. Still - I don’t think it’s as scary as its original jam prototype, Showtime Massacre. That music creeps me out!

Blueprint Bob is a physics-based construction game where you have to build as tall of a tower as possible, so Bob can climb up to the tippy top and surpass a specific hight. This one made my list of favourite games from the 2024 jam - and now it’s here with new levels, improved gameplay, a level editor, and that classic Zachtronics-style histogram leaderboard. We love a histogram on GMTK. Oooh baby - there’s no better way to visually represent the distribution of quantitative data. Show me a better way and I’ll show you a liar. Uh...

Boomerang X is an arena shooter, ala Devil Daggers, but with - and you’ll never believe this - a boomerang! This game is sick - it’s lightning fast, and you can chuck your boomerang, teleport to its location, spin around, go into slow motion, and take out an enemy from across the room. I bet you can’t do THAT in GTA 6. Anyway, the jam version was cool but hard as nails. The final version, published by the certified sickos at Devolver Digital, is much more fair.

Breezy is a golf game where you play as the wind. Who wouldn’t want to be the wind. The wind is awesome. I haven’t actually played this one and it might suck but I feel bad that it only has one review on Steam so I had to include it out of pity.

Cappy & Tappy: Temples of Peril is a block-stacking platformer where Cappy the Capybara jumps around dusty ol’ tombs, while buddy Tappy the Tapir helps out by dropping massive Tetris blocks into the environment. You can play solo but it’s surely more fun with a friend - especially as you can trap or squash your buddy if you suddenly decide you hate them, actually. Forget them, get new friends. It’s not out yet but there is a demo on Steam.

Cleaning the System is actually still my favourite jam game I’ve ever played on one of my live streams. It’s a stupidly simple game. You play as a stick with a spring on each end. And you can rotate the stick left and right. That’s it. But it leads to a deliciously fun movement system that is effortless to learn, but deviously difficult to master. I love it. This Steam edition adds more mechanics, levels, and even multiplayer and a level editor.

Coasterama is a rollercoaster builder where you have to satisfy all the desires of the tiny guys who’ve paid good money to get into your monochrome theme park. So if some riders want to go fast, others want to go high, and another dude will lose his lunch if the track is too curvy... you’ve got to figure out how to make them all happy. The M. C. Escher-inspired art style is a particularly nice touch.

Deadly Heart Gambit is, according to the Steam description, an Anime Gambling Visual Novel. I don’t think I like any of those things. But, I gotta say I dig these vibes all the same. This one is all about flipping cards, and whoever gets the joker loses. And you’re playing with Alice-Sama, who is very interested in stealing your heart. And I don’t think she means it in like a cool, Persona 5 kinda way. She wants your heart. For real.

Diceomancer is not just a Slay the Spire-alike with a quirky first-person perspective. Because in this one, you can replace any number with the result of a die roll. So suddenly your attack goes from 6 damage to 18 damage... or one damage. It’s all the roll of the dice, baby. YouTube’s favourite egg seemed to dig it. And, also, the trailer opens with the phrase “unexpected trip to fishing”, which... uh... I’m really not sure what that means. But I’m intrigued!

Frogmageddon is pure Vampire Survivors, but with its own twist. And no, the twist is not that it features frogs. That’s not a twist. The twist is that instead of just gaining more power over time, you actually gain more team members - slowly bolstering your army from a loan toad to a massive battalion of bullfrogs. Looks addictive!

Okay, this is gonna take forever. So let’s do a rapid-fire speed round of more games available right now on Steam.

There’s also... the cozy collectathon BunnyFly: Season of Summer, object stacking game Clean Stack, rollercoaster design sim Coaster Crash Course, claymation timeloop Metroidvania Collapse Relapse, cyberspace puzzler Control:Override, Inscryption inspired Death in your Dice, the non-Euclidean Deep Snake, Tetris and Dice mash-up Dicetris, co-op puzzler (Dis)assemble, tricky block connector Dyestributor, roguelike rabbit racer Faster Bunnies, the looping low rez horror game Fragment, cursed twin stick shooter GIMMIKO, train defender Gunslinger Express, absurdist cereal collector IM LOOPY FOR DOOP LOOPS!!!, cosmic clean-up game In Space No One Can Hear You Clean, and the non-stop spinning game Loop my Crank.

Phew. Okay, back to the list.

The Gardener and the Wild Vines is a fun little score-chasing platformer where you leap and slash at flowers to bounce your way up a tower. All so the knight can save his true love - the prince. There are also challenge modes and quirky cutscenes.

Globs is a puzzle game about chunky cubic characters who can resize in different directions to move around, push blocks, make platforms, and more. The pixel art is giving Nitrome and - is that how you use that phrase? It’s giving Nitrome? Do you have to say It’s giving Nitrome vibes, or do you not have to say vibes? Anyway, it’s short, sweet, has immaculate art and music, and is a super cute package on both Steam and mobile.

The Good Overlord is an auto battler bag builder, where you play as the evil overlord of some dusty old dungeon. Your job is to guide the hero to your lair - by placing down baddies and collectibles along the way. I’m not 100% sure why you’re helping the hero instead of snuffing him out as quick as possible - but I’m sure the demo on Steam will explain more. The full game is coming soon.

Happenlance is sort of a Getting Over It style rage game, where your valiant knight uses a massive great lance as a lever to clamber up hills and vault over gaps. But it’s also a jousting sim, where those same kooky controls are used to outmatch foes in online multiplayer. And if you’d rather play with your friends than against them, it’s also got co-op with physics based puzzles. Clever stuff.

Hate Me Not is a role-swapping twin stick shooter. For half of the game, you’re in the black world - protecting and powering up a menacing monster yin yang from incoming enemies. But then you flip to the white world and you’ve got to dodge bullets from that very same yin yang. It’s a clever back-and-forth twist on a classic arcade formula.

How Many Dudes? comes from Butterscotch Shenanigans - the peeps behind Crashlands and Levelhead. And it’s a ridiculous game that asks the question... how many dudes would it take to beat up a gorilla? What about a horse-sized duck AND 10 duck-sized horses? How many dudes would that take? Obviously you can also power up your dudes, unlock perks and synergies, and bend the rules in your favour, making for a game that looks to essentially turn a Reddit thread into a roguelike. This one’s out at the end of July but there’s a demo available right now.

Imagine Sisyphus Happy, goes the saying. Well I don’t need to imagine it, Camus, you big dork. Because now I can play your philosophical thought experiment as a video game which is obviously the BEST medium for such a thing. Anyway - in this one you watch poor Sissy push his boulder up the hill, while you type words, play mini-games, stream your struggle to the masses, and unlock perks and boulders. This is absolutely, 100% what Albert Camus was thinking and I’m sure he has already downloaded the demo on Steam.

Koromeke is described as “the deckbuilder dissected”. It’s a simple board game where you roll dice to do loops around the board - and get upgrades and perks to score more points as you go. That sounds fun enough, but the real treat is the idea that this is some lost game made in 1993, and has been reconstructed in 2026 after decrypting its source code - all suggesting a sort of spooky Inscryption style meta-layer to uncover. It’s out soon but there’s already a demo on Steam.

KoroNeko is, I should say, almost nothing like its game jam prototype Qube - an ugly, fiddly game about rolling dice. A game that didn’t even get into the jam on time. But the developers assure me that this unfinished turd eventually turned into the adorable, cozy, and ultra kawaii puzzle game KoroNeko. And yes, that is the first time I’ve said kawaii on this channel, for anyone keeping track. Add that to your Mark Brown wiki.

Mingle is an adorable hidden object game where you hunt through a crowd of monochrome mates in order to find your exact doppleganger. Which transforms you into a new doodle, so it’s off to find your twin once again. The game’s got a great sense of style, thanks to Kiwi studio Cookiecrayon’s focus on simple shapes and colours. This one isn’t out just yet, but there’s a demo available already.

Molecoole is another Vampire Survivors-esque game, but this time you build your character bit by bit by shoving together different atoms to make a molecule. Each one has unique powers, which means there are millions of possible combinations. And you’ll almost certainly learn nothing about chemistry as you play, which is good because learning is for losers and it’s fine if you fail all your exams and have to become a YouTuber. That’s actually good and fine and cool.

Moonbrella is a bonkers-looking mix of Hollow Knight and Getting Over It - it’s a Metroidvania, but with a miles-deep physics-based movement system where you can use your brolly to pole-vault, grab onto ledges, and float safely over big gaps. Mastering this acrobatic character controller is going to be just as important as puzzling out the layout of the world. It’s got a vague 2026 release date on Steam, but there is a demo you can play right now.

Time for another break, with another lightning round.

There’s also... the dice-rolling dungeon crawler Maces and Dices, time-travelling bullet hell Negative Time, soul-swapping puzzle platformer ODA, picture editing gem PhotoLoop, the dual protagonist puzzler Roku and Rei, die-rollin’ game Roll it to the End, parcel puzzler Scale Mail, sadomasochistic sokoban game Side by Size, tower stacking Skyblocker, tricky seafaring ship sim Sloop, one button scrap dodger Spaceship for Newbies, object estimator Split Happens, first-person adventure game Trigger of Time, time-looping FPS Wait, Is That Me?, slippy car park game Where Is My Parking Spot, genre-bending adventure Willow Guard, the programming puzzle game You Are The Code, and VR cockpit game Ziggy’s Cosmic Adventures.

Deep breath!

Negative Nancy is a visual novel - or as the devs call it, an “interactive sitcom”, where your only way to respond to people is to say “no”. Basically, the opposite of that Jim Carrey movie. The Mask. Anyway - a Steam review sums it up nicely: “How you can manage to make this many outcomes and story branches by choosing between staying quiet or saying no is a mystery to me.” I guess you’ll have to play it and see for yourself.

Neon Shift is a simple, speed running, Geometry Dash-style platformer where you have to sprint through a level as quickly as possible. But then you switch to a phantom version of yourself and have to chase down your previous player before they hit the finish line. Clever stuff. And it won’t be the last time we talk about a game where you play against yourself.

Nested is a grid-based Sokoban-style game, but with an ingenious twist: you play as a set of Russian Nesting Dolls, which means you can stack multiple characters together or split them into separate entities. And you can use their unique shapes and sizes and colours to solve puzzles. I really enjoyed the jam version of this one - it made my list of best games for 2024. And I always felt that this could be even better with more time dedicated to designing puzzles. It’s due for release in August but there’s a demo on Steam right now.

Omelet You Cook is a chaotic cooking roguelike where you have to create the perfect omelet by mashing together the ingredients on a conveyer belt. Each customer has unique tastes and requirements so you’ll have to adjust your recipes to match. This is actually a super popular game with over 1000 overwhelmingly positive reviews on Steam, which is pretty great for a game where you feed a peanut butter and screwdriver omelet to a hungry chicken.

PortalSnake is a mind-bending Sokoban game from the mad genius behind SquishCraft. It’s all about guiding a snake to delicious apples using portals that don’t just teleport the snake through space - but also change its size and rotation. Or even duplicate the snake. It will mess with your bonce. And it has one of the best trailers on Steam. This guy is clearly a marketing genius.

Press Ctrl is a recursive meta adventure where you control a character, that controls another character, that controls another character, that controls another character. The jam version was an honourable mention in my video about 2020’s top games.

Push The Cat with WASD is a Baba Is You-lookin’ puzzler where you move certain letters on the screen - based on specific rules. So the letters W, A, S, and D can move up, left, down, and right - but others can move diagonally. This is all you need to solve a bunch of tricky screens - which are also telling you a meta narrative about kitty cats.

Rollerdrome is... yes, that Rollerdrome. Okay, this one has definitely had the biggest glow up between its jam version and the final release. It started as a fiddly top-down twin-stick for the very first GMTK Game Jam. But the central idea - a game where you do rollerskating tricks to reload your guns - stuck with the designer. So he ended up turning it into a 3D game, getting a publishing deal with Roll7, and making one of the coolest arena shooters of the decade. It’s slick as hell, feels great to play, and has a really compelling scoring system. All told, it totally nails that idea to marry DOOM with Tony Hawk’s Pro Skater. Sadly, mega-publisher Take-Two shut down Roll7 a few years ago, and so I doubt any of the original developers are getting paid if you buy this on Steam. I’m sure there are other ways to get it, mind you.

Scale the Depths is a satisfying fishing game where you catch critters, remove their scales, sell them for cash, buy upgrades, and repeat. The original jam project was one of my picks during the 2024 competition, so it’s wonderful to see it on Steam - and doing so well with over 800 very positive reviews. The developers knew they had something good when Chinese scammers started cloning the game and uploading them to Android. Frustrating, but at least they knew they’d made something special.

Singled Out is a rapid-fire identification game where you’re given a bunch of targets, a basic description of their features, and a few seconds to find and shoot the correct one. It was one of my favourites from the 2019 jam - which is perhaps apt as developer Matt Glanville’s previous game, Switch ‘n’ Shoot, was actually the inspiration for the jam’s theme “Only One”.

Sisyphus Is a Bug is... hey! Our old friend Sisyphus is back. And now in bug form! This one is more of a Metroidbrania - the only way to finally get that dung ball up the mountain is to discover new movement techniques that you’ve always had access to, you just probably didn’t know about them until you find the tutorial someway up the hill. It’s clever and squeezes a lot of game out of only a few buttons.

Slider is a game about moving chunks of the world in order to explore a large space and solve puzzles. Kinda like that dungeon at the end of Skyward Sword, you can open up a sliding block puzzle and move the tiles around to completely reinvent the layout of the world. I believe this is GMTK Game Jam’s first IGF winner, taking home the prize for best Student Game in 2023.

Final break! This time - games that have pages on Steam, but they’re not available and there’s no demo to play yet. They’ve all paid Gabe $100 though, so they’re pretty serious about getting these games on the store at some point.

That includes... Draftula, First Class Rescue, Galactic Diner, Getting Goopy, House Always WINS!, Kickochet, Laser Guy, Last Round at the End of the World, Loaded Dice, Looplash, Mycelium Mayhem, Robot Reckoning, Songrunner, Splatterpus, and Time Travel Shanty

Okay, final stretch!

Sol Cesto is a very popular roguelite that’s sold over 100,000 copies. This one is heavily focused on luck and is all about using magical items and sinister curses to bend the odds in your favour. Now I’ll admit, the link to the GMTK Game Jam is a little tenuous - the game is based on a failed jam entry that was never actually submitted to 2022’s competition. But you can see the first inklings of Sol Cesto in this prototype so god damnit I’m counting it, it’s my video, I don’t care.

Squarepinski is a cozy jigsaw game where every piece is a simple square. So that means instead of relying on the shapes of the tabs and holes, you can only use the pictures. Also, the tiles can be resized and mirrored - which adds a unique challenge. The built-in puzzles are cool, but you can also import your own pictures to generate infinite puzzles. I recommend this one - though I might be a little biased as it’s made by one of the mods on the GMTK Discord! Hi Dotwo!

Swan Song is a cozy... sorry, I need a new word for cozy. I’ve already said it too many times. Swan Song is a peaceful puzzle game about constructing a song that will manipulate the internal gears and springs of a music box - so that a swan can get to the goal. It’s got over 100 puzzles, plus also a nine chapter storyline with voice acting. Move over goose game, it’s all about swans now.

Tape Man is a puzzle platformer about a builder who can use a tape measure to solve puzzles. Not only can your expanding measure be used as a platform, but it can also interact with various elements - in surprising and mysterious ways. You’ll have to figure that out for yourself. It was in June’s Steam Next Fest with a free demo - but expect the full game sometime in 2026.

Tenvader is described by developer Rikerat as Tetris meets Space Invaders. What the heck does that entail? Hey shut up I’ll tell you. So you’re blasting at alien baddies with your lil’ spaceship. And then legally-distinct blocks drop out of the sky - which you can slot into your ship. Each one confers new abilities and powers - giving the game an addictive roguelike loop. It’s got one review, but it sums it up nicely - “an incredibly ambitious game that packs a lot of great ideas into a very cohesive package”.

TetherGeist is a pixel art platformer that’s all about precision and perseverance. It’s definitely inspired by Celeste but has its own core mechanic: namely the ability to eject the hero’s spirit into the astral plane, move it about, and then snap back into the new position. Marry that up with dozens of clever mechanics and enemy types and you’ve got a truly terrific platformer. With an indepth story to boot. Plus, developer O. and Co. games did a long devlog series about evolving this game from a jam entry to a full Steam and console release, which I’d highly recommend watching through.

Threadbound is a puzzle platformer where you can connect two objects in the environment, and have them share the same movement. So now you can shift one block in order to move another object on the other side of the screen. This charming pixel art head scratcher was one of my top picks in the 2021 jam - so it’s great to see that a full version will come to Steam in the future. There’s a demo right now if you want to check out the current progress.

Tiny Traffic is a relaxing turn-based puzzle game about redirecting the flow of traffic. You can tell I’m running out of steam on writing these descriptions because that’s literally what it says on the game’s store page. But anyway - it’s a clever take on this familiar set-and-forget puzzle design. And I feel like I definitely played the jam version during a stream, maybe? I’ve played a LOT of jam games at this point. Well over 1000. Where’s my congressional medal, Obama?

Two-Timin’ Towers is a clever twist on the tower defence game, with the simple change that your character runs around the battlefield... and your towers have friendly fire. So that means the more defensive structures you place down, and the more bullets they spew out, the more difficult it is to stay alive. I often use it as a go-to example of taking a familiar genre and doing something unexpected. Sadly this one has been stuck in “coming soon” mode for years, but there is a demo on Steam at least.

UvsU: You vs. You - the game so nice they named it twice - is the next game from dietzribi - developers of the genre-bending puzzle game Toodee and Topdee. This one is equally clever: you must finish the level. Then play it again as the villain, killing your previous self. Then play it again as the hero, avoiding the villain. And so on. A recursive self-defeating spiral of death. All with a funky claymation art style. There’s a demo on Steam right now.

Word Factori is a simplified Zachtronics game, where you create a production line to spell words. Only one factory can actually make a letter: the letter I. The rest can bend, flip, and combine that shape to make the other letters. So if you want to spell out the word cat, you need to make a complex mega factory with dozens of processing steps. And best of all: it’s got a histogram! We’ve come full circle!

Word Play is... hey! That’s my game! Okay this one is a bit cheeky as yes, I am promoting my own game. And also it wasn’t in the official GMTK Game Jam, but the secret Patreon-only jam for my Discord community. Which means I’m promoting that too. But as previously established: my video, my rules. BYE.

And finally, Wrangle Ranch is a roguelike about lassoing animals on a farm. And then, in classic Balatro style, you can get upgrades to score more points, more quickly. It’s a cute and fun game, but then YouTuber and developer BenBonk said he made minus 200 dollars on the game which, uh, I feel partly responsible for, as the host of the game jam. Sorry about that, Ben!

And so there we have it. One hundred games that went from jam to Steam.

If you want to find out more, I’ve got an ever-growing list on Backloggd where you can see all the games that have made the jump from the jam.

And so the final thing I have to ask is... who’s next? Who’s the next person who will discover a great game idea during the jam - and turn it into a game that makes millions of dollars, or wins a BAFTA, or gets released on Switch, or is just a cool ass video game that people enjoy. Could it be you?

There’s only one way to find out. GMTK Game Jam 2026 starts on July 22nd. I’ll see you there.

Sources, comments, and more on YouTube

Video transcript:

Hey, it’s Mark.

Now, I don’t do a lot of videos about programming on this channel - mostly because I have no idea what i’m doing in that regard. And the codebases for my two published games - Mind Over Magnet and Word Play - are held together with, like, sticks and glue.

But! I am really proud of one aspect of Word Play’s programming - and that’s the way I implemented the game’s modifiers.

So - for context. Word play is a vocabulary roguelike where you spell words to score points. And then you can unlock perks that let you bend the rules and score even bigger.

These perks come in three flavours. Gifts typically add more letters to your bag. Upgrades let you change and improve your tiles. And modifiers are the passive perks that give big benefits under the right conditions.

Like... this one that doubles your score if the word has the same first and last letter.

So they’re basically the relics from Slay the Spire or the Jokers from Balatro.

Now when I first started making the game I had no idea how I would implement these modifiers. There wasn’t an obvious and easy way to do it. But, ultimately, I came up with a system that I’m really happy with. One that fits the requirements for the system, can be easily scaled to fit all 120 modifiers in the game, and makes it effortless to test, change, and debug the lot of them.

So - I’m hoping this might prove interesting or inspirational to, let’s say, intermediate level programmers, to show what goes into making a system for a live video game. Sound good? Okay, then let’s jump in.

A quick note that I’ll be talking about it in the context of making the game in Unity, because that’s what I used. But the basic idea should apply to pretty much any engine.

Okay, so the modifiers in Word Play touch so many aspects of the game, with modifiers that fiddle with refreshes and bonus rounds and special tiles and the basic rules of spelling.

And I’ll come to that - but for now, let’s make things simple by looking at the ones that impact the way your word is scored. Like - the modifier “If the first letter is a Vowel, multiply the Final Score by 1.5”.

How would we implement that?

Well, first I’ll write some code to tally up the word’s initial score by adding up the points on all of the submitted tiles.

And then... well, in Word Play all of the modifiers are separate GameObjects. And I have a list of which ones the player currently owns. So we could check if the players owns the modifier “VowelFirst”. And, also, if the first letter is a vowel.

If both are true... multiply the word’s score by 1.5.

And that would work. It would work fine. In fact, that’s exactly how I did it for the scrappy game jam prototype of Word Play. And it did the job,

But when it came time to turn this into a real video game, I knew I needed a better system. Because... well, a couple reasons.

The big one: every time we add a modifier, this script gets longer and longer. A massive spaghetti mess of if statements. And that makes it harder to find things, change things, and fix bugs and problems. Like, what if I actually want it to multiply the score if the LAST letter is a vowel? Now i’ve got to hunt through all this code to find that. And what if I want to change something about how modifiers work in general - I might end up having to change the same code in 100 different if statements. Yikes.

and another problem is that the order these modifiers trigger is defined by the order they show up in the code, not the order they are in your modifier list. Now this didn’t actually end up being a big problem for Word Play - you almost never need to worry about the order of your modifiers. But it’s a huge deal in Balatro, and so - with that game serving as my initial inspiration - it was a requirement when I was first coding up this system.

Okay. So my plan was to give each modifier its own super short script, with a function for calculating word score. I could pass in the word and its score, then check if the condition is met. If so, multiply the word score by the desired amount and pass it back to the main score script.

Now the main score script just needs a list of the player’s modifiers, and loop through the associated scripts one by one - testing that onWordScore function.

The only problem is... well, you can’t just have an arbitrary list of scripts, and try and run a function on all of them.

I guess, think of it like this. I ask you to bring me a list of 10 items. And then I say... eat them. You’d be like, wait, what? But this is a motorbik... EAT THEM.

It wouldn’t work, right? The list needs to be more specific. They all need to be edible. They all need to be of the same type. The same class.

Aha!

You see, every script you write in Unity is based of the same class - MonoBehaviour. This lets you use all sorts of default functions like OnCollisionExit, OnMouseOver, and OnDestroy.

But you can actually write your OWN class - and then make scripts that inherit all of that class’s functions. Ehhh... it’s probably easier if I just show you.

So, it’s easy to do. We first make a blank script called something like ModifierClass. And then simply add “abstract” in-between public and class. That turns it into a base class that we can build on.

Then when we write a function, like OnWordScored. We just need to put virtual in between public and, in this case, float.

For now, I’m going to have it return -1. You can never score minus numbers in Word Play, so I can just use -1 as a sort of stand in to say “nothing happened”.

So this is our base class. And now we can make scripts that will inherit from it.

So, I can now make a new script for my modifier that checks if the first letter is a vowel. But instead of it using the class MonoBehaviour... I can make it use the class ModifierClass. And now, I can add the OnWordScored function - and you’ll notice that it’s listed as an override. That basically means - ignore the function in the base class, and use this one instead.

And in here, I can write the code to check if its a vowel. And if so return the wordScore multiplied by 1.5. Otherwise, return that “nothing happened” -1.

And now I actually CAN make a list of scripts - specifically, a list of ModifierClass scripts, because now Unity knows that every script will definitely have an OnWordScored function. Either the base function that just returns -1, or an override that may return a higher number.

And so back to our wordScore script, we now can loop through that list of owned modifiers, run the OnWordScored function on all of them... and if they don’t return -1, we can update the word score. And maybe, like, I dunno, make the modifier do a little ping animation and sound effect.

And that’s that! we’ve gone from the script having 100 if statements in a big unwieldy chunk, to having the script just do one elegant loop. And all of the score functionality lives in separate scripts - which makes it easy to find them, change them, and, of course, add new ones. Plus, the modifiers are triggered in the order they are in the list - so you just need to make sure the list syncs up with the modifiers on screen.

This system also let me make the modifiers work in other areas of the game too.

For instance, there are modifiers that do things when you use an upgrade. So I add a bool to the base class, called OnUpgradeUsed, that returns false by default. Then I can make a new script with an override that, say, increases its bonus points when you use an upgrade. And then finally, in the “use upgrade” code, I just loop through the owned modifiers - and trigger the function onUpgradeUsed.

And the nice thing is - this is completely compatible with the other modifiers. Like, the “is the first letter a vowel” modifier doesn’t have an override for OnUpgradeUsed, so it will simply use the code on the base class - which is to do nothing and return false.

And so with that, I can have these triggers all throughout Word Play. Like, when you sell a modifier. Score each letter. When you lose a life. Get a game over. Open the perk screen. Re-roll the perks. Reach a new round. Refresh the tiles. Add a tile to the letter bag. And so on and so on and son. Each one is a tiny script, simply containing a few lines of code needed for the modifier’s functionality - plus details like the modifier’s icon, description, rarity, and special numbers... so it’s easy to change them from the Unity inspector.

I also used this same system for the upgrades, which are the yellow perks that change a tile when you drag it on. Each upgrade inherits from a base class with a TransformTile function - and then each upgrade’s script has a unique override for that function, depending on whether it destroys the tile, turns it into an emerald tile, increases the tiles’s score, and so on.

So - this is the power of inheritance. A really useful way of working with classes. And you could imagine a lot of ways to use this. Perhaps a base class for enemies that all share the same basic movement, health system, and death functions - but with unique code for their attacks. Or a base class for guns with the same code for aiming and reloading, but unique functions for their special abilities.

The possibilities are endless. But this system ended up providing the perfect architecture for my game - letting me add, remove, debug, balance, and change the more than 150 perks in Word Play.

Thanks for watching. I hope you found that interesting. Word Play is available on Steam and is coming very, very soon to iPhone and iPad. Check them out!

Byeee

Video Transcript:

Hey, it’s Mark. This is not like a real episode of GMTK.

It’s just... I was working on the video about GTA3 and memory and streaming and I used this short clip from Cyberpunk 2077, and was like, “Oh yeah, that was from that super impressive montage sequence. Wait. How did they pull that off?”

And basically, the only way to not get distracted and go down another rabbit hole while I was already going down a rabbit hole was to promise myself I’d look into it as soon as the GTA video was done.

So. Here I am.

Okay, let’s have some context.

After you finish the prologue, whether that’s for the Nomad, Street Kid, or Corpo Life Path, we get a sequence that follows protagonist V and best buddy Jackie Welles across six months in Night City.

And it’s this super cinematic montage that jumps from location to location and from scene to scene as it depicts their successes and setbacks.

This whole sequence is about one minute twenty seconds, and it has fifteen location changes, seventy-nine cuts, three mirror scenes with three different costumes, and about forty-five unique NPCs.

Now the sensible thing to do, from a technical perspective, would be to just do it as a pre-rendered video.

Like take Expedition 33. Most of the cutscenes in this game are done in engine, but for this scene, which is a montage of the group collecting Chroma from all around The Continent, the developer just plays a pre-rendered video file.

But Cyberpunk’s montage is entirely In engine, which means it’s happening in real time in the game itself. That’s what makes it so impressive.

Now probably the most impressive part is the sheer number of locations.

This sequence takes us from a street in Japantown to Mama Wells’ house to Lizzie’s Bar, to a basketball court in Heywood, a car dealership, Misty’s shop, Viktor’s Clinic, a random alleyway, a nightclub, back to the Welles’s, then to a garage, to a gun shop, another alleyway, back to the Welles’s one last time to move out, and then finally to V’s apartment.

Phew! This is a whistle stop tour of Cyberpunk 2077’s open world.

But as we know from the GTA 3 episode, it’s not advisable to move so quickly between locations as the streaming tech needs time to load in the next place.

I mean, look how long it takes to swap between protagonists in GTA V. It’s quick, but there’s still a big delay to load in the assets for the next area. They’re certainly not doing it in a single frame.

So how does Cyberpunk do it?

Well, I decompiled the source code... and no, just kidding, not this time. I just installed a bunch of mods. One of which let me pause the montage and fly around, revealing that we’re not in Night City at all.

It’s actually a special back lot hidden away in the far corner of the world, and it contains all of the places we’ll visit in the montage. Like, here’s the street in Japantown. But it’s just a small part of that location. There’s nothing to the left and right of V’s starting point, and if you travel to the end of the road, you’ll quickly reach the bounds of the world.

And if you turn around, hey, that’s Mama Welles’s house. There’s also the basketball court, the alleyways, and the gun shop. The nightclub is floating directly above the Wells’s house, and the car dealership is sitting on top of the club.

Then as the scene plays out, V will teleport from area to area in order to swap location. All while the game snaps between day and night, depending on the current scene.

So the benefit of doing it like this, instead of literally moving around the real Night City is that we only need to worry about streaming the assets we need for this montage sequence and nothing else.

Like this scene’s environment designer says “it was quite a challenge to make the streaming load in such a fast pace, but each scene was optimised and close to each other, to facilitate streaming”.

And you can see how CD Projekt Red aggressively optimised these scenes to keep everything as lean as possible.

They used low-poly models and textures for stuff in the distance, and they didn’t include things you can’t see, like side rooms or the backs of certain buildings.

The montage also implies more life and detail than it actually shows. For instance, this scene at the very beginning makes it look like you and Jackie are walking through a bustling city street, but it’s actually just a bunch of low-poly cars covered up by a smoke particle effect. And literally four NPCs. The rest of the street is completely empty, but it feels alive in the moment.

Speaking of NPCs, the game also loads all of them at the start of the sequence. You can find them just hanging out, ready to pop into action. They’re also often on top of each other, which looks pretty funny, or just floating in space below their spawn spot.

But by loading them in early, we can avoid that nasty thing when a character’s model loads a few frames before their texture.

The montage also uses cuts, which makes it look like time has been compressed. For instance, in this scene with the arms dealer, it looks like V and Jackie are trying all sorts of different weapons, with chunks of time cut out between movements. But in reality, all that’s happening is that the characters are teleporting to new locations and starting new animations, which looks like a jump cut.

It’s the same thing that powers this funny animation in Super Mario 3D World.

Then there are the mirrors. Ah, so this is why the scene couldn’t be pre-rendered. It features the player character who is completely customisable. So every version of this montage is unique.

But yeah, mirrors are as much of a technical challenge as streaming in the world. But by going out of bounds, we can see the trick in action.

So Jackie’s room features a mirror but it’s actually not a mirror at all. It’s a window to an identical copy of Jackie’s room on the other side of the wall. And so when you’re looking in the mirror... Jackie and a clone of V exist in the mirror room, while the player character is in the real room.

And then when Jackie grabs you by the shoulders, he instantly teleports from the mirror room to the real room to make this shot work. Clever stuff.

Though, admittedly, not that new. It’s the same technique that gave us the mirrored room in Super Mario 64, after all.

And so, with these tricks, right, the fake Night City set, the spawned in NPCs, the staccato animations, and the mirrored rooms CD Projekt Red pulled off a real time cinematic montage that plays out perfectly every single time.

Okay, yeah, okay. We can’t talk about Cyberpunk 2077 without talking about its disastrous launch.

It’s no secret that this game was full of bugs and issues and unoptimised code, and it ran like an absolute dog on platforms like the non-Pro PS4 and the Xbox One.

And this sequence was certainly not immune. You’d see pop in, weird lighting changes, low poly models and textures, Jackie just being straight up invisible, and other issues.

But after a lot of updates and patches, this cutscene does now work pretty much flawlessly in 2026. On PC, at least. Well, almost.

This character’s coat still renders in front of the smoke, which is hilarious. And if we’re being really nit-picky, then why are the posters in the mirror not flipped horizontally, eh? Is that some super clever 2077 technology or just a mistake?

Anyway, there we go. That’s out of my system. I can stop thinking about it. I know how it’s done now.

And now, just because this took a ridiculous amount of time and hackery to produce, I’ll end with the full cutscene, but this time from a fixed camera perspective. Thanks for watching. See you soon.

Sources, comments, and more on YouTube

This is Grand Theft Auto III. And this game... this game is a lot of things.

I mean, it laid the foundations for one of the most profitable franchises in video game history. It started an unstoppable tidal wave of open world games, and it kept the news channels very, very busy. But this game is also nothing short of a magic trick.

A trick where Rockstar North, then DMA Design, managed to squeeze an entire city, an entire four kilometer wide pastiche of New York City, into the tiny 32 megabyte memory chip of a PlayStation 2.

So, I’m Mark Brown. This is Game Maker’s Toolkit. And here’s how they did it.

Okay, so I think we need a super quick, and super simplified primer on how memory works in games... or else none of this is going to make sense.

So on PlayStation 2, you’ve got the DVD - which is a whopping 4.7 gigabytes. That’s big enough to hold the entire game - every model, texture, sound effect, and radio station. But it’s also painfully slow, with a transfer speed of about 5 to 6 megabytes per second. Way too slow to load assets on the fly.

But then there’s the memory inside the system itself. This thing is lighting fast. It can move files at 3.2 gigabytes per second. Gigabytes! So that makes it perfect for assets. But it’s also tiny at just 32 megabytes - less than 1 percent of the DVD’s capacity.

Now in a traditional game, well, the console loads all of the data needed for the current area, from the DVD and into the memory. All of the models, textures, animations, sound effects, and so on. Then when you leave the area, those assets can be flushed and the assets for the next area can be copied into the memory - with a loading screen while the assets are shifted across.

But this set-up would not work for Grand Theft Auto 3.

It doesn’t have individual areas, with walls and doors to separate the zones. Instead, it’s a sprawling metropolitan city that you can explore, freely in any direction. A city made up of thousands of unique models, for every building, vehicle, and pedestrian in the game - about 77MB in total. And then thousands of different textures for every material, sign, and character skin - another 45MB.

And so with a total of about 130MB of data, Liberty City is way too big for our paltry 32MB budget.

Well - okay, let me be perfectly honest. Liberty City is split into three different islands. The industrial zone of Portland. The downtown sprawl of Staunton Island. And the luxury suburbs of Shoreside Vale. And these are treated as separate levels with quick loading screens whenever you cross between them.

But even still - each one of these islands is still too big to fit into memory.

Take Portland: it’s got over 600 unique models, requiring roughly 40 to 50 megabytes of data for the models and textures. Still way more than the 32MB budget.

So what could Rockstar do?

Well, the studio’s first idea was to just make the city in a much more simplistic way. To use basic 3D models with low resolution textures - so each island could comfortably fit into memory. But the game’s technical director Obbe Vermeij said this “resulted in the city looking bland”.

So the game’s other technical lead, Adam Fowler, came up with a much more ambitious solution.

It works like this: Liberty City is split into thousands of small sectors. Then, there’s an invisible square around the player. And all of the models needed for the sectors in that square are loaded into memory.

Then, as you move around the city, the game loads the models and textures for the buildings in the sectors you enter - and unloads all of the assets from the sectors you as leave them.

And because the new models will always be far away from the player, it doesn’t matter that it might take a short while to load them off the DVD. It’s a technique called “streaming”.

Now I’d love to show this happening in a more real-time, visual way.

But because the streaming system is tied to the camera, and not the player… well, even if I take control of the camera and fly far away, the streaming will just follow me and the city will materialise around me.

The only solution, ultimately, would be to somehow get the GTA 3 source code, rewrite several sections of the program, and compile a brand new executable.

So, I got the GTA free source code, I rewrote several sections of the program, and I compiled a brand new executable.

Now you can see the trick: how Grand Theft Auto 3 loads and unloads a small number of assets into memory as you move around Liberty City. It secretly builds the world in front of you, and it silently deletes the world behind your back.

Basically: instead of trying to fit a whole city into memory, Rockstar instead built a moving window which shows just enough of the city to make the illusion work.

And with this, Rockstar could build a city as big as it wanted, with artist Aaron Garbut saying “getting it right effectively removed a lot of limitations for scale and detail. We could build what we wanted as long as the area around the player would fit in memory and we could get it off the DVD fast enough”.

Now: I should note... GTA 3 is certainly not the first game to come up with the idea of loading assets into memory while you play.

In fact - this was already happening on the previous generation.

Take the PlayStation 1 game Legacy of Kain: Soul Reaver, developed by Crystal Dynamics. The developer realised it could chop an entire world into lots of small chunks, and then lay them out, end to end, in a line. Then, when a player approaches the boundary of one chunk, the game would quickly start loading the next part so it was ready for you by the time you got there - giving you a whole world to explore without needing to stop for a loading screen.

That same technique has been used in hundreds of games - from Jak and Daxter and Metroid Prime, to Uncharted and God of War.

It also explains why video game characters spend so much time slowly squeezing through tight gaps: it gives the console time to load the next area into memory.

But, back to Grand Theft Auto - this probably WAS the first game to use a streaming solution to create a dense urban environment. A world where the developers can’t predict which area you’ll go to next, or hide its loading behind doors or elevators or tight gaps. This was an actual open world that you can explore freely, in literally 360 degrees.

But pulling it off was not easy, and this ambitious plan led to 5 big problems for Rockstar to solve.

Pop-in

If GTA 3 worked exactly as I just described, with assets loading and unloading in a tight rectangle around the player, then it would be a game where bits of the world materialise out of thin air in front of you, with the dreaded pop-in effect of a last generation game.

So it’s not enough to just stream in the models nearest to the player - but, at the same time, you can only cast your net so wide before running out of memory.

So here’s the solution: the game checks a large number of sectors - but in those far away zones, the game only loads the bigger structures, like buildings, bridges, ships, and cranes.

So the area around your character is fully loaded in, but anything more than a few sectors away will only feature the largest structures.

Of course, this would still eat up a lot of memory. So instead of using the actual models, these far away structures use a simplified, low polygon imposter with a crummy low resolution texture. Then, when we get close enough, the real model and texture can be streamed in, and gracefully fade in over the top of the low poly copy.

It’s a system called LOD, or level of detail.

And it even extends to the other islands. Portland, Staunton, and Shoreside each have low poly models that sit on the horizon as stand-ins for the other bits of Liberty City. Actually, Staunton has two - with different models depending on whether you’re looking over from Portland or Shoreside.

There’s no point rendering the back of these skyscrapers if you’re in the east, after all. Better to save on triangles.

All of this lets Liberty City maintain its distant skyline, for immersion and navigation purposes. But by using low resolution stand-ins for these faraway buildings the game can save a lot of memory.

More in memory

The next problem is that, well, Liberty City is not the only thing we need to store in memory.

That 32 megabyte pile of RAM also needs to keep track of animations, pedestrian AI, physics calculations, the location of your last car, and so on.

So this means only a small slice of the total memory overhead can be dedicated to streaming.

I don’t know exactly how much memory GTA3 uses, but if you take San Andreas, which is using the same engine on the same console, take the DVD and put it into your computer... you’ll find a file called stream.ini, which shows that just thirteen and a half megabytes of memory are dedicated to streaming the city.

On top of that, Liberty City is more than just buildings and lamp posts. This is a bustling city filled with cars and pedestrians, which must also share that memory space. So, all of these systems had to be heavily optimised to save enough room for the city streaming.

Take the cars: GTA 3 has about 60 unique vehicles. And I’d estimate that the full fleet is about 10MB worth of models and textures. So it wouldn’t be sensible to keep them all loaded in at the same time.

Instead, the game keeps a very strict pool - only eight vehicle types can be loaded into memory at any one time.

When a car is spawned in - it must either use a vehicle from one of these eight slots, or boot an older car out to make room.

And because that tiny pool also needs to contain the car you’re currently driving, the last car you drove, any parked cars in the nearby area, and common vehicles like taxis and police vehicles - including SWAT cars, helicopters, and army trucks if your wanted level is high enough... well, this is why you’ll sometimes see the same cars repeating again and again.

There’s only a few spare slots left for vehicle variations.

This also explains that spooky GTA phenomenon of looking for one very specific car, only to finally track it down, and now everyone in Liberty City seems to be driving that car. It makes sense: because it’s now in one of those eight car slots, it’s much more likely for it to be spawned in as the next vehicle.

So basically, with only a tiny amount of memory to go around, everything in the game must be as optimised and streamlined as possible to make sure there’s room for the main city streaming system.

Memory fragmentation

So, in super, super simplified terms, if you remove a small asset from memory, then it leaves a nasty little hole in your total memory space.

And then if we load in a bigger asset - well, even though there’s space for it... that space is fragmented into lots of little holes, and so the asset just has to lump on the end - exceeding the total amount of memory and causing the game to crash.

Not good.

That’s fragmentation.

And so if you’re constantly pushing assets in and out of memory, fragmentation becomes very likely and memory blockages are almost guaranteed to occur.

So smarty pants Adam Fowler had to write a bespoke memory manager that could deal with these issues, with functions that can merge and move blocks of memory around to avoid fragmentation.

It can also aggressively free up memory if needed - going in and deleting stuff behind the camera, or big buildings in the distant skyline.

Another clever trick was to keep many of the asset files as the exact same size. Like, among the models and textures, over 500 of them are exactly 2 kilobytes, and 416 are exactly 4kilobytes. That way, assets can seamlessly take each other’s place as they load in and out - filling those gaps in memory and avoiding fragmentation altogether.

To check if this all worked, Fowler would attach the player character to the train that zips around Portland. And then set the game running on five or so PS2 dev kits before he left work for the night.

And as Vermeij explained on the podcast Kiwi Talkz...

“And then he was hoping that he’d come back in the morning and that the games were still running.

And the first time he did that, they all crashed after five minutes.

And then he would spend a couple of days figuring out what had gone wrong.

And then he did it again. And then it was they crashed after thirty minutes. And then maybe an hour.

And eventually they got it through the night.

So that meant that for the whole night, the game was loading models and unloading, loading models and unloading”.

Seeking

So here’s the thing. CDs and DVDs are designed for the continuous read of something linear like an album or a movie.

And so the data is embedded in a non-stop spiral that goes from the centre to the edge - and the laser reads that microscopic groove in a continuous line.

But if you want to jump around to another part of the disc - called seeking - then the laser head has to move. And the acceleration, movement, and deceleration of that mechanical part all takes time.

So now imagine how GTA 3 is constantly asking the PS2’s DVD reader to seek out different bits of the disc.

To speed things up, Rockstar had to literally optimise the position of assets on the disc to reduce seeking time. That meant bunching up assets from the same part of the city, so the laser wouldn’t have to travel as far to get the next batch of buildings or roads.

Also, the DVD reader code is given a queue of assets to load in, but it will prioritise the ones closest to its current read position, instead of going through them in the order they came in.

And Vermeij says the team experimented with duplicating assets on the DVD - because it’s more actually efficient to have multiple versions of common models like trees and lampposts at different points on the disc, instead of having to constantly seek back to find that single tree model.

And this technique is still used in game development, with a game like Insomniac’s Spider-Man having hundreds of the same mailbox model at different points on the Blu-Ray.

Now you might wonder if all that seeking is safe for the DVD drive. And Rockstar shared that worry. Here’s producer Leslie Benzies, in the documentary The PlayStation Revolution...

Streaming was a risk. Could a drive inside a machine constantly move for 24 hours and not melt?

Now it obviously didn’t melt your PS2’s drive - and Vermeij assumes this was cleared with Sony before the game was sold. But it’s possible this did do some damage to the console, and you can find reports of disc read errors and even busted DVD drives from this era.

Speedy players

Even with all of these techniques and optimisations, the disc simply couldn’t read assets fast enough - and if players moved through Liberty City too quickly, they could outrun the streaming system and see the world appear in front of them.

Or even worse: fall through the floor and into the blue hell beneath.

So the only real solution was to just slow players down.

Cars were given a strict maximum speed - and in certain parts of the city, the developers secretly increased the air resistance on vehicles by a few percentage points - not enough for players to notice, but enough to let the disc catch up.

Flying vehicles were completely ruled out, and the only plane in the game had its wings clipped so it was impossible to fly. Well, almost impossible.

And the layout of the city needed adjusting, too. Vermeij explains that in its original form, Portland featured a main drag that ran practically across the entire island - a nightmare scenario where players could build up to max speed while the streaming code struggled to load in the buildings on both sides of the road.

So in the end, this big building was placed directly in the centre of the main thoroughfare so it wasn’t possible to go so fast.

So - that’s the trick.

Rockstar had to fit a roughly 130 megabyte city into the tiny 32 megabyte memory chip on your PlayStation 2.

And the only way to do it was to move assets from the disc and into memory at rapid speeds, constantly chucking out the buildings and trees in your rear view mirror to make room for the shops and bridges just around the corner.

But with this, and a lot of clever optimisations, Rockstar showed it was possible to have a rich and detailed open world on console - and it laid the foundation for the rest of the series.

That, of course, includes two more GTA entries on PlayStation 2, which are even more impressive in their implementation.

Vice City added helicopters - by only showing low poly models when you’re above a certain altitude, so there’s enough memory spare for the most distant skylines.

And San Andreas got rid of the separate islands and halting loading screens - partly by having so much empty countryside in between the cities that one will be entirely out of memory by the time you reach another.

But even as gaming hardware got better, these techniques still needed to be used - because the worlds became more detailed and rich in step.

Like, on a modern computer with gigabytes and gigabytes of RAM, I can easily fit all of Grand Theft Auto III’s Liberty City in memory at one time. No streaming needed!

But I can’t fit the whole world of GTA IV or V - so it’s back to the cheeky optimisations. Like putting roadworks in the middle of Algonquin’s longest road, to stop players from driving so fast that they outpace the streaming tech.

Now in more recent times, a move to solid state hard drives is making things easier for developers. There’s no more seeking - so you don’t need to optimise asset location or duplicate dustbins everywhere.

But even still - I can promise you now that Grand Theft Auto VI will still stream bits of its world in and out of memory.

And it will use some of the same techniques, optimisations, and maybe even a few lines of code from Grand Theft Auto 3 - the game that performed a magic trick, and proved it was possible to squeeze an entire city into just 32 megabytes of memory.

Sources, comments, and more on YouTube

Hey, it’s Mark.

So, back in 2013, longtime Zelda producer Eiji Aonuma promised that the next big Zelda entry - coming soon to the Wii U - would completely reinvent the Zelda franchise.

He said...

“Our mission ... is quite plainly to re-think the conventions of Zelda.

I’m referring to the expectation that the player is supposed to complete dungeons in a certain order, or that you’re supposed to play by yourself.

The things we’ve come to take for granted recently.

We want to set aside these conventions, get back to basics and create a newborn Zelda so that the players can best enjoy the real essence of the franchise.”

And he wasn’t lying.

This next game wouldn’t just shake up the formula - but it would fundamentally change how Zelda games would be designed, perhaps, forever.

Here’s how it happened.

So, by 2013, Zelda had settled into a familiar and formulaic groove.

In each game, hero Link visits dungeon number one, finds some handy new item, uses it to defeat the boss - and open up the next part of the world map. Where he’ll visit dungeon number two, find a handy new item, defeat a boss - and, well, you get the picture.

Sure, each Zelda game might have a unique world... tone... cast of characters... and clever mechanical gimmick - but this strictly linear structure had barely changed since Ocarina of Time back in 1998.

It had become stale and a little bit stinky. And it all culminated in the disappointing Skyward Sword - which adhered so closely to this formula that it felt like a pale paint-by-numbers parody of a Zelda game.

Something had to change, and Aonuma knew it - saying...

“We started to feel that fans may not continue playing this franchise unless they can think independently and try various things freely on their own, rather than following a set path”.

So it was time to mix things up.

For the next game - which, spoiler alert, would be Breath of the Wild - he gave his team a wildly ambitious mandate: to rethink every aspect of the Zelda franchise.

To consider every convention that had become an essential staple of the series and ask if it truly had to be that way.

Aonuma says “I told the staff, “This time, let’s build the game without shying away from anything. Let’s stop thinking about what Zelda has been up until now.”

Of course, the first ideas were pretty out there.

Link fighting aliens in UFOs. Or dodging missile attacks from the sky. Or meeting Lovecraftian monsters. Or bombing around Hyrule on a motorbike.

That was all a bit off. Well, except for the motorbike.

But eventually the team came to look at the more fundamental, structural bones of the Zelda formula. That’s ultimately what needed changing.

The game’s director, Hidemaro Fujibayashi, said...

“We wrote down all of the stress points, the things that make Zelda games less enjoyable, and we replaced them with new ideas.”

And this led to five key changes to the Zelda formula.

So, as stated, Zelda was a series where you must do things in a strictly linear, and predetermined order.

But for this new game, there would be no predetermined order. The four divine beasts and the 120-odd shrines could be tackled in literally any order you see fit. Or not at all, actually. The only mandatory bits in Breath of the Wild are the tutorial and the fight against Ganon.

The rest is entirely optional.

This idea was trialled in A Link Between Worlds on 3DS. And it was part of the very first Zelda game on NES. But with Breath of the Wild, Zelda would finally see true freedom and complete non linearity.

But it wasn’t just about linearity. Zelda was also a series made up of individual areas connected by doors, gates, and gaps in the clouds.

But for Breath of the Wild, the world would be a single contiguous space that can be explored in any direction. And nothing can block your path as Link can swim across rivers, climb up virtually any vertical surface, and soar over the landscape with a paraglider.

Zelda was also a series that told you where to go and what to do, often with an overbearing companion character who would offer, uh, helpful hints and advice.

Breath of the Wild would be the opposite. Outside of quest markers that show the location of the four divine beasts, players would be left to explore by themselves.

And the developers would stay silent - choosing to guide the player through the design of the environment, or the placement of harder enemies, or with a breadcrumb trail of curiosities that lead you to the next important place.

Zelda was also a game with hand-crafted puzzles that would require you to find the designer’s single, intended solution.

But for Breath of the Wild, that would be totally reversed. Instead, the world is constructed from a set of interconnected physical and chemical properties like wind, fire, electricity, and gravity. And so players could exploit these relationships to discover their own way to solve puzzles, defeat enemies, and bypass obstacles.

Even if that means skipping the puzzle entirely.

Fujibayashi says “I really want players to relish those moments where you do something and think, oh my gosh, I’m a genius.”

And Zelda was a game with a pretty limited moveset, and mostly contextual actions - down to having an automated jump when Link runs off the edge of a platform.

But in Breath of the Wild, Link would have a surprisingly extensive moveset. That includes a dedicated jump button, the ability to throw weapons, climb up walls, surf on his shield, and more.

This would give players way more control over Link’s actions and let them travel through the world in a more freeform manner.

And look, there are even more ways that Breath of the Wild mixed things up. Like how you no longer get hearts from chopping down grass, but from hunting and cooking. Or how Link now wears a blue tunic rather than a green one.

But, it was those five key changes to the Zelda formula - a non-linear structure, a fully open world, laid-back hints, systems-driven problem-solving, and an expanded moveset - that really made Breath of the Wild feel so different.

But here’s the thing.

The result wasn’t just a fun game, and a Game of the Year winner. The result was actually a fundamental shift in how the game relates to the player.

You see, Fujibayashi refers to the previous Zelda games, like Skyward Sword, as passive games.

Games where players go through the content in a predetermined order, solve puzzles by finding the developer’s intended solution, and play within the tight confines of walls and obstacles.

Games where the player merely follows the designer’s lead in a docile, passive way.

But with these changes, Breath of the Wild became the opposite of a passive game: an active game.

Here, the player is very much in the driving seat. They choose what content to engage with and what order to do it. They choose where to go and how to get there. They figure out their own solutions to problems. And they have to think, plan, explore, and experiment in a much more active way.

In a passive game, the designer’s job is to make hand crafted content for players to experience.

But in an active game, the designer’s job is to make a sandbox, give the player some toys, define a goal, and just let the player get on with it.

And I think this is quite an interesting lens through which to look at game design as you can plot almost every game on this spectrum.

From very passive games - you know, linear, story-based games like Call of Duty and The Last of Us - to super active games - playful, sandbox sims like Minecraft and Factorio. As well as everything in between.

Now in order to get here, the developers had to rethink everything.

Art director Satoru Takizawa said “to create an appealing, never-before-seen world filled with new surprises, you need to break conventions. You need to be creatively reckless.”

But there were some other important factors, too.

Like having fresh eyes - Aonuma gives credit to the contributions of younger staff members who would come up with fresh ideas and weren’t set in their ways from decades of working on the same franchise.

Fan feedback was important too - the producer says that Breath of the Wild’s open world was made partly in response to players wanting to explore the areas in between the disconnected chunks of Skyward Sword.

And there were also outside influences - the team played games like Minecraft, Terraria, and Skyrim to get ideas. Historically it’s rare for Nintendo to name check other games - especially ones made outside of Japan.

But, whatever the case, the result was a truly incredible game. And for the Zelda team - it would prove impossible to ever go back to that more passive style of game design.

Like, seriously - Aonuma was practically radicalised by what he saw in Breath of the Wild. It fundamentally changed his mind about how Zelda games should be designed.

For instance, he’s said that he once thought it was a “sin” to let players get lost in a game - hence the overbearing helper characters and the linear level design of the older Zelda entries.

But with Breath of the Wild, he saw how it could be fun to let players wander off the beaten track and find their own adventure.

He says...

“I wanted to basically have the players experience this ‘getting lost’ while getting to the next destination, because this journey leads to something they’ve created with a path they’ve chosen.

I myself learned that getting lost wasn’t so bad after all, so I was able to reflect this idea to this game.”

So for Aonuma, Breath of the Wild showed a new, and - to him - much better way to design games.

In fact, when he replayed the Wind Waker, he thought it felt outdated. He’s described set puzzles with single solutions as being an “old design”. And as for future Zelda titles? Well, he says...

“I think for me … the incredible freedom that this game offers you and how well that’s been received…to me, it means that level of freedom is something that needs to be maintained in Zelda games going forward. My eyes have been opened to how important that is”.

And so Breath of the Wild has continued to influence future games in the franchise.

And the sequel, Tears of the Kingdom, doubled down on this active, player-led, problem-solving gameplay.

In fact, Nintendo was so inspired by how players used the Breath of the Wild sandbox to pull off wild and unexpected things that the sequel gives you even more radical toys - including the ability to glue physics objects together to build vehicles and machines.

And players are given even more opportunities to exploit the game and find their own solution.

Take the ability “ascend”, which lets you dig through the ceiling and pop out on the other side. It started as a debug option during development, to let the creators quickly skip past content, but it later became a defacto power in game.

Technical director Takuhiro Dohta said...

“We’ve placed more importance on creating a game that enables players to do exactly what they think they can do, rather than how we want them to play the game”.

And then there’s the spin-off title, Echoes of Wisdom, which was made by outside studio Grezzo. The team pitched a clever copy-and-paste game mechanic and Aonuma was like “yeah, cool, whatever, great - just make sure the player has tonnes of freedom”.

So the game lets players be mischievous and exploit all sorts of cheesy tricks to solve problems.

And players are allowed to tackle challenges out of order, and are encouraged to find their own path through the story.

Co-director Tomomi Sano says...

“We put effort into maintaining player freedom. Limiting mechanics makes development easier, but it frustrates players when things don’t work as expected. We wanted to avoid that.”

So if you ask Aonuma - this is how every Zelda game will be, going forward. He says...

“Games where you need to follow a specific set of steps or complete tasks in a very set order are kind of the games of the past.

The games of today are ones in which that can accept a player’s own decisions and give them the freedom to flexibly proceed through the game, and the game will allow for that”.

In fact, he went so far as to question why anyone would want a game that felt like the older Zelda titles.

He said...

“Why do you want to go back to a type of game where you’re more limited or more restricted in the types of things or ways you can play?”

Well... sir, if you’re asking, then, I have some thoughts.

Because here’s the thing: as much as I love these three new games - and all of the advantages that this “active” design philosophy brings - I have to admit that these games don’t always feel like Zelda titles.

And that’s because the changes necessary to make an active game ended up removing some of the stuff that makes Zelda... well, Zelda.

For example, consider that old formula where you get a new item in each dungeon.

This might be linear and restrictive but it does mean that each Zelda game has a wonderful sense of mechanical progression - where you get brand new toys at fairly even points throughout the adventure.

As the game goes on, Link becomes more powerful and adept - and the player is always left wondering what cool new ability is just around the corner.

But for Breath of the Wild - well, the team never wants you to reach a shrine or a dungeon that requires some ability that you don’t have. And so the solution is to just dump all of the games mechanics on you, during the tutorial.

Which means you’ll see pretty much everything the game has to offer in the first couple hours. And then only one of the dungeon rewards - Revali’s Gale - adds a fun new wrinkle to Link’s moveset.

Then there are the dungeons themselves. Zelda’s old temples are beautiful puzzle box playgrounds. Clever spatial reasoning conundrums that you slowly unpick by finding keys and items. Or by changing the water level... moving back and forth in time... or flipping the entire temple on its head.

But that sort of level design requires heavy restrictions, a narrow lock and key structure, and hand-crafted puzzle solutions. That doesn’t gel with Breath of the Wild’s freewheeling movement system and puzzle design.

But that leads to lax dungeon layouts that can be cheated and cheesed. And so, the new dungeons often end up feeling messy and frankly unsatisfying. I didn’t even bother doing a Boss Keys episode on Tears of the Kingdom’s dungeons because what’s the point if you can just phase through the floor and skip half the puzzles?

And more broadly, Nintendo’s focus on player-driven solutions can sometimes lead to moments of surprise and delight - but just as often it can lead to repetitive solutions, cheesy exploits, and outright cheats.

There’s also the storyline. I’ve never been too fussed by the narratives in Zelda games - but they do at least have a structure. Breath of the Wild’s plot, meanwhile, is essentially, a bunch of disparate memories told in a jumbled up order. There’s no emotional escalation, no dramatic timing, no pace, no structure, no arcs.

And finally - the older Zelda games were full of beautiful, curated, and hand-crafted spaces that have resonated for decades. But the new Zelda worlds - by dint of needing to be utterly massive - can feel empty, soulless, and repetitive.

These worlds even veer into the realm of feeling like they were generated by an algorithm.

So when Aonuma asks why anyone would want to return to the older style of Zelda - well this is why. There are hugely important aspects of those older titles that have simply been lost in the shuffle.

And ultimately, I’d argue that the active design philosophy of Breath of the Wild isn’t inherently better than the passive design philosophy of Ocarina of Time - it’s just different. In fact, both styles have their own pros and their own cons.

They offer different experiences that suit different players, or just different moods.

Sometimes I want to go on my own adventure across an uncharted landscape. But sometimes I just want to be delighted by a developer’s carefully curated experience.

But here’s the question - does it have to be one way or the other? Because in the last year, I’ve played two games that had me thinking about how Zelda might reconcile its different eras.

The first is Resident Evil: Requiem. No spoilers, don’t worry. So this game cleverly mashes up two very different parts of the franchise’s history.

When you play as Grace Ashcroft, the game is a first-person, haunted house spook ‘em up, ala Resident Evil 7 and 8. But when you play as Leon Kennedy, the game turns into a campy, third-person, action-packed romp, calling back to Resident Evil 4, 5, and 6.

This two-hander not only satisfies fans of both styles of game - but it works as a nifty little palette cleanser.

Just when your anxiety is amped up to breaking point, the perspective shifts to Leon and you get to blow off some steam.

Could you do something similar for Zelda? A game that incorporates aspects of both eras? Maybe! I could imagine some future game more clearly delineate between the overworld - which would play like Breath of the Wild - and the dungeons - which would feel more like classic Zelda.

And, as for the other game?

Well, it’s no secret that Breath of the Wild has inspired other Nintendo franchises, beyond just Zelda itself.

Fans and critics have noticed its influence on games like Mario Odyssey, Bowser’s Fury, Mario Kart World, and Pokemon Legends. Plus, Metroid producer Kensuke Tanabe has specifically said that Breath of the Wild’s open world design led to the inclusion of the desert hub in Prime 4.

But then there’s the most Breath-pilled game of the bunch. Donkey Kong Bananza.

Now this is a very active game. This is a game where you don’t just climb up and over any surface - but can actively smash right through it.

It’s a game with an ultra expansive move set. It’s got heaps of optional content - did you know you can complete this whole game without ever picking up a single Banandium Gem? And there’s a lot of freedom in how you go through the game.

Co-director Kazuya Takahashi says…

“There’s a lot of leeway in terms of progression in this game.

For example, locations or parts of the story you can skip without suffering any consequences.

Typically, in game development, restrictions are put in place to prevent [sequence breaks] from happening, but since this game is all about destruction, we wanted people to play freely.”

But it’s not total freedom. The animal transformations are doled out across the entire adventure, just like classic Zelda’s items.

And there’s a storyline with an actual plot and, like, stuff that happens. I know, crazy right?

Now I don’t think Zelda needs to copy Bananza’s structure entirely.

But it just goes to show that it’s possible to balance player-driven freedom with designer-set progression. A mix of active and passive elements.

So, look - we don’t know much about what Nintendo has planned for the next Zelda game.

We’ve got some clues.

We know it won’t be a direct sequel to Tears of the Kingdom - “that would be a sequel to a sequel, which is getting a bit wild”, says Aonuma.

And it won’t feature Ultrahand - that’s a one-off mechanic, exclusively for Tears.

Oh, and it may take some inspiration from the new Hyrule Warriors game. But it’s very vague what that might mean. And also I haven’t played that game. Is it any good? I dunno. Lemme know.

But - here’s what we do know: just like how the previous games worked off the formula set by Ocarina of Time, this next game will build upon the template set by Breath of the Wild.

But I don’t think we’ve seen the final form of this new style yet.

And I’m confident that we’ll see Aonuma and his team tweak and experiment with this new design philosophy - inspired by both fan feedback, and competing games.

And, if we’re lucky, they’ll find a some mythical middle ground - halfway between the imaginative, freeform experimental fun of Breath of the Wild - and the beautifully handcrafted majesty of Ocarina of Time.

To be honest, I just don’t know.

But I’m looking forward to seeing what they come up with.

Happy 40th anniversary, Zelda.

I can’t wait to see where you go next.

Sources, comments, and more on YouTube

Hello Substack readers!

It’s time for the final Digest of the year. It’s a tad shorter - partly because there’s not a lot a new games releasing in December, and partly because I’m writing this on the 19th so I can stop for Christmas!

But one big thing did happen - the Game Awards, aka The Keighleys. And while I’m loathe to treat these as the defacto Oscar-like winners, they are at least a good indication of what people liked playing in 2025. So I’ve picked some articles and videos about the development of some of the winning games to get more info on how they were made.

This is my last Digest for a while as I’ll be on paternity leave for the first few months of 2026 (I’ll pause billing on Substack during that time). But I’ll likely be watching YouTube at 3AM, covered in baby sick, and so will have a big batch of stuff to recommend when I get back!

Game of the Year

Read - Ask the Developer Vol. 19, Donkey Kong Bananza - Nintendo’s designers explain the design of this year’s “Best Family” game.

Read - Interview with Team Cherry for Hollow Knight Silksong - In an interview at the Australia media museum, Team Cherry discusses the making of this year’s “Best action/adventure” game.

Watch (17 mins) - How Hades 2 Was Made and Why Zagreus Had to be Replaced - ThisGuyGlen breaks down the development of the “Best action game”.

Read - How the Creators of Xbox’s ‘South of Midnight’ Made a Southern Gothic Fantasy - Rolling Stone interviews the devs behind this year’s “Games for impact” winner.

Read - More Games Need Doom: The Dark Ages’ Accessibility Settings - More details on the winner of “Innovation in accessibility”.

Read more

Welcome back to Boss Keys.

This is my series about the design of games with sprawling, interconnected, maze-like worlds. Including dungeons, alien planets, fantasy kingdoms, and gothic castles.

Now the most viewed episode in this series - by far - was on the game Hollow Knight - a modern day masterpiece, and perhaps the finest example of the Metroidvania genre.

That’s a genre, if you still don’t know, that describes games that take place in a single, contiguous world map. Where progress is made through careful exploration, by making connections between abilities and obstacles, and by slowly untangling the knot at the centre of the world.

Now Hollow Knight did this with aplomb. And then combined it with beautiful graphics, a deep well of lore, sharp controls, and terrific boss encounters.

And so a sequel had a lot to live up to. But after seven long years of waiting, it’s finally here, in the form of Hollow Knight: Silksong.

So in this post I’m going to investigate how Australian developer Team Cherry has put together this new world, and how they’ve thought about topics like progression, non linearity, navigation, back tracking, optional content, and more.

There will be major spoilers for every inch of this game. There will be charts and graphs. And there will be hot takes about corpse running.

I hope you’ll join me. I’m Mark Brown. This is Boss Keys. And this is the world design of Hollow Knight: Silksong.

Part 1 - Scaling Pharloom

So, Hollow Knight Silksong takes place in the withered kingdom of Pharloom - a twisty, maze-like land with 12 distinctly different areas - including a crumbling road made of bones, an industrialised forge, a gusty garden, and a farmland beset by eternal rainfall.

And the game starts with a very clear goal.

Hornet has been kidnapped, caged, and transported to this broken land. But with the help of a mysterious silken fly, she manages to escape. Now she can climb out of Moss Grotto - taking on her first boss along the way - and emerge in Bone Bottom, where she is given her first quest:

To climb to the very top of Pharloom to find The Citadel, and track down the ones who tried to imprison her.

Or in other words: go up.

Yeah, it’s a lot more to-the-point than the first Hollow Knight, where the protagonist has, essentially, ended up in Hallownest by accident, and just starts wandering around, looking for something to do. Your goal in Silksong, on the other hand, is super clear.

Now in terms of progression, Silksong mirrors Hollow Knight 1 by starting with a pretty linear and guided trip through the first few areas of the game.

After we emerge in Bone Bottom - our first town - we can journey through the Marrow, until we get our very first ability: the Silkspear.

Well, it’s technically not an ability, just one of a handful of different silk skills that we can use in combat - but, we do need it to blast through a web, so we can take on the Bell Beast.

That’s the second boss, but once defeated, it becomes our means of fast travel throughout Pharloom. Handy.

With the Beast on our side, we’ll then get safe passage to the Deep Docks, where we’ll find ability number two: the Swift Step, aka the dash. This lets us hop over a gap so we can open up a gate, and then face another boss: the dainty swordfighter Lace.

With her defeated - for now - it’s off to the Far Fields. Here we’ll meet the Seamstress who offers us a sort of side quest: basically, if we can gather up the spines that are shot out from some floating fleeces, she’ll sew a new layer into Hornet’s dress, which turns it into the Drifter’s Cloak.

This lets us catch updraft winds and fly up to higher areas.

On the way back we’ll face boss number four: a giant, creaking metal automaton called the Fourth Chorus.

With that thing dead we can ride the winds up, and up, and up... all the way to Greymoor. Here we’ll face all manner of flying critters - including the fifth boss, Moorwing. This guy’s a right pain in the butt.

But, with Moorwing defeated, we’ve reached the town of Bellhart - but right now it’s ensnared in giant webs of silk. So we’ll pass straight through to Shellwood where, after a bit of climbing, we’ll fight boss six - Sister Splinter. And with her defeated we can get another ability, the Cling Grip, which lets us climb up walls.

Okay! Press pause.

So far Silksong has proven to be a pretty straightforward game with a tightly plotted sequence of events. If we chart it out - this is Boss Keys, after all, we love a chart, we see a linear set of boss fights and new abilities.

And if we lay this path over the world map, we see this: a giant U-bend through Pharloom, taking us, in turn, through Moss Grotto, Bone Bottom, The Marrow, Deep Docks, Far Fields, Greymoor, Bellhart, and Shellwood. It looks like the world’s most simple metro map - without the usual zig-zagging and backtracking you’d see in other Metroidvanias.

Of course, it doesn’t feel so simple when you’re on the ground. This is a game of winding pathways, and one-way doors, of enemy gauntlets and tricky platforming sections, of side rooms, elevators, chatty NPCs, and secret areas. So while you’re in the thick of it, the game certainly feels labyrinthine and knotted.

But, from a more zoomed-out perspective, the path is actually pretty straightforward. And here’s why:

For starters - if we think of an ability like a key, and the obstacles that ability helps you overcome as a lock - well, at this point each key only opens a few locks.

Like, when you get the Drifter’s Cloak... there’s no point going back to the Deep Docks or The Marrow, or Bone Bottom because you won’t get access to any new areas. And as for the lock that actually lets you make further progress through Pharloom? Well, it’s always literally just around the corner from the key.

Plus, there are only a few diversions you can make away from the intended route.

For instance - from very early on, you can climb across these perilous hanging cages, and dodge this acidic barf, to enter an area called the Wormways.

But, then again, almost all of the area is guarded behind a locked door, the only key we can get right now is from the shop in Bone Bottom and it costs… 500 rosaries?! In this economy? You’ve gotta be kidding me.

You can also mess around in the eastern side of Greymoor to find Craw Lake and unlock a new silk skill. And then the biggest diversion of the bunch is Hunter’s March. But I’ll come back to that later on.

And finally - it’s hard to ever get lost because you can always just use the ol’ map-poking technique.

Uh, sorry, that’s a term I totally made up. But if you’ve played a few Metroidvania games you’ll be familiar with map poking. It’s when you open the map, look for paths and doors you haven’t been through yet, and then methodically go to each one and see where you can make progress.

And it’s easy to do that in Silksong. The game has the same map system as Hollow Knight 1 - you start each area with no map at all, for that old school, Metroid 1, ‘where the hell am I?’ sensation. Until you find a cartographer - in this case, the warrior Shakra - who will sell you a map of that zone.

Now you’ve got a scrappy, half-finished layout of the world. But when you next rest on a bench, the map will fill in with the areas you’ve visited. And crucially - you’ll see open paths and passages to the areas you haven’t visited.

So if you’re ever stuck, you can just poke at each unexplored area of the map until you make progress. You can also place markers on your map to remind you to come back to specific places later.

So, so far, this is a pretty straightforward game. You’re always moving forward through the map. Each key only has one or two locks, and the one that lets you make progress is always just around the corner. And the map gives you all the info you need to get going.

But - once we get that Cling Grip, things change dramatically. Unlike before, this one key lets us open up a lot of different locks. And owning this ability suddenly opens up a bunch of new rooms and areas.

So there are high-up areas all across Pharloom that we can now reach, letting us get to more goodies and items. We can even face new optional boss fights - like the charging Skull Tyrant in The Marrow.

Also, once we have the Cling Grip - new enemies will start to appear in a handful of locations. They’re called Wardenflys. Little snotty jerks with big cages. And if they manage to trap Hornet, she’ll be kidnapped, again, and whisked off to The Slab.

This is a prison complex, high above Pharloom, where Hornet has no weapons or cloak until she tracks down her attackers.

Now it’s always a surprise in Metroidvanias when you’re transported to some random corner of the map - like the teleporting chests in Elden Ring or the Snatcher in Bloodborne. For a genre that’s so much about knowing where you are and where you’re going, it’s a big deal to suddenly be in completely unfamiliar territory.

And so if you ARE taken to the Slab, you’ll have to use the Bell Beast to fast travel back to Pharloom and carry on.

Right. Let’s get back to the actual progression.

So, the Cling Grip lets us go to two very important places. For one, we can use it to climb to the upper area of Bellhart, and uncover the mystery of what has trapped this town - turns out, it’s a fleet-footed weaver named Widow.

And with her defeated, the spell is broken - reverting Bellhart to a normal town where we’ll find NPCs, shops, side-quests, a needle upgrade, and more.

Plus, we get the fifth upgrade in the game: the Needolin.

This turns our needle into a musical instrument, which can be used to unlock a number of specific doors throughout the game.

It’s also essential for finishing this first act, but more on that in a second.

The other important place to use the Cling Grip is to gain access to the next part of Pharloom - the Blasted Steps, which holds a perilous, wind-swept platforming obstacle course that will finally take us to the front door of the Citadel.

Only... to find that it’s, maybe, locked up tight. That’s because it’s jammed shut until we’ve rung five massive bells throughout Pharloom. You might have found and rung them already, or you might have missed some and have to go back and find them now.

This is... a slightly strange quest. After triumphantly climbing all the way to the top of the Blasted Steps, you may find you are stopped at the very last moment and have to go all the way back down again.

It’s like getting to the top of a mountain and realising you left your camera at base camp.

Luckily there’s a fast travel point nearish to the summit.

It’s also a strange quest because finding these bells just isn’t that tough - I mean, two of them, the ones in the Marrow and the Deep Docks, have to be rung to make progress through those areas, anyway.

Another two, in Shellwood and Greymoor, can be missed if you’re rushing, but they’re not hard to find and aren’t behind a boss or a platforming challenge or a puzzle.

And then the final one is rung after defeating Widow and freeing Bellhart. Which might make you think that this bell quest ensures we’ve saved Bellhart before entering the Citadel. But... we need the Needolin, which we got from defeating Widow, to open the door anyway.

Now it’s not a bad quest, and it certainly ensures that you’ve explored a bunch of Pharloom before the next act. But it also feels like there could have been more to it. And, as it turns out, these bells may have had a different purpose during an earlier stage of development. So who knows.

For now though, let’s add them to the chart.

Anyway! After ringing the bells, reaching the summit, and opening that gate, we face the hardest boss fight yet - and what might prove to be one of the hardest boss fights in the whole game - The Last Judge.

She’s the powerhouse guardian of the Citadel, who fights with flames, does two masks of damage upon impact, and can even kill you, right at the very end, with her final, dying explosion. Not cool, love.

She is an extremely hard boss. Certainly a suitable cap to the end of Act 1. I had a lot of trouble with this fight, almost to the point of wanting to quit the game altogether.

But, as it turns out... you don’t have to fight The Last Judge at all.

Part 2 - Another Route

So I’ve just described Act 1 of Hollow Knight: Silksong as having a pretty straightforward and predictable structure. A sequence of bosses and abilities that must be fought and unlocked in a very specific order.

And for most players, that is how they will scale Pharloom. It’s certainly how I did it. But there are actually a bunch of ways to break that sequence, leave that intended path, and do things your own way.

So, for starters - in The Marrow, you’ll find the Flea Caravan. This works a lot like the Grubfather in Hollow Knight 1. You’ll find fleas dotted around the world and when you get enough of them, you can go back to the caravan to get a reward.

But unlike the Grubfather, the caravan will actually move to a new location on the map after you hunt down a certain number of fleas.

From the Marrow to Greymoor after getting five of them. To the Blasted Steps after getting 12. And to the Pale Lake at the end of the Putrified Ducts after finding 22.

And because the caravan site in Greymoor is literally in the boss fight room with Moorwing, well... if you manage to find 5 fleas before that boss fight, you can ride the caravan to that space and Moorwing simply won’t appear in this room.

Now it will appear later in the game in a different spot - but you won’t have to fight it at all if you don’t want to, turning this once mandatory boss fight into an optional encounter.

Also - remember that expensive simple key from earlier? Well, 500 rosaries is a tall order at this early part of the game, but it’s not impossible with a little grinding and farming. And purchasing this key lets you really throw a spanner in the works.

For one, we can open a door in the Deep Docks. This lets us sneak through a bunch of rooms and pop out in the Far Fields - bypassing that boss fight with Lace entirely. Again, turning this essential encounter into a completely optional fight.

Or, even better - we can go through a door in the Wormways. This lets us climb up to the top of Bone Bottom and enter Shellwood from the western side, meaning we’ve skipped Lace, the Drifter’s Cloak, Fourth Chorus, and Moorwing.

(Though, we will need to go get the Drifter’s Cloak at some point as it’s needed in Act 2).

All of this stuff is totally intended and planned by the developers at Team Cherry, of course. Like, if you skip Lace by either method, she’ll show up in a later location with a unique cutscene to tell you that she’ll save the fight until later.

But then there’s the biggest skip of all.

You see, among the many new places we can go with the Cling Grip, is Sinner’s Road. This is a whole new area, off the northeastern corner of Greymoor, that’s filled with nasty guard dogs, swinging traps, and pools of acidic goop that cover Hornet in a bile that slowly saps her silk. This place sucks.

But within this area is another, even more hidden zone, called The Mist. It’s one of those, you know, Lost Woods type of deals - a loopy, impossible maze where taking the wrong path will just whip you back around to the entrance.

So you’ll have to figure out that playing the Needolin will send these fireflies off towards the correct door. And if you successfully follow them a bunch of times you’ll end up in an entirely new place called The Exhaust Organ.

This is the home of another boss - Phantom, who is perhaps the closest to the Hornet fight in the original Hollow Knight. And defeating him not only unlocks another silk skill, ooo, a parry - but - surprise! - it also opens a path to the Citadel, and Act 2 - meaning you don’t have to fight The Last Judge.

That’s right - there are two completely different ways to finish this act and enter the Citadel. There’s the typical route through the Blasted Steps and past The Last Judge, and the more hidden route through Sinner’s Road and past Phantom.

And I like this a lot - because if you’re struggling with The Last Judge (hello!), the game gives you an alternative route to tackle instead.

One that’s less about sheer boss fighting skills - I mean, Phantom is a lot easier than the Last Judge, and has less total health and does less damage with each hit. But is, instead, more about exploration, secret-hunting, puzzle solving, and survival.

Plus, it just makes the world feel more complete and interesting. Of course there’s not just one single entrance to the Citadel through some massive gate. There’s also a sneaky side door by breaking in to its Underworks.

And so here’s the thing - when I first played through Silksong, taking that predictable path I showed in chapter one - I was ready to make this video and say that Act 1 is very linear. A strict sequence of events that must be followed exactly as Team Cherry intended.

But it turns out there are sequence-shattering secrets all over Act 1 that let you tackle the game in all sorts of ways.

Which is great. This offers different ways to play in future playthroughs. It rewards those who explore, or save up money, or find fleas. And while most players will take the quote-unquote normal path, it gives those who do stumble upon these alternative routes a very unique and special experience.

And this doesn’t even mention that you can also skip the Bell Beast, Sister Splinter, and Fourth Chorus... but with cheeky jumps and pogos, that are more like speed-running exploits than normal Metroidvania choices. So I won’t go into them properly in this video.

All of this means that Act 1 is perhaps more open-ended and non-linear than you might first think. With more optional bosses and alternative routes, and a more player-driven structure, than you initially realise.

But if you think Act 1 is open, just wait for Act 2.

Part 3 - The Citadel

Okay, now that we’ve made the pilgrimage up through Pharloom - however you ended up doing that - it’s time to tackle The Citadel.

This area is shown as just one big blob on the zoomed-out area map - alongside the Cradle and the Underworks - but, inside The Citadel, you’ll realise that it’s made up of a bunch of distinct areas.

Well. Sort of distinct.

To me, Pharloom - the land of Act 1 - very much feels like a bunch of different zones. Each one has its own colour palette and vibe and can be easily identified from a screenshot.

But in the Citadel, the different areas feel slightly more same-y, and I felt like I was exploring one single space rather than lots of different areas.

Though, I’ll admit that each area does have its own ideas. The Cogwork Core is an industrial zone with spinning gears. The Whispering Vaults has box-shoving puzzles. The High Halls are all up-and-down elevators. And Whiteward is a small slice of a psychological horror game.

Plus, there’s some great environmental storytelling here - like how the enemies in the Choral Chambers carry loads of rosaries, but the worker drones in the Underworks have hardly any. And how the benches down here are pay-per use, showing how the staff are getting squeezed dry.

So perhaps I’m talking my way out of this position. But still - as I reminisce on the game, I can’t help but think of Pharloom as a bunch of distinct places and the Citadel as a single zone.

Maybe it’s like exploring a Metroid map, which then leads to a Castlevania map. Perhaps Pharloom is Zebes, and the Citadel is Dracula’s Castle. That could work.

The Citadel also feels - suitably - self contained and disconnected from all of the stuff we just did. For starters there are separate fast travel systems, with the Bell Beast in Pharloom, and pneumatic elevators in the Citadel - and only one connection between the two. So going back to the Act 1 areas has the extra step of getting to the Grand Bellway first.

Plus, there’s actually no reason to ever return to Pharloom once you’ve reached the Citadel. Now I’m sure you will for wishes, pick-ups, needle upgrades, and whatnot - but all of the upgrades, bosses, and paths forward through Act 2 are found exclusively within the Citadel and the Underworks.

But the biggest difference between Acts 1 and 2 is in how progression works.

Act 1, as we just explored, certainly offers variance - but that was by skipping and bypassing bits of an otherwise linear structure. Act 2, on the other hand, is completely open-ended by design.

Okay, so both paths into the Citadel start in the Underworks. But once we’ve made our way through there we’ll enter the actual building. A grand, opulent palace with stages, dining halls, and a spa.

At the very top of the Choral Chambers is a boss fight against the Cogwork Dancers. Side note: what a brilliant boss. That’s, like, an all-timer right there.

But anyway. Beating these ballroom bots gives us a new quest. We need to collect three bits of a melody and then bring back the completed tune.

These three song-lets are spread across the Citadel. And each one tests a different skill.

The Conductor’s Melody in High Halls is found after a frankly ridiculous enemy gauntlet with 10-or-so phases. The Architect’s Melody in Cogwork Core requires serious platforming skills as you bounce over deadly gears. And the Vaultkeeper’s Melody in the Whispering Vaults is behind a more traditional boss fight - against the theatrical ham, Trobbio.

Best of all, these three melodies can be found in any order. Lemme show you how.

So, after we defeat the Cogwork Dancers, we enter the Cogwork Core. This gives us access to the Whispering Vaults where we can fight Trobbio - which, in turn, gives us access to the eastern side of the Underworks, where we find the Vaultkeeper’s Melody.

(It’s actually just a cylinder, so we’ll need to take it to Vaultkeeper Cardinius, also in the Vaults, to actually play it. But whatever).

So that’s the first part of the chart.

Now, the other two areas with songs - the High Halls, and the upper part of Cogwork Core - are locked until we get the next ability.

That would be the Clawline - a speedy hookshot-style ability, which can also latch on to hanging rings. This thing is in the Cauldron, on the east side of the Underworks. But there are actually two very different ways to reach it.

For one, we can get to it after defeating Trobbio. We’ll take this elevator down to the Cauldron and get the Clawline from there.

Or, we can find the White Key in Songclave, use that to open Whiteward, work our way through, and then drop down into the Cauldron from that area.

So... take your pick.

Now, I love a Metroidvania upgrade that can be found through multiple routes. It gives players more agency, and it doubles their chance of finding it through exploration - or simply stumbling around.

It also means that we don’t have to fight Trobbio first, as he’s not necessarily between you and the Clawline. And it means that it’s actually possible to get to the High Halls, finish that tricky gauntlet, and meet Conductor Ballador before we’ve even defeated the Cogwork Dancers and gotten the quest to find the three songs.

We’ll just have to come back for the melody after the quest begins.

So, you can hopefully see now that Act 2 is a very different beast to Act 1. Not only do you have three destinations, which can be completed in any order you like - but the whole business with the Clawline creates a tangled spaghetti mess.

With the outcome being: how you finish this act is very much up to you.

But you’ll still unlikely to get lost. You can always rely on the map-poking technique I explained earlier. And the game even puts icons on your map to show you where the melodies are. They don’t really spoil the game, if you ask me - they’re just there to stop you getting completely lost, as you’ll always know the general location of these all-important tunes.

And so once you have all three pieces of the melody, that is, well, ultimately... the beginning of the end of the game.

We can now take this elevator up, to the Cradle. Here we’ll fight Lace - again, or for the first time, depending on what when down in Act 1. And then continue through the Cradle to face down the final boss of Hollow Knight: Silksong: Grand Mother Silk. Beat her. Roll credits. You’re done.

But - in truth, this is not an accurate representation of playing the game. Not at all. Because we really have to talk about...

Part 4 - Optional Goodies

So far, I’ve been talking almost exclusively about the main progression of Hollow Knight: Silksong. You know: the things you have to do to advance the story, finish the main quests, and get to the end of the game.

But that is only a small part of what this game has to offer.

Because here’s the thing. In every Metroidvania, there’s the stuff that you have to do to finish the game, and there’s the optional stuff you can do that doesn’t impact the progression either way.

For instance, in Metroid Prime you’ll need the high jump boots, the Varia suit, the boost ball, the missiles, and so on to overcome the obstacles between you and the final boss.

But there are also energy tanks, missile upgrades, and even special weapons like the Wavebuster which all make Samus more capable, but they aren’t required for any locks, bosses, or puzzles. So: totally optional.

That means very Metroidvania has two piles: the mandatory pile and the optional pile. And in Hollow Knight: Silksong - well, the optional pile is massive. Like whopping. Like astronomically big. The amount of stuff you CAN do far outweighs the stuff you HAVE to do.

I mean, to put it in numbers - there are 12 bosses that must be defeated before finishing the game. Moss Mother, Bell Beast, Lace, Fourth Chorus, Moorwing, Sister Splinter, Widow, The Last Judge - or Phantom, Cogwork Dancers, Trobbio, Lace - again - and Grand Mother Silk.

And, yeah, that includes all the ones that can technically be skipped.

But then are 17 more bosses on top of them. 17 beautifully drawn, designed, and developed baddies that are totally optional and do not stand between you and the end credits. 17 boss fights that you just might never come across. That’s wild!

Another way to think about it is in terms of items. All the goodies and pick-ups and bits and bobs you can stuff in Hornet’s pockets. All of this stuff is optional - but just think how much crap there is in this game.

There are 20 mask shards that increase your health, 18 spool fragments that boost your mana, 8 pouch upgrades, and 3 bottles of pale oil to increase the power of Hornet’s needle.

There are 6 different silk skills, which are powerful combat moves. Seven crests, which change Hornet’s moveset and unlock special attacks. There are 20 red tools, which are sub weapons. 21 blue tools, which help in combat. And 12 yellow tools, which help with exploration.

Plus 20 memory lockets, which let you equip more tools at the same time.

There are also 3 silk hearts to improve mana regeneration, a charge attack move, endless bundles of the game’s two currencies - rosary beads and shell shards, plus rare objects that can be sold to merchants for even more cash. There’s craftmetal which can be used to make special tools. There are fleas, simple keys, and bags of frozen peas. There’s so much stuff.

Some of these things are simple upgrades that give you more health, mana, ammo, and money. But others completely change the game - giving you a chance to make a totally unique build that works for you and the way you like to play.

It is a lot.

And all of this stuff is going to be useful because Silksong is, famously, a very hard game. A game with gruelling boss fights, exhausting enemy gauntlets, and difficult obstacle courses.

Now Team Cherry knows this - but has always said that the solution to getting stuck in a Hollow Knight game is to just... go somewhere else. To stop beating your head against a brick wall and instead try another route.

Ari Gibson says that players who get stuck “have ways to mitigate the difficulty via exploration, or learning, or even circumventing the challenge entirely, rather than getting stonewalled”.

So, as Ari said, sometimes that means skipping the fight through a cheeky secret path. Other times, that means finding someone to help. For instance, you can find Garmond and Zaza in a side room in Greymoor, and get them to assist in the fight with Moorwing.

But the most obvious way to help improve your chances is to find some of these optional items and goodies. Some extra health or a tool that speeds up healing can be the difference between success and failure.

Though - I do want to note that this intention, for players to go somewhere else if they get stuck, is largely undercut by Silksong’s use of corpse running.

That’s a gameplay system where - when you die, you’ll lose some of your stuff. Specifically: all of your loose rosaries - the game’s main currency. And also any spool upgrades you’ve found.

So you’ll need to go back to the place you died and smack your cocoon to get your stuff back.

This idea was popularised by Dark Souls, and then borrowed for Hollow Knight. And the idea is once you’ve died, you’re in a heightened state of play because you’re less strong - and if you die again before getting back to your corpse... you’ll lose your stuff for good.

So it can be a good tool for amping up the stakes. But I just don’t think it works for a Metroidvania. We want players to die... and then go explore somewhere else instead. But the corpse running system just encourages them to keep going back to the same place, over and over again.

Now, Silksong does have some ways to reduce its impact - like these tools that turn loose rosary beards, that are lost on death, into strings of beads, which aren’t. And the Silkeater item gives you a few chances to retrieve your corpse.

But overall - I just don’t think this is a good system for this sort of game. I didn’t like it in Hollow Knight, I didn’t like it in Elden Ring, I don’t like it here, either.

But, okay, another thing worth noting is how many items you can possibly find before each boss fight.

For instance - in terms of mask shards, there are 20 total in the game, meaning you can double your health with 5 additional masks.

But the vast majority of them can only be found in Act 2. And that’s true of almost every other pick-up, too.

So if we consider the bosses at the ends of Act 1 and 2, The Last Judge and Lace. Ignoring Phantom for now.

Well, before fighting The Last Judge we can theoretically find 6 mask shards, (which gives 1 extra mask) 6 spool fragments, 1 needle upgrade, and about 25 tools.

And considering that the Last Judge does two masks of damage with every hit, one extra mask doesn’t help all that much.

But then before fighting Lace, we can find 10 more mask shards, adding 4 more masks, 12 more spool fragments, 2 more needle upgrades, and almost every tool in Silksong.

That makes a big difference. Like, there’s just not a huge amount of stuff we can do to improve our chances against The Last Judge - the biggest one being this magma bell that nerfs incoming fire damage - but when it comes to Lace, well an industrious player can massively tip the fight in their favour.

And so I think this is why I found Act 2 to be almost easier than Act 1. In the first act I felt like I had to just constantly retry boss fights like Moorwing, the Fourth Chorus, and The Last Judge. There wasn’t much else I could do. But in Act 2, if I was struggling with Trobbio or Lace or Grand Mother Silk I could just go somewhere else, get more power, and come back later.

But okay. With those caveats aside, let’s ask the question: where you do find all this stuff?

Part 5 - Walls and Wishes

Okay, so we’ve established that Silksong has a buttload of optional stuff to find. But the important question, in terms of Boss Keys, is asking how players will discover them.

And well, ultimately, there are four main sources for finding optional items in the game - backtracking, secret walls, shops, and wishes.

Let’s start with backtracking.

For the main progression, Silksong almost never asks you to go back on yourself. Act 1 can be finished through endless forward progression. And Act 2 is entirely self contained - with everything you need to complete that quest being found exclusively within the Citadel and the Underworks.

That’s a bit unusual for a Metroidvania, where part of the joy of the genre is zig-zagging back and forth across the entire map. Perhaps unlocking a new ability in one place and then going all the way back to the very start of the game in order to use it.

Silksong doesn’t really do this for its main path. But it does have stuff for those who return to a previous zone with a new ability in tow - it’s just almost always optional.

For instance, there’s a path off of the Marrow with a big beefy baddy who will promptly turn you into mashed potato. It’s teaching you, very early on, that sometimes you’ll want to leave and come back later.

So if you return after getting the dash you can outfox this guy, defeat him, and enter a whole new area: Hunter’s March, which holds a flea and a secret shop. And if you come back with the Drifter’s Cloak, you can fly up here... in order to face the Savage Beastfly and - if you manage to defeat that - unlock the powerful Hunter’s Crest.

And this happens throughout Silksong. Each time you get a new ability you can use it NEARBY to move forward and make more progress. OR you can turn around, go back to somewhere you’ve already visited, and use the ability to unlock something cool.

The Clawline is a particularly good example, actually. Technically, it’s only needed within the Citadel - to open the paths to High Halls and the Cogwork Core, in order to get parts of the Threefold Melody.

But we can also backtrack to the zones we visited in Act 1, and use it to unlock a bunch of new places in Pharloom - including a whole new area called the Sands of Karak, a new part of the Deep Docks, plus a mask shard back in Far Fields.

Oh! And it can also be used in the Slab to reach another whole new zone: Mount Fay.

This is an icy alpine area on the western edge of the kingdom. It’s so cold that Hornet will take damage if you’ve been exposed to the elements for more than a few seconds. So you’ll need to dash between these heating lights to stay warm, and reset the countdown.

At the very top of the mountain is, actually, another upgrade - the Faydown Cloak. This fuzzy winter coat not only protects you from the freezing temperatures, but also gives you a double jump.

And seeing as I’ve only just mentioned it, after talking about how to finish the game, you’ll hopefully realise that this thing is - again - optional.

I mean, if you do find it while playing Act 2 then it will certainly come in handy. It will make the Cogwork Core platforming section a lot, lot easier. And you can also use it to find a secret area in the High Halls, where you can stop these silk-sucking marionettes from ever appearing throughout the Citadel - which will help out during that ridiculously long gauntlet.

But, still, the Faydown Cloak is not technically essential and you can finish the game without it.

But, again, with this thing under your belt you can now backtrack across the map and use this new power-up to find a whole bunch of new areas across both Pharloom and the Citadel. Like Wisp Thicket, which is above Greymoor, and Memorium, which is at the top of the Citadel. There are new optional bosses like The First Sinner, Father of the Flame, and Voltvyrm - all of which drop powerful silk skills and tools. And there are lots of useful goodies to find.

So - the point is, Silksong almost never asks you to backtrack - but if you do it, you’ll be rewarded with all sorts of helpful items.

And I think this approach to backtracking works really well. It means casual players can just stick to the main quest and make progress by always moving forward. But it also means that Metroidvania obsessives still get a big advantage for, you know, committing the world map to memory, and being super fastidious about exploration.

Okay, so that’s backtracking. The other way to get these optional items is through secret walls.

All across Silksong are these crumbling walls that you can break open. Or they’re sometimes just cheeky passageways that are covered up by bits of the foreground.

These things defy map poking because, until you find them, they do not show up on your map at all. It looks like you’ve explored the whole space - but, ahh, no you haven’t.

Now these things are everywhere - I counted more than 50 across the map on my second playthrough for Boss Keys. And they hide items, boss fights, shops, and even entire areas.

Seriously - both Bilewater and the Putrified Ducts are hidden behind secret walls, and so it’s entirely possible to play the entire game without ever knowing these places are there. Seriously - I beat the whole game without stepping foot in Bilewater. The only clue was that suspicious empty space on my map.

Though, it should be noted that you are given some help with finding these locations. Bilewater can be discovered through two different hidden walls - in Sinner’s Road and in the Whispering Vaults, meaning you’ve got two chances to find it.

Meanwhile, there’s a side-quest where you track the shopkeeper Jubilana - which shows a trail through a secret passageway in Memorium, and takes you to a hidden area above Songclave. From there, it shouldn’t be too hard to find this breakable wall that leads to the Putrified Ducts.

We can also get optional items from shops. That one’s pretty self explanatory. There are shopkeepers across Pharloom and the Citadel, and each one will sell tools, spool fragments, mask shards, and more in exchange for rosary beads.

So if backtracking and secret walls reward exploration, then these shops reward being economical with money, and not losing your cocoon - and the rosary beads you were carrying - by dying a lot.

And then there are wishes. These are side quests that can be picked up from bulletin boards in the major towns, or by talking to specific characters.

Many of these are simple quests that you’d see in other games. You know: kill specific enemies and gather their spoils. Track down an NPC by following their trail. Find materials. That sort of thing.

But there are more notable ones too. There are wishes with special boss fights, like the Skull Tyrant in the Marrow and the Broodmother in the Slab.

And there’s the wish Great Taste of Pharloom, where you need to find four ingredients across the entire map of Silksong, and deliver a rasher from Bellhart to the Citadel without it disintegrating - all of which gives you the necessary item to boost your needle’s power.

Oh, and there are also other side activities that aren’t listed as specific wishes. Like in the Citadel, you can find three cogheart pieces - which then go together to power up this robot sentinel.

And so between the cogheart pieces, the taste of Pharloom, and the threefold melody - you could be doing three completely different “find all of these objects” quests, simultaneously, throughout Act 2.

Overall, I think side quests fit really well in a Metroidvania. It’s not typically a part of the genre - I can only think of a few examples with them, like Prince of Persia and some of the Castlevania entries on DS. But they make sense.

They give you something else to do if you’re lost or banging your head against a boss fight. They often give you a reason to revisit an area you’ve long forgotten about. And they reward other types of activities like scouring the map, or collecting things.

So - Silksong is jam packed with optional goodies to find. And as I said in the previous section - they’ll prove very helpful in a game this difficult.

But this stuff actually has a much bigger impact than just letting players modulate the difficulty.

Because the real side effect of having so much optional stuff is that it makes the adventure inherently less linear. At almost every moment in Silksong, you can make a choice about where to go and what to do.

You could follow the main quest progression - or you could go and work on a wish, or explore some new area of the map, or backtrack to an old place with a new ability, or retry that boss you skipped earlier, and so on, and so on.

And this makes each player’s experience way more personal.

The areas you visit, the bosses you fight, the secret walls you uncover, the wishes you find and complete, the crests and silk skills and tools you find and use... all of this changes from player to player.

Which ones will you encounter, and what order will you discover them?

By having so much stuff be entirely optional, it means that each playthrough of Silksong is unique - and not at all defined by the developers.

But. With all of that being said. Perhaps I’m being a bit hasty to describe all of this stuff as being truly optional. Because, as it turns out, a lot of this stuff ends up being way more important than I’ve first let on.

Part 6 - Silk and Soul

Okay, so you might have felt like the final boss fight was a bit easy. And that the final cutscene before the credits was a tad bleak. Like... this can’t be the real ending of the game, can it?

And you’d be right. Just like Hollow Knight 1, there’s a lot more to do if you want to see this thing through and unlock the true, final ending.

This all comes in the form of a wish, Silk and Soul, which is given by the Caretaker in Songclave. This quest will kick off a sequence of events that will lead to the real ending of Silksong.

But actually getting wish thing to appear means fulfilling a hodgepodge collection of prerequisites and requirements. Including...

A whole bunch of specific wishes, many involving big donations of rosaries and shards to wish walls. That also includes the wish Trail’s End, which involves buying all of Shakra’s maps - meaning you’ll need to have been to all the major areas in the game.

You’ll need the Faydown Cloak from Mount Fay. You’ll need the Bellhome key, which you get from doing wishes in Bellhart. You need to move the Flea caravan to Fleatopia - which meaning finding the Pale Lake and hunting down 22 fleas. And you’ll need 17 “quest points”, which come from completing miscellaneous wishes.

Now this list is not actually in the game. You can of course find it on Wikis and Reddits and whatnot. But inside the game itself - well, I guess it’s supposed to just be a thing that will trigger when you’ve proven that you’ve done a lot of the game.

Lore wise, I think it’s supposed to show that Hornet truly cares for the people of Pharloom, and has shown this through charity, effort, and bravery. Which is very cute.

But because it has so many specific prerequisites - like the Bellhome key and that one wish with Shakra’s mentor - well, a lot of players may not trigger it, and just want to know how to continue with the game. I mean, it’s quite telling that the first results on Google for “Silksong how” are all about unlocking Act 3.

So perhaps there could have been another, less obtuse, way to have this thing unlock.

But okay. Once we have this quest, the Caretaker has a plan to defeat Grand Mother Silk without plunging Pharloom into chaos. We just need to gather up four items.

We need to get the Maiden’s Soul from the Chapel Maid in Bone Bottom, and the Hermit’s Soul from the Bell Hermit in Bellhart. Both of those are just about asking nicely.

The Seeker’s Soul is a lot tougher because we have to defeat Groal the Great in Bilewater, which is perhaps the worst boss in perhaps the worst area. Joy. And then finally, we need the Snare Setter item, which can be found in Weavenest Atla, back at the very beginning of the game.

What’s interesting is that three of the four items needed are behind hidden walls, or in hidden areas. And as I explained earlier, these things aren’t shown on your map and so can’t be discovered through simple map poking.

And on top of that - this is where Silksong finally makes backtracking part of the main progression. We’ll need to return to Bone Bottom, to Moss Grotto, and to Bellhart in order to grab items that weren’t available on our first trip there.

Basically: now that casual players have seen the ending and left somewhat satisfied - Team Cherry can amp up the challenge for us sickos who want more. Not just the difficulty of the bosses, mind you - but the difficulty of the exploration and navigation.

Though, it should be noted that the Caretaker will at least give you clues for each part of the quest in his dialogue.

So that’s Silk and Soul - it adds another dose of non-linear exploration to the game.

And with the Soul Snare finished, we can fight Grand Mother Silk again - but, this time, we’ll play the Needolin at the end.

This gives us a very different ending cutscene. And if we load up the game again, uhhh, our save file looks rather strange. Let’s load it up.

Part 7 - Act Three

Welcome to act three.

There’s a bit in Hollow Knight 1 where the familiar Forgotten Crossroads area has become corrupted. Useful pathways are shut off, and basic enemies have mutated into more powerful forms.

Well, this is the premise of Silksong’s third act - only it covers the entire map. Entire areas are damaged and destroyed, there are broken pathways, and enemies have a chance of being corrupted by the void - which makes them more deadly and resilient.

It’s a case of reimagining a place that you thought you had come to know well. A twist on the familiar. Kinda like the upside down castle at the end of Symphony of the Night.

And for the first big twist - when we go to use the Bell Beast to fast travel back to Bone Bottom, we’ll find that it has been usurped by the Bell Eater - our first new boss of Act 3.

Killing that pooping weirdo gives us back the Bell Beast, plus a handy new ability. We can now use the Needolin to call out for the Bell Beast’s kids - which warps us back to the nearest fast travel point.

Now I’m always a fan of holding back quality of life stuff until the back part of the game. You wouldn’t want to have this throughout the whole of Silksong - it would make things too easy. But now we’re in the final stretch, it’s nice to remove some of that tedium.

Anyway - while it’s nice to have the Bell Beast back, I think it would have been fun to have to walk all the way back to Bone Bottom on foot, rather than just fast travelling there. It would have been a good way to see how both the Citadel and Pharloom have changed for Act 3, it would give you a bunch of new combat challenges on the way there, and it would be a clever twist on the goal of the first two acts - to descend, rather than to climb.

Whatever the case, once we’re back at Bone Bottom we find the Shamans, who give us a quest to dive deep beneath Pharloom, into the Abyss.

(I guess that’s the twist on the Act 1 goal - to reach the lowest part of the map, rather than the highest. Fine. Whatever. Ignore me).

So it’s time to go back to the Deep Docks, where there’s a diving bell. Now we might have already found this thing in Act 2 - it’s just sitting there, behind a Clawline ring and a boss fight against the Forebrothers Signis and Gron. In fact - that might be for the best as the boss encounter is actually harder if playing in Act 3, thanks to the addition of mutated enemies.

Whatever the case - we take the diving bell down. Deep down. Very deep down. Into the abyss.

This is a slightly more linear section, and also the location of the game’s final ability: the Silk Soar. This lets us fly up vertically. It’s useful for getting to areas that are too high for even the double jump - and it’s a bit like the Shinespark in Metroid, but without the run up.

We can also use this new move to leave the abyss in a thrilling escape sequence. Which is also quite Metroid coded, let’s be honest.

Now we’re back in Pharloom, we can use the Silk Soar to finally reach every single area on the map. Nothing is off limits now. And we can use it to backtrack to various areas and unlock a number of optional goodies.

That includes a mask shard on Mount Fay, a silk skill in the cradle, a new quest involving Mr Mushroom in Pale Lake, and an optional boss fight against the Watcher at the Edge in the Sands of Karak. Who drops a completely useless memento. Great. Cheers. What was the point of that?

But our actual goal is to go back to Bone Bottom and get the real quest of Act 3.

That quest is to gather three hearts by tracking down dead or otherwise defeated warriors - and then using a new Needolin song to pop into their memories. There we can defeat them and win their heart.

Those hearts are...

The Heart of Might. This comes from Crust King Khann in the Coral Tower at the top of the Sands of Karak, past two boss fights against giant Conchflies. Again, like with the diving bell, we might have already been here in Act 2. But either way, we can now use the Needolin to enter Khann’s memories, fight through several tricky gauntlets, and then take down Khann himself.

Then there’s the Heart of Pollen. This one will take us back to the elevator shaft that goes from the Grand Gate to the Underworks. And by using Silk Soar we can leap up to this higher section. That leads us to two bosses in a row - Shrine Guardian Seth (rest in peace) and then Nyleth. This is a tricky platforming-focused fight that will have you bouncing back and forth from wall to wall, over a pit of spikes.

And finally there’s the Heart of Wild. For this one we can go back to Hunter’s March and use the Silk Soar to jump over this statue. This takes us to a new part of the Far Fields and eventually, to a showdown with the operatic Skarrsinger Karmelita. One of the hardest bosses in the game.

Like in Act 2, the locations of these bosses are put onto your map as icons - to help you track them down. But because this is Act 3, the actual entrances to these boss areas are often hidden, and thus not shown on the map. The paths to Nyleth and Karmelita are behind secret walls - and so they’re immune to map poking.

But - like Act 1, there is also another way to complete the quest. You see, there’s actually a secret fourth heart.

If you saved the Green Prince from his cage in Sinner’s Road, using a simple key, you can find him in Greymoor. Here you can enter his memories to find an entirely new zone, Verdania, and fight the Clover Dancers - a clever storytelling callback to the Cogwork Dancers in the Citadel.

Defeating the Clover Dancers gives us the Conjoined Heart which can be used in place of another of the other three - useful if you can’t find the entrance to Nyleth’s shrine, or you suck too much to beat Karmelita. Hello!

So, those are the four hearts we can find. But we only need three of them, at which point we can take them back to the Shamans to kickstart the ending of the game.

We’ll go through Hornet’s memories to get the Everbloom. Then return to the Abyss and fight the real, actual, promise this is the final boss - Lost Lace, before seeing the true ending of the game.

All of which means Act 3 looks something like this... with a few steps being completable in Act 2 if you were being particularly thorough. Like Act 2, this final act is non-linear and free-form - guided by the player’s decisions and abilities.

Oh, and before I wrap up this section - it does mean I’ll need to recalculate that previous claim that there are more optional bosses than mandatory ones in Silksong.

So, at the end of Act 2 there were 12 mandatory bosses and 17 optional ones.

We’ll need to promote the Great Conchflies, Groal the Great, and the Forebrothers to mandatory, as they’re part of the progression through Trail’s End, Silk and Soul, and Act 3.

If we ignore Khann and go after the Clover Dancers, we’ll need to add Seth, Nyleth, Karmelita, Clover Dancers, and Lost Lace as mandatory fights. It’s not looking good...

But! There are loads of optional bosses in Act 3, including a fight with the Pinstress, an annoying chase after a mythical Palestag in Verdania, and a rematch with a tormented Trobbio.

Which all, to my count, means there are 20 mandatory bosses in Silksong, but 25 optional ones. My point still stands! The video still works!

Part 8 - Boss Keys

Okay, let’s bring this all together.

When we consider the entire game, Silksong has a very interesting structure.

Act 1 has this tightly paced, pretty linear progression, with a non-stop parade of boss fights, upgrades, and new areas. But then Act 2 is more loose and freeform - a non-linear sequence where you’re free to wander and explore in a more meandering fashion. And then Act 3 is a mix of the two - with linear sections like the Abyss, and then exploratory parts like finding the three hearts.

This concertina structure, which dramatically bounces between moments of forward propulsion and slower discovery, keeps the game interesting. It’s a constant change of pace that means Silksong never sings on the same note for long enough to get boring.

On top of that, the game constantly offers ways for players to break and bend the obvious structure of the game. Act 1 is full of cheeky skips and sequence breaks. Act 2 has the stuff with the Clawline. And Act 3 lets you pick 3 hearts from a total of 4.

This gives players more agency, it makes each playthrough different, and it gives options for speed-runners and challenge players to tackle the game in a different way.

And then on top of that, there’s just endless optional content to complete. Wishes, bosses, pick-ups, and secret areas.

This makes each playthrough completely unique. It lets you modulate the difficulty of the game - by giving you a way to turn the tables against bosses. And it lets Team Cherry keep the golden path through the game quite simple and straightforward - but provide lots of rewards for the players who explore, who save up money, who poke at suspicious walls, and who backtrack to earlier parts of the game.

And then on top of all of THAT, this is a world that feels coherent and cohesive.

This is not just a bunch of arbitrary zones stitched together for the sake of a Metroidvania. But the world layout makes sense. Each location has a reason for being - and a reason for being where it is. And ultimately the grand story of the game - about this pilgrimage from the depths of Pharloom to the heights of the Citadel - is represented on the world map itself.

The level design and the way you transport through it tells the story just as clearly as any lore dump.

Put all of this together. A changing pace. Alternative paths. Optional goodies. Coherent storytelling. Put all of this together and we get a Metroidvania made with such obvious skill and thought - taking it to the highest echelons of the genre, and making it the new gold standard for other developers to aspire to.

Hello GMTK backers! Here is your Digest for November.

This month was actually a little light on new releases - the biggest new games were COD BLOPS 7 and Kirby Air Riders. So I’ve picked some different, but still recent, titles for the new releases section.

Hope you’re doing well and keeping warm and cozy in these winter months. Chat soon!

New Releases

Read - Call Of Duty: Black Ops 7’s Awful AI Use Is A Watershed Moment - “Without question, Call of Duty: Black Ops 7 is the most egregious use of AI I’ve seen in a mainstream game”.

Watch (41 mins) - Silent Hill 𝑓 & Women as Suffering Machines - “Silent Hill f is the story of women as machines for suffering, of women as goddesses and doormats, and the men who tread upon them”.

Watch (35 mins) - What Is Silent Hill f REALLY About? - “It’s a horror game about the curses we carry in our blood and how we kill them when we transition between the past and future”.

Read - Hades 2’s Story Mistakes Repetition For Revelation - “Hades II Isn’t A Story—It’s Maintenance”.

Behind the Headlines

Watch (19 mins) - The Rockstar Workers Fired Before They Could Finish GTA 6 - “Rockstar claims these firings were due to “gross misconduct”, but the workers have another two words they prefer to use instead: “union busting”.

Read more

On the last episode of Game Dev 101 - my series about the indie game production pipeline - I shared a whole bunch of tips for coming up with ideas for games.

So, now you know what sort of game you want to make - it’s time to start building it, right? You know, make the art, write the code, hire the composer...

Wrong!

In fact, rushing straight into production is one of the biggest mistakes that new game developers can make.

But check this out: before studio MDHR drew thousands of hand drawn animation frames for Cuphead, it made a version that looked like this...

And before Nintendo built the enormous open world of Hyrule for Breath of the Wild, it made a simple top-down 2D version, using sprites borrowed from Zelda 1.

These are prototypes - scrappy, sample versions of games, built to test the idea before starting development.

It’s the game dev equivalent of building a bridge...by first making it out of wood, string, and zip ties.

Now you might be thinking - oh man, I have to make a whole other game, before I get to make the game itself? That seems like a lot of work! What’s the point of doing a prototype, anyway?

And, well, that’s what this post is all about.

And to help me out, I chatted to Luke Muscat - the superstar designer behind games like Fruit Ninja, Jetpack Joyride, and Feed the Deep - to offer some expert advice.

I have made literally hundreds of prototypes at this point in my career and I would say like at least half of them weren’t fun at all. - Luke Muscat

So, this is Game Dev 101, and this is episode two:

What’s the point of prototyping?

Okay, so basically...a prototype lets you answer a question.

And perhaps the most important question of all is...“is this game fun”?

Because, as it turns out, our brains are terrible video game simulators. Almost any idea can sound good in our head. But it’s only when you actually get to play something that you can see whether or not it works.

That’s why Nintendo designers have the mantra “make before we talk”.

Whether its splatting ink in Splatoon or making vehicles in Tears of the Kingdom or coming up with wonder transformations in Mario, they always make something, to test and prove that the game will be fun, before pitching the idea to the higher ups.

It’s just so hard to conceptualise whether things are going to work or not once it becomes a complete system.

We can’t know whether something’s going to be fun without actually playing it and experiencing it, even when we think we have a pretty good imagination for that.

There’s just too many nuances, too many things that are unexpected.

A relatively recent example for me was: I had this idea for like a one-button shooting game and the idea was was like you would shoot like a Uzi and it would push you back at the same time so shooting became your locomotion and your attack.

And it is just really like not fun to move in a backwards position, like always getting pushed away from something didn’t feel assertive in the way that a game about shooting heaps of people felt like it should be.

That one’s so simple, right? Like there’s very little to it, but there is a feeling when playing it that just isn’t right.

But perhaps “fun” is a little restrictive. Because, in truth, a prototype can be used to test almost any aspect of the game.

So when Jonas Tyroller and co were making Thronefall, they certainly made plenty of different gameplay prototypes to see if it was fun - but they also made prototypes exclusively for testing different art styles and camera angles.

You could even a prototype to test out the narrative. At Pixar, the studio will often make the entire film as a scrappy comic book animatic, with stand-in voices from the animators, to they see if the narrative makes sense.

And all before they ever make a single 3D model.

Another question a prototype can help answer is...

Is this idea viable?

The act of making a prototype will give you a lot of insight into what it would be like to make the whole game, and give you a preview of how easy or difficult production is going to be.

I think puzzle games are a really good example.

Like if you’re prototyping a puzzle game, if it’s really easy to make puzzles for the puzzle game, you know you’re somewhere fertile.

If it’s really hard, it’s like, well, okay, this is Clearly, going to be an uphill battle the whole way.

And as Jonas explained in the previous episode - you can make a pretty good guess for how long a project will take, based on how long the prototype takes to build.

Like, I was working on a prototype for a game about running your own video rental store, where you’d decide how to stock the shelves and answer customer queries.

But even that small sample took weeks of work - and showed me that this project would take much more time than I have to spare right now.

And, besides, a few months later, the game Tiny Bookshop launched which is basically the exact same thing but better. So, dodged a bullet there.

And then the other important question a prototype can help answer is...

Will other people like it?

A prototype lets you share your vision with other people - but not as a pitch or a concept. But as something actually playable.

Like, when Rare was trying to get Microsoft to green-light the pirate party game Sea of Thieves, the studio got Phil Spencer and co to play a scrappy prototype version of the game, and filmed their session - to prove the point that this game would lead to cooperation, conversation, and banter.

And so, ultimately, you can use a prototype to help figure out if this game is going to be a hit, once it comes out.

For my roguelike spelling game Word Play I was able to build a simple version of the concept in just two days for a game jam - and then immediately got strong feedback that people liked the game and would play, and buy it, if there was a full version.

That enthusiasm basically kept me going through the rest of the development.

And, I should say, it proved true - the game has already sold more than twice the number of copies as my previous game.

But, of course, it can also go the other way.

I was working on a prototype called Luck of the Draw, which was Luke Be a Landlord meets Super Auto Pets, so, kind of like a roulette wheel, deck builder, versus mode.

And when I was making that prototype, I’m like, this might be the smartest thing I’ve ever made.

This is such a, I can’t wait for everyone to play this, and this is just gonna blow up. This is it, this is like THE game.

And I released it, and no one cared. It was like complete crickets.

And that was perfect. That’s exactly what you want from a prototype. It gave me so much information and it told me this isn’t it. Especially, this isn’t it in a way that I thought it was. I would have gone and spent two years on that game based on me playing it.

But that’s the thing - this failed prototype meant Luke was able to can that game after a few weeks of work - rather than after two years of development.

And so to complete that definition from before - a prototype lets you answer a question, quickly.

Prototypes let you try out ideas, and then throw them away, before spending a lot of time and resources on them. A prototype lets you pivot, tweak, or completely nope out of a project before you’ve invested any actual resources into its creation.

That’s what makes prototypes so powerful, and so important.

You want to learn whether your bridge design works now - not when you’re halfway across the river.

So, if a prototype is about making something quickly... here are four tips for getting to the point, as fast as possible.

Tip number one is...

Don’t make it flashy

If you’re making a gameplay prototype, avoid the temptation to make it look and sound good.

Instead, use ugly programmer art made in Microsoft Paint. Use grey boxes and the default Unreal Engine dummy. Use sprites stolen from other games. Use free assets from the store.

You don’t need music. You don’t need UI. You don’t need hyper extensible, clean, commented code.

All of this stuff will slow you down and lead to wasted work.

A prototype is designed to be disposable, after all, so don’t spend time on stuff that’s only going to be thrown away.

And, besides, it can distract from the point of the prototype - it’s hard to evaluate if a gameplay mechanic is fun if your playtesters are caught up on the story or the art style.

The only caveat here is when it comes to game feel and juice.

The juice can be part of the fun, especially if it is a visceral game. And so if you’re just like, no, we can’t do any particle effects, we can’t do any kind of feedback.

It’s completely possible that you’re missing part of what’s going to help it be fun.

Fruit Ninja. Super good example of that. The prototype for that was really, really simple, but even that just had the splatters.The prototype had some particles to at least be: yes, this is going to be like a viscerally satisfying experience.

On the other hand, it’s very easy to be like, well, okay, all I need to do is add these 20 things. And then the game will be fun.

And so you spend like a month adding these like extra juice, like, and you what you’re doing then is you’re kind of band-aiding over the problem that the core game isn’t fun.

So, I think when you’re creating a game, you kind of have to be the judge of that, but you have to judge it very honestly.

Tip number two is...

Use other mediums

You don’t have to make your prototype in the same engine that you’re planning to make the final game in.

Thatgamecompany made a prototype of Journey in Macromedia Flash before ever making something that would run on a PlayStation 3.

So use whatever tools let you get to the idea quickly.

In fact - you don’t need to make it in a game engine at all. Paper prototyping is the art of planning a game out on, well, paper. When designing the puzzles for the game Storyteller, the developers created cardboard mock-ups of the assets to try out levels outside of the game.

And when I was making Word Play and trying to imagine the gameplay loop, I used Scrabble tiles, and wrote the modifiers on a deck of hand-made cards to help inspire ideas for different upgrades.

In fact, you can use all sorts of materials.

For the upcoming detective game Locator, the developer played a prototype over Discord by sharing sketches of the current viewpoint and then asking his players what they’d choose to do next - kind of like running a D&D campaign to simulate the gameplay.

Meanwhile, Hideo Kojima planned out levels for Metal Gear Solid by making them out of Lego and then looking at them through a tiny camera.

And for a complex simulation game with lots of interlocking systems, you could make that in a spreadsheet, or as a tabletop board game, before writing a single line of code.

And while a prototype is technically something payable... it doesn’t have to be.

Ori and the Blind Forest started as simple animations, so the devs could play around with ideas for moves and attacks. And Arnt Jensen used this cinematic concept video to hire the team and raise the funds to make Limbo.

Tip number three is to…

Keep your prototypes small and specific

A prototype is not just a crap version of the whole game. It’s not supposed to contain every system.

Instead, it’s a tiny sample of a specific feature, mechanic, or idea to prove its viability, before you go on to make it for real.

So keep your prototypes small. And also keep them separated, with distinctly different project files for your gameplay, art, technical, and audio prototypes.

Keeping your prototypes small also allows you to make way more prototypes. Developer Dinosaur Polo Club made almost 20 different versions of Mini Motorways before entering production - with each one offering some new insight into the best way to make the game.

More prototypes means more questions are answered, more decisions are made, and more risks are removed.

And you might find that, through this fast and iterative process - you’ll stumble upon new ideas that you couldn’t have come up with just by brainstorming. Sometimes, the process of writing code and fiddling with numbers and noodling in your engine can spark incredible ideas.

I actually made a whole video about this phenomenon - so check out “The games that designed themselves” for more inspiration.

And then - tip number four is realising that…

You don’t need to prototype everything

I think there’s a misconception that once you create the prototype you’ll have all the answers.

I think really it just gives you confidence that you’re heading out into hopefully fertile soil.

But that doesn’t mean you know what You’re gonna plant out there.

You know, I would say most of my games, like, the prototypes are pretty thin, and then we just go, like, okay, I feel like we’re heading out into a place that has good possibilities and we’ll figure out the rest of it out there.

The prototype for Jetpack Joyride was Literally, just getting the machine gun jetpack in Monster Dash and kind of just removing the level, adding a roof, and then adding some obstacles.

And we were like, okay, this is pretty fun. There’s something here.

But we never prototyped like the challenge system, the mission system, you know, the vehicles all came after the prototype. Realistically, the meat of that game and what made it stand out came well after we were into production.

Because here’s the thing.

Making prototypes can be really fun. In fact, it’s pretty addictive. You’re constantly trying new things, starting new projects, and you’re free from the constraints of good code and polished assets. Plus, there’s always that nagging voice that says... what if the next prototype is even better?

But if you never stop making prototypes, you won’t make a video game.

So an important skill you have to learn is knowing when enough is enough, and when it’s time to actually start making the game.

Which means - these initial prototypes should help answer the biggest and most risky questions. Is this fun? Is it viable? Will people like it?

But as for the hundreds of thousands of small questions you must answer as a game designer... well they can be figured out as you go along.

But, the thing is - you can use prototypes for those questions too.

Because prototyping is not some one-and-done process that must be buttoned up before production. It’s not something you leave behind once you start making the game for real. Prototyping is simply a cost-effective way to test an idea or answer a question.

And so under that definition... well, it can be used at any point of the process.

Want to add a new feature? Prototype it. Want to add a boss fight? Prototype it. Want to add a new level? Prototype it!

Can’t decide whether the game should have a double jump or a dash? Make quick prototypes for both, put them in front of playtesters, and go with the winner.

Thanks to Luke Muscat for his help on this one. Check him out on YouTube and watch him prototype an overly complicated golf game, a Vampire Survivors-like with cars, and more.

Plus thanks to all the other designers who shared footage of their prototypes with me.

But okay. We’ve picked a game idea. We’ve made a bunch of prototypes to prove that it’s fun and viable. Now... we can actually make the game, right?

Well, I’m sorry, but not quite yet.

Because we should take a moment to do something really important, often forgotten, and not at all fun or sexy... I’m talking about planning. That’s next time, on Game Dev 101.

Hey all! Mark here.

So I’ve been thinking. Whenever I make the GMTK Digest I find that I love sharing my recommendations for articles and videos. But I’m actually way less interested in writing about the new releases or the news headlines.

Video game news is just so depressing. It’s all about rich morons wasting money and ruining lives. It’s about layoffs, buyouts from dodgy companies, price increases, and wage cuts. I don’t really follow it and I definitely don’t enjoy talking about it.

And as for the new releases - well, you don’t really need me to tell you that new Pokemon and Battlefield games came out this month. And while I would love to share indie game recommendations... I’m just not playing as much at the moment as I get ready for the birth of my son!

For instance, The Séance of Blake Manor looks awesome, but I haven’t had a chance to check it out yet.

So I’m going to try and rejig the digest a little going forward. I’ll start by finding more articles and stories about the new releases and headlines. Instead of simply telling you that Ghost of Yotei has come out this month, I’ve shared some pieces about its design, development, and reception. Instead of telling you that Tomonobu Itagaki has died, I’ve shared James Mielke’s tribute to the man.

I’ll keep rethinking it as I go along - I want it to provide as much value to you as possible, and that only happens when I’m personally excited to make it!

On the new releases

Read - Ghost of Yōtei – tech deep dive - “Sucker Punch delves into the tech that helped them bring Atsu’s engrossing journey to life”.

Watch (10 mins) - I’m Getting Bored of Every PlayStation Game Telling the Same Story - IGN points out that games like God of War, Ghost of Yotei, and The Last of Us Part II all feature the same theme: revenge.

Read - Searching for Cool in Open-World Games - “Is Ghost of Yotei itself cool? I’m not sure that it is. Being full of cool things does not make a thing cool; that’s a collection, not a quality”.

Read - I’ve Turned on Battlefield 6’s Senseless Destruction - Watching buildings fall down in a city doesn’t feel so fun when the same images are on the news every day.

Read - Pokemon Legends: Z-A makes the world feel livelier than ever before, but it’s not because of Lumiose’s design - “In an age when the vast majority of open world games try to create (or recreate) entire regions to get lost in, the prospect of being limited to a PS2-era-sized map was enticing”.

Read more

Hey. I’m Mark. Let’s talk about destruction.

Let’s talk about smashing cars and levelling buildings. Let’s talk about terrain deformation and fire propagation. Let’s talk about games where things can break and bend and blow up.

But I’m not talking about those fancy pre-canned animations that play as part of a cutscene or scripted moment. I’m not talking about buildings that fall down in the exact same way every single time.

I’m talking about doing destruction in real-time. In response to a player’s actions. As part of the gameplay.

That’s a whole ‘nother thing. That’s really hard.

And it’s hard because you’re dealing with unpredictable physics and particles and lumps of debris and if you’re not careful... you’ll tank the frame rate.

And so that’s what I find interesting about this topic. Doing dynamic destruction in games is about being clever. It’s about using cost-cutting algorithms and outright magic tricks to blow up the world, without blowing your memory budget.

So in this post I want to look at 6 games that have done destruction in wildly different ways - and reveal the ingenious techniques they’ve used to make their worlds fall apart.

I’m Mark Brown. This is Game Maker’s Toolkit. And here’s how games do destruction.

Control

Okay - let’s start with Control.

This is a game where the environment just falls to pieces as you play. Your bullets rip through tables and chairs. And you can use Jesse’s telekinetic powers to rip chunks of concrete out of the wall.

Now this is perhaps the most common kind of destruction in games - almost every shooter is filled with props and small objects that can be broken upon impact. And they all use basically the same technique.

You store two versions of the object in the memory - an intact one, and a broken one. And then when that object is attacked you replace A with B. It’s simple, but it works.

And Control uses the same idea - it just takes it a lot further.

For one, objects are made up of loads of individual pieces that can all be destroyed independently - that means objects can break at almost the exact point where you attack them.

And because those broken pieces are distinct physics objects that are connected with joints - well, if you break a table in half then the table will fall into two parts and collapse in a realistic way.

The game will also play a particle effect when the object breaks to cover up the transition. Props are tagged with different materials like metal, concrete, and wood - and will then play a specific effect on impact.

And finally - decals are used to really sell the impact. Decals are basically stickers that you can slap on the environment - like bullet holes and craters. And control has some really good ones that trick you into thinking that you just made an actual dent in the wall.

Put all of this together and you get an environment that is chaotically crumbling around you in a truly satisfying way.

Astro Bot

Next - let’s talk about Astrobot.

So I just said that the way to do destruction is to secretly prepare a broken version of whatever object you want to destroy. But it turns out that you don’t have to fake the fracture.

So in AstroBot, there are these logs that you can cut with your little laser tootsies to make smaller logs. And they seem to cut exactly where you damage them. What’s going on there? Does the game just have thousands of possible variants of this log, tucked away in memory, ready to swap out?

Turns out - no. The game actually projects a flat 2D plane, underneath Astro Bot, and literally divides the 3D model at the place where you cut it. It then adds new edges and surfaces, creates new collision shapes, and spits out two models to represent the cut logs. All in a single frame.

This is also how Rainbow Six: Siege lets you shoot through walls: it slices up the model in real time. It’s the tech behind Hardspace: Shipbreaker - a game where you use a big ass laser cutter to slice up bits of spaceship. And it was even being used in Red Faction back on the PlayStation 2 to dynamically cut spheres out of the walls.

This turns out to be quite expensive on the hardware, so it comes with limitations. It only works on simple shapes, and the games will typically hide the broken pieces after some time to avoid clogging up your memory. But still: in certain circumstances, you can do the destruction in real-time, instead of preparing it in advance.

BeamNG.drive

Okay, so physics objects in games typically use something called rigidbody physics. Which means they always retain their exact, rigid shape. But other games use soft body physics to create something more malleable and bouncy. It’s the technology behind squishy games like World of Goo and Jelly Car.

And this idea can be scaled up to create a realistic vehicle - as seen in BeamNG.drive - which can lead to extremely detailed and realistic deformation.

Here’s how it works. Every car in BeamNG is made up of a bunch of nodes. Just floating points in space. And then they’re held together with beams - which act like stiff springs that try not to stretch.

So when one of those nodes is moved by a collision, it will yank on all of the connected beams, which pull in all of their nodes - creating that realistic crumbled effect.

And so where most racing games just shaders and textures to represent damage in a purely visual way, soft body games like BeamBG and Wreckfest can literally squish and squash the model of the car to fundamentally change how the object moves, and interacts with the environment.

There’s so much more to BeamNG like aerodynamics and fuel consumption and so on - but most people just use it to recreate that one scene from Final Destination.

Far Cry 2

Okay. Let’s go to Far Cry 2. This game is famous for its fire propagation system. In this game a molotov cocktail can set fire to a patch of dry grass and it will spread into a raging inferno that will engulf nearby trees and buildings.

How does that work?

Well, basically, the game world is represented by an invisible grid that stretches over the floor. And each cell has a certain number of hit points. If the cell is flammable and loses all of its hit points - say, when you shoot a rocket at it - it will set on fire.

Then - a burning cell will start to damage its neighbouring cells. And when their HP reaches zero they’ll set on fire too, creating a chain reaction that will make the fire spread from zone to zone.

There are other bits of nuance to the system, of course. Each cell has a finite lifetime so it will eventually stop burning.

Different environments will have more or fewer hitpoints to represent, say, a dry patch of grass vs a wet jungle - the more hitpoints it has, the harder it is to set alight.

And burning cells will deal more damage to neighbours that are in the direction of the wind, to capture the effect of a fire being corralled by a strong gust.

it’s a deceptively simple system, but creates a really strong effect.

Teardown

Next up - Teardown.

But first, let’s start by looking at 2D games. So in a game like Spelunky, we can see that the world is divided into a grid of tiles. Then we create a simple destruction system by removing those tiles to create tunnels, pathways and chutes. The devs can just swap out the sprites of the neighbouring tiles to hide the rough edges.

And if you want more fidelity in your destruction, you can just reduce the size of the grid. In fact, you can go all the way down to destroying individual pixels as seen in Noita. In this game, every single pixel can be removed in order to dig through walls and floors with practically infinite granularity.

And so if we take that grid of tiles and extend it into an extra dimension we get... well, Minecraft. Minecraft is built out of cubes called “voxels”, which can be independently added and subtracted from the world to allow for completely free-form destruction.

And, in fact, lots of games use this voxel approach for destruction - including No Man’s Sky, 7 Days to Die, Astroneer, and the terrain smashing Donkey Kong: Bananza, where every layer is made up of hundreds of cubes that are packed together like dirt.

Most of these games use smoothing techniques and algorithms to avoid that super blocky Minecraft style in favour of something more naturalistic.

And then - like how we went from Spelunky to Noita by just reducing the size of the blocks, we can do the same in 3D to get a game like Teardown.

Teardown is basically Minecraft but with way higher resolution. Where Minecraft represents a piece of furniture as a single cube, Teardown represents it as hundreds or thousands of tiny blocks - all of which allows for much for precise destruction.

Though the other difference with the other voxel-based games is that Teardown has physics. There are no floating platforms here!

Basically, the game makes objects from a cluster of voxels. But if some of those voxels get loose from the cluster they are instantly turned into a new, rigidbody physics object - separate from the original. This means it can fall down, hit into other objects, and roll around.

Red Faction Guerilla

And then finally... it’s the reigning champion of video game destruction. Red Faction Guerilla.

This is a game where almost every structure on the Martian surface can be brought down with an explosive - or just a big swing of your trusty hammer. This is a game where buildings react realistically when you remove their foundations. And can even fall down and crush other structures in the process.

Now there are other games where you can destroy buildings. We just saw it in teardown. There are small structures in the Just Cause games that you can topple. And there’s fan favourite Battlefield: Bad Company 2.

In this game, buildings are made out of a large number of prefabricated pieces, each with two versions - an intact one, and a destroyed one. and when you do enough damage it will, you know the trick by now, swap the piece out for its destroyed variant.

So, that’s all cool, but all of these games struggle with a fundamental question: when does the building fall down?

Teardown does it when the object is completely detached from the ground - which means massive buildings can be held up by a single strand of voxels. And Bad Company does it after a certain number of those prefab pieces are destroyed - at which point it plays a pre-canned destroyed animation - but that all feels kinda arbitrary.

Ideally, we’d want to perform some kind of structural analysis to see if the building should stay standing, or whether it should fall. And that is exactly what Red Faction does - with a system that developer Volition calls “stress”.

Here’s how it works.

So, a building in Red Faction is made up of hundreds of individual parts that you can destroy. Let’s think of them like bricks. And then each brick has two stats. It’s force - which, for simplicity’s sake, let’s just think of it as how heavy that part is. And it’s strength - how much mass it can hold up.

We can then group all of the bricks into individual layers. For a simple building that will be a bunch of horizontal tiers like a big stack of pancakes. But it can be more complicated for horizontal structures like bridges and overhangs.

Then, we can tally up the stats on those bricks to get each layer’s combined weight and strength - keeping in mind that the weight must include everything above it.

And now we can do a stress calculation by simply asking - is the strength of one layer greater than the weight of the all the layers above it?

Now a building is designed, by Volition, to be architecturally sound and hold up its own weight. With different values for different materials - so a giant supporting concrete pillar has a much higher strength value than a pane of glass.

But, as we start destroying bricks we’re reducing the strength of that layer. And we can keep going until the strength is now less than the force of everything above it... at which point, the layer will fail and everything above it will turn into a bunch of physics objects that can come crashing down.

If you go frame by frame, you can see the exact moment where the upper layers change from static stuff to floppy physics bits.

And because those physics objects can inflict damage themselves, we can get into a situation where the top layer of the building will fall down and crush the layers beneath it. Or even fall to the side and make a big hole in the building next to it. Super satisfying.

Now there are plenty of limitations to this system. The game actually takes a while to actually do the stress calculation, which is why buildings sometimes take a while to fall down. But still - this technique allowed for an unmatched level of realism, detail, and player freedom... and that’s despite it being over a decade old, and running on a PlayStation 3. Which I think all goes to prove what I said at the start - that destruction is about ingenuity. It’s not a problem you can solve by just throwing more processing power at it.

....right?

Well, that is true. But what if we took Red Faction’s system and added the last 10 years of software and hardware advancements to it. Well then you’d get The Finals.

Yeah, this is a multiplayer arena shooter ala Overwatch, buy with truly absurd levels of destruction. Every building can be busted, blown up, bifurcated, and turned into bits. Over the course of a match the game world will crumble to pieces in reaction to rockets, mines, and explosive barrels. And it’s even possible to level the entire play space.

And the thing is - the system is basically identical to Red Faction. The developers were directly inspired by Volition’s work and so they used a very similar set-up. The buildings are made up of hundreds of fractured parts. The game calculates stress in real-time. And if a layer’s force exceeds its strength, then the whole thing comes down.

Except... it’s now running on Unreal Engine 5, with Nvidia PhysX, and the min spec asks for 20 times more memory than you’d find in an Xbox 360.

Conclusion

So those are some of the most effective techniques used for adding destruction to games. Swapping props. Splitting the world into breakable cubes. And performing quick stress tests on buildings. All of which can make for an impressive imitation of real-world demolition.

And now you know the tricks - you might be wondering why we don’t see this level of deformation in every game.

And I think it’s because, well, each one of these techniques has some big knock-on effects for the rest of the game. Not just for performance and memory - but also stuff like lighting, sound, and AI pathfinding - all of which must be able to adapt to a world that’s constantly changing and moving.

And then - on top of all of that - there’s game design.

Teardown designer Dennis Gustafsson says “the idea of a fully destructible environment is compelling for the player but a nightmare for the game designer”. And Noita maker Nolla says “It’s easy to think that having a complex or highly realistic simulation will auto-magically result in interesting gameplay. Often, that’s not the case”.

And that’s because, well, game design is usually about setting limits and restrictions, about rules and constraints. and destruction can throw all of that out of the window.

Donkey Kong director Kazuya Takahashi says “In most games, level design is a bit like designing a course. You’ve got a main route you want players to follow, and you build paths to guide them along it. But in this game, even those paths can be destroyed. If you instead dig a tunnel and sneak around to grab [a Banandium gem] from behind, that’s equally viable. Designing levels with that kind of flexibility in mind was a huge challenge”.

So it’s like anything else that can be added to a game - it must be done for a reason.

Perhaps it’s for new combat gameplay, like denying enemy cover, creating new lines of sight, or collapsing structures on enemies. Maybe it’s to let players be crafty by making tunnels, trapdoors, and slides. Perhaps it’s for creativity, or immersion, or replayability, or a deeper sense of simulation.

Whatever the case, and however you implement it - destruction should be there to support your game design goals, not blow them to bits.

I’m Mark Brown, thanks for reading.

Hello lovely GMTK backers!

Oh boy, the video game business just woke up from its summer slumber! This month saw new Silent Hill, Borderlands, and Sonic games, plus massive indie hits and loads more. Even this big list below is just scratching the surface.

I’ve also got a bumper reading list for you to enjoy - with tonnes of articles and videos to keep you busy. So let’s jump on in.

New Releases

Blockbusters

Hollow Knight: Silksong - It’s the big one. Seven years in the making, Team Cherry is back with the follow-up to one of gaming’s masterpieces. It’s another sprawling, atmospheric Metroidvania with rock hard boss fights - and a twitchy, precise moveset.

I’ve been glued to this game all month. I’m about 45 hours in and I’m still not done. The world is just so big - it’s dense and detailed, and splinters off into every direction. You explore one tiny cave or unlock an unassuming door and stumble into a whole new world of ideas, enemies, and secrets.

And, yeah, it’s hard. I’ve died hundreds of times. But you can always explore somewhere else - find new powers, get more health, get more silk, try new tools and traps, change your build. Then come back and succeed. Every failure is the start of a comeback story.

I can’t wait to dig deeper for a Boss Keys episode - but I need to finish my own playthrough first!

LEGO Voyagers - A cute co-op puzzle game from the same peeps who made Builder’s Journey. I played a big chunk of this with my wife and I’m sad to say it completed bored us. The puzzles are simple and repetitive, and the building mechanic has fussy controls that makes it tedious to construct a simple bridge. We dropped it before the short run time was up.

Read more

Transcript:

This week, the game design discourse is all about Hollow Knight: Silksong. Which, don't worry, I won't show any footage beyond Act 1. No spoilers here.

But, yeah, this is Team Cherry's much anticipated, much hyped sequel to 2017's incredible Hollow Knight. But pretty much the only thing people are talking about is the game's rather brutal level of difficulty.

For context, this is a Metroidvania about winding through an interwoven map - looking for upgrades and secret passages. There's nothing too strenuous about that. But along the way you'll have to go toe-to-toe with a huge cast of terrifying bosses - many of which will put your skills to the test in a punishing war of attrition.

Now, the first Hollow Knight was pretty tough at times - but Silksong seems to turn the dial further, with even early and mid-game bosses - like the bullying Last Judge at the end of Act 1 - requiring dozens of failed attempts as you learn the patterns, safe spots, and phases you must go through to win.

All of which has left some players feeling dejected and frustrated. Review site VGC gave the game 3 stars out of 5 - praising almost every aspect, but trashing it for a level of difficulty that feels "demotivating" and "sadistic".

And so, with all this chatter going on, I thought it might be interesting to talk about the design of difficulty in games through the lens of Silksong. To look at why game designers choose to implement a higher level of challenge. And - crucially - how they should go about it.

I'm Mark Brown, and this is a GMTK Mini.

Okay - so the first question is, why would a game designer choose to make their game hard, anyway? What can you gain by amping up the challenge?

Well, I think there are three main reasons to do it.

Number one is that the harder a challenge is, the more satisfying it feels to finally overcome it. If you breeze through a boss fight first time, it might barely even register as something that happened. But if you struggle against a boss for an hour, slowly improving your skills, your reflexes, and your knowledge with each failed attempt, and then you finally topple that monster - well, that moment of triumph is something you might never forget.

That's the idea behind the Dark Souls series - creator Hidetaka Miyazaki says "I just want as many players as possible to experience the joy that comes from overcoming hardship".

And Hollow Knight: Silksong certainly plays into this. When I finally beat the Last Judge, my palms were sweaty, my heart was racing, my eyes were straining from not blinking - and I let out an embarrassing, involuntary noise like "ahhh!" to mark my victory. You don't get that with an easy game.

The second reason to amp up the difficulty is to tie in with the game's narrative or aesthetic goals. Like, if a game tried to tell you that this kingdom is a place of misery and death, a no man's zone where hope goes to die. And then you play it and it's full of enemies who die in a single hit... well that's going to be pretty dissonant.

So if you want to tell a story about suffering and failure, then the gameplay should reflect that. That's how Celeste works - the game is about hero Madeline stubbornly persisting through hardship and setbacks as she climbs a legendarily high mountain. And so the player must also stubbornly persist through hardship and, well, you get the idea.

Silksong plays into this too. The game's world - Pharloom - is a withered, haunted land. There is death everywhere. Other characters are seen failing on their pilgrimage to the Citadel. Main character Hornet must grow in strength and wisdom to succeed. And so must you.

And the final reason for increasing a game's difficulty is to ensure that the player approaches the game in a certain way.

Take the game Furi - a game about being patient and observant as you dodge and parry every incoming attack. Before exploiting a rare opening to deal some damage. The game is designed as a brutal back-and-forth ballet of defence and offence.

But if you could also just run up to the boss, dodge a few simple attacks, and then wail on it until it dies... well you wouldn't have to do any of that stuff I just said. And the game would be pretty dull. So in Furi, difficulty acts like a teacher: punishing you if you play it wrong, rewarding you if you play it right.

Where easy games let you get away with simple and repetitive strategies, hard games require you to use every mechanic, every system, every tool, and every inch of the level design.

And Silksong does play into this also. You'll need to learn every boss's patterns, make use of every charm and tool, and understand Hornet's moveset on a deep, molecular level.

So - I think those are the three main reasons to make a game more challenging. To increase the sense of satisfaction. To mirror the storyline. And to discourage boring playstyles. All of which turns difficulty into another tool that a game designer can use to create the exact experience they want.

But! But but but. Big but. A tool can't be used carelessly or thoughtlessly. I mean - it's simple to make a really hard game. Just put spikes everywhere and send the player back to the beginning of the whole game every time they fail. That's done - but you've now alienated 99% of players.

No. Instead, difficulty is a careful balancing act. We want players to feel challenged. Maybe even frustrated. But we also want them to get that moment of catharsis from winning before they go all the way into rage quit.

Now, I think if a game feels fair - like you can see how you might learn from your mistakes, you can feel yourself getting better, you blame yourself for slip-ups, and you have ample opportunity to repeat and improve - then the game can keep you in that challenge zone for a long time.

And oftentimes, this is where Silksong lands. I spent a long, long time fighting the Last Judge - but I kept going because I could see a path to victory. I was lasting longer and longer. And mistakes were my mistakes.

But if the game is bullshit, then the player is going to dip into frustration, and perhaps even rage quit, much more quickly. And, sadly, Silksong has a fair amount of BS.

Things like the long, repetitive journey to the boss fight each time you repeat. Things like unpredictable troll moves. Things like boss encounters that use extremely random patterns. Things like how the ammo for your secondary weapons don't regenerate upon death, so you'll have to farm enemies to top up. All of this stuff... this is what pushes players to get mad. To stop playing. To write reviews where you call it sadistic.

And that also means it no longer fits one of those design goals from earlier. The first one - where overcoming a tough challenge feels rewarding. Because the player either doesn't overcome the challenge at all, because they stopped playing. Or they feel a different emotion when they finally win - just sheer relief that they'll never have to do that again.

So as a game designer, we can't think of difficulty as a single monolithic thing that we can add or remove from our game. But instead, it's a broad category that includes a whole catalogue of different tools - each one changing the game, and the player's relationship with it, in a unique way.

But there's another type of tool we can use. And that's giving the player an opportunity to change the difficulty in some way. To decide just how hard or easy the game should be.

Perhaps the most obvious version of this is to provide difficulty modes, or assist options that let you modify the game. And Silksong - like the original Hollow Knight - does not include anything like this.

Which, in some ways, is a shame.

Because, well - difficulty is a really subjective thing. What's "challenging" is different for every player. If you asked me to run a 100m dash in, say, less than 12 seconds that would be a really tough, but plausible challenge to overcome. But it would be practically impossible for a young child. And completely laughable to Usain Bolt.

Difficulty adjustments let everyone find a place where they feel challenged.

And also - there's more to Silksong than just beating bosses. This is also a game about untangling an interconnected web of maps and areas. And a game about understanding the deep lore of its storyline. It would be a shame to deny people of those aspects just because they can't beat the bosses.

It reminds me of Tunic. My wife loved playing this game and uncovering its many secrets and riddles. But she simply would not have been able to play it without the accessibility adjustments that let you tone down the difficulty of its crushing Souls-like boss battles.

But - with that being said, Silksong does have one big thing that allows you to adjust the difficulty. And that's being able to explore elsewhere.

You see, one of the bosses that pushed me into the frustration zone was this guy - the Savage Beastfly. It's not actually a very difficult fight - he has pretty simple patterns to avoid and exploit. But, he also spawns in loads of enemies who just get in the way and drain your health. The overlapping patterns make it almost impossible to predict where will be safe.

But, instead of deleting the game and cancelling Xbox Game Pass, I decided to just explore somewhere else. I found new powers, more charms, I increased my maximum health. And when I finally went back to the Beastfly I was able to defeat it a lot more easily.

And in many ways, Silksong is designed around this idea.

The game is full of completely optional bosses, and there are very few times where a boss acts like a complete bottleneck to progressing further into the game. I'll explore this more fully in the inevitable Boss Keys episode on this game, but, basically: you can almost always go somewhere else and come back later.

Also - Silksong has a mechanic where you lose your cash - your rosary beads - when you die and you've got to go back to your cocoon corpse to pick them up. It's a mechanic that, in most games, forces you to keep going back to the same area again and again. But it's not as egregious as other games. Like, in Hollow Knight 1 you'd have half as much mana until you found your corpse. Not so in Silksong. You can also find ways to bank your rosary beads. All things that make it okay to leave your cocoon behind and go somewhere else.

And perhaps you could even point to the way ammo works. Like I said before, ammo doesn't regenerate on death so you'll need to go and kill some enemies to top up your reserves. This breaks the pattern of going back to the boss over and over again, and may encourage you to just do something else.

In all of these ways - and more - Silksong is telling you that the solution is almost never to just keep hitting your head against a brick wall until you get through. Instead, if you get stuck you should explore. Go elsewhere. Get more powers. Maybe you'll find a friend who will help you in battle. Or that one special item that will make the boss fight easier. Or maybe you just need a break.

So - that's where I'm at.

In a nutshell: difficulty is a tool that game designers can use to craft a specific experience. But difficulty is a huge set of disparate tools - which must be deployed carefully. And finally, it's always worth thinking about how to give players ways to adjust or circumvent the difficulty.

Right now, Silksong is still in my sweet spot - it's tough, but I'm surviving. But how about you? Let me know in the comments down below. See you soon for a new episode of Game Maker's Toolkit.

I'm technically on holiday today. Taking some much-needed time off after the madness of releasing Word Play and the Game Jam. But I couldn't leave you without the Digest.

Enjoy!

New Releases

Blockbusters

Metal Gear Solid Δ: Snake Eater - Konami's flashy Unreal Engine 5 remake of that all-timer PS2 game. It's been described as a super faithful remake that barely deviates from the original. But I guess when Hideo Kojima has, ahem, moved on, it was probably best not to mess with things.

Mafia: The Old Country - Hangar13 returns to Mafia for this prequel, which is far more linear than the open world Mafia 3. I've seen it described as a throwback to mid-budget Xbox 360 cover shooters - but there's also plenty of praise for its storyline.

Drag x Drive - It's Nintendo's tech demo showcase for the Switch 2's mouse mode (you use two joycons to simulate pushing wheelchair wheels). While I'm always down for accessibility-adjacent stuff, the reviews are rouuugh. Gamespot calls it "tiring and imprecise".

Sword of the Sea (Pictured) - Giant Squid (Abzu, The Pathless), return with "an atmospheric surfing adventure with fast movement inspired by classic skateboarding and snowboarding games". I guess Journey meets Tony Hawks, which sounds good to me.

Smaller Stuff

Öoo - The developer of Elechead returns with another super smart Metroidvania platformer. This time you're a caterpillar made of bombs who can blow up its own body to get around and solve puzzles. Looks to be a design tour de force.

Read more

Earlier this month we held the 9th annual GMTK Game Jam which is… well, you know the drill by now. It’s a game-making marathon where people race to make a game in just four days, and that game must fit within a specific theme.

As always, we saw even more people enter the jam than last year, with more than 37,000 participants making over 9,600 games. Which means that it’s still the biggest jam itch.io has ever seen. And boy, do their servers feel it.

So, what was this year’s theme? Well, it was a single word: loop. And so we saw thousands of games about programming loops, about musical loops, about running loops around people, about loops in time and in space. The way you all managed to take this theme and spit out so many distinctly clever, imaginative, and fun games… well, it’s absolutely inspiring.

And now it’s time for me to pick my 20 favourite games from the jam…

Now, normally this is where I’d say I’m choosing the “best” games from the jam, or even the “winners” of the jam. But the truth is… as GMTK Jam grows bigger and bigger, I don’t think it makes sense for the whole thing to come down to a single judge who really likes puzzle platformers.

Especially when I’ve only played like 1% of the total entries. So, from now on, I’ll let the community pick the winners through rating and ranking the games. You can see those over on itch.io. But I’ll simply choose some personal favourites. That cool? Cool.

Let’s start with Rhythmetric. This is a game about a circle who fights rather more pointy shapes, using laser beams, dodges, and shields. But it’s up to you when those moves are triggered. You see, there’s a track on the right side of the screen, and here you can plop down moves at different points. Then, when the looping indicator scrolls across the track, it will play each move on the beat.

This turns your fighting style into a personalised anthem. Attack, attack, attack, shield… dodge, dodge, dodge. You know, simply by feel, when it’s time to aim at the enemies and unleash a barrage, and when it’s time to swing the mouse in the opposite direction to dodge away. It turns the player into a deadly DJ, mixing between fighting and composing. And then, as you pick up more beats and add to the track, the game naturally gets more complex. It’s addictive, smart, and puts you into a dizzying zen-like state.

The next game, Voodloop, has a somewhat similar premise. This time, though, it’s a 2D platformer and we can assign moves like jump, kick, and dash - which will then trigger on the beat. So you’ll need to build the right loop to do double jumps over pits, or to fight simple enemies. What I really like about the game, though, is that each level is like a miniature Metroidvania. You must hunt for new moves to lay down on the track - which then lets you access new areas, in order to find new moves - so you can reach the exit and do it all over again.

Also, by changing the style and tempo of each level’s soundtrack, the game takes on a very different feel from stage to stage. Super smart stuff, with a rather dashing, chunky pixel art visual style.

With a theme like loop, we were bound to see a few entries inspired by the ultimate time loop game, Outer Wilds. But my favourite of the bunch is Night O’Clock - a quick-fire murder mystery where you have just one minute to scour a mansion, collect clues, and gather knowledge… before time’s up and you have to start again. A clever twist on the formula is that you keep hold of any item you’ve got when time runs out, so you have to think carefully about, say, which key you want in your hand before restarting the loop.

Night O’Clock is absurdly polished for a game made in four days, and it kept me rapt for the entire run time. An easy recommendation among a sea of games about repeating the same day over and over again.

Now, game ideas can come from the weirdest places. Like, what about that shower thought of… if you plug a power strip into itself, wouldn’t that give you an infinite energy source? Forget conservation of energy and other boring realities - in Infinite Energy, you can cheat physics and unlock an eternal electrical source.

But it’s more than a meme - this is actually a very clever, mind-bending puzzle game. You need to understand a simple set of rules - about where power comes from and where it goes. Deal with restrictions like the short length of your cables. And think outside the box as you make temporary connections and purposefully undo your actions to make everything work. It made my brain hurt… but like, in a nice fuzzy way.

Talk about thinking outside of the box. When developer SI Interactive heard the theme loop, they thought about the looping lasso tool in Photoshop. And so we get PhotoLoop: a game about being a freelance photo editor. Each day you get a request in your inbox. Can you take this photo of me and add hair? Can you block out my ex? Can you add friends to my birthday party? You can then use the loop to select stickers, people, and cartoon characters to place into the photo. Send it back and hope the client is happy.

There’s even a hint of detective work in here, as your clients ask you to figure out relationships from their email. It’s only a sprinkle - but you could see this being expanded into a clever deduction game where you provide your answer through doctored digital images. But as for the game that’s actually here - it’s cute, clever, and as someone who spends their life selecting crap in Photoshop, it was nice to play a game where my pointless life skills are finally validated.

Circuit Breaker starts out as a pretty simple puzzle platformer. Walk, jump, manipulate machines, plop batteries into sockets. But when you finally make it to the far end of the room, you realise you can loop back around to the start and do the whole thing again. Or, you can walk back on yourself and play the level in reverse - all while making changes to the level that may help or hinder your progress depending on which direction you’re going.

It turns a single level into a galaxy-brained spatial reasoning puzzle. I want to reach this spot but it’s impossible from where I’m standing. But if I loop all the way around the other side, I can make my way up! It reminds me of those puzzle box Zelda dungeons I love so much. It scratches a very specific itch, and that’s why I dig it.

Walk It! takes Crypt of the Necrodancer - that excellent roguelike where you have to move on the beat of the music - and turns it into a series of single-screen vignettes. You play as a dancing dude who cuts shapes from pavement slab to pavement slab, while dodging dogs, pushchairs, and Pilates posers. You have to move to the beat. You have to start on the first beat. And you have to reach the goal on the last beat.

It should be effortless. And maybe it is if you have any sense of rhythm. But I do not - so it turned each one of these short levels into an epic showdown where I had to learn the beat, internalise the music, and feel the rhythm before I could finally dance my way to victory. Great animations, great art style, make 100 levels and put it on Steam.

I’m always impressed by people’s ability to make a Zachlike - a game inspired by SpaceChem and Opus Magnum - during a 4-day jam. It shouldn’t be possible! And yet, here we are. This is Cooking by the Numbers, a game where you build an engine to turn one number into another, using nodes that can add, minus, divide, and multiply. But most importantly, there are the conditional nodes: do one thing if the number is greater than 15, and another if it’s less.

With these, we can - going back to the theme - make loops where the same sum is performed over and over until the condition is met, and then a different path is taken through the flow chart. You’re constantly thinking about how to use loops to get the result you want. And - if you didn’t realise it by now - you’re also learning the basics of programming, as you figure out the inner workings of if statements and while loops. You clever clogs, you.

Okay, this game - Caw Across Time - needs a bit of set-up. You get a phone call - it’s you from the future. Future you delivers a key that helps present you get through a door. Handy! But now the space-time continuum is all goofy, so you’ve gotta fix it. But aha - you’re now future you. And you can deliver the key to present you. Who is now past you. Thus finishing the time loop, fixing that nasty rift in the timeline, and letting you move on to the next level.

Did that make sense? Does your brain hurt? Whatever. Keep up. Because this game is really clever. A whole sequence of single-screen conundrums where you have to not only make it through the level, but also tidy up the timeline by putting back any time-travelling keys and blocks you end up using. It’s an absolute belter of a game - and it’s utterly frustrating to play a puzzle platformer made in 4 days that’s better than the one I made in 4 years. Burn this jam to the ground.

Nokia classic Snake got some love in this loop-themed jam. But my absolute favourite take on the idea came from Goop Snake. It’s a Sokoban-style grid puzzler that works on the basic rules of Snake - eat apples to grow in size. Your lengthy body mimics the exact movements of your head. But then it adds loads more ideas. Eating soft apples lets you shift past your own stomach, but hard apples block your head entirely. You can spit out apples to shrink back down, or hold down buttons. Stuff like that.

With each of these additions to the rule set, Goop Snake can create some truly clever levels, with those all-important “a-ha” moments.

Alright, let’s look at Curse of the Amethyst Skull. This is a dungeon-crawling puzzle game, where each time you die - you have to start the whole thing from scratch. But you can at least keep your items, your notes, your knowledge, and most fun of all, your hand-drawn map of the environment. Maybe I’m just a massive frickin’ nerd, but I really enjoyed charting out the contours and contents of this underground world - making notes of traps and treasure and dead ends.

I think I hit a bug where my map didn’t show properly, because I think I should have been able to see my current position on the map screen. But I guess I just played on hard mode, because I had to draw things out without knowing where the hell I am.

Another nice touch - you can light lanterns to know where you’ve been on previous runs. And when you come back as a new adventurer, you have a new, randomly generated name. No idea how Frances Grumblejack got access to Roswell Chapman’s notebook, but we’ve seen stranger things in this jam.

Here’s another common take on the theme: looping around from one side of the screen to the other, like in Pac-Man. Mobia’s Trip uses it as the main mechanic in a cute platformer - hold a key to lock the camera in place, and then you can waltz off to the left of this frame to appear on the right - which lets you bypass a big ol’ wall and be on your way.

This kicks off a number of clever scenarios where you’re manipulating the position of the frame, perhaps shifting the frame up with each jump so it’s high enough to reach a platform. Another mechanic comes from looping through the loop multiple times to build up speed, before using that excess energy to crash through the floor. It’s very nicely done. Shout out to Pimby’s Workaround, which had a similar idea and is also lots of fun - there wasn’t much between these two for me!

How’s this for a unique spin on the card battler? Like a Record puts your moves on a spinning disc - playing each card as the record shifts position. Only - your opponent is doing the same thing. So you’ll need to be smart about how the two discs will line up. Things get even more strategic when you add in cards that can change the direction of your disc - or your opponent’s. I’ll be honest, this was too much for my pigeon-sized mind, but even this simpleton can see the potential of putting Hearthstone on a turntable.

Here’s a clever one - Side Effects. It’s a puzzle game on a Rubik’s Cube. You’re presented with a starting point. Perhaps some seeds. And a goal. Say... a forest. By rotating the cube, you move the seeds into different zones - the water zone turns the seeds into saplings, and the sun zone turns the saplings into trees. Done, smashing.

But just like solving a real Rubik’s Cube, things get complicated fast. Sometimes you’ll need to manipulate things so two elements perfectly line up. Other times you’ll realise that one turn will solve one part of the puzzle, but utterly ruin another part. It becomes a sticky strategy game of thinking several moves ahead to get everything to the goal simultaneously. Maddening, but good.

Next up: Killover. A super-stylish game where you hunt down and kill a target. Then you do it again - only you also have to kill yourself from the previous loop. Then do it again and again and again, and now you’ve got to murder an entire army of yous from past loops.

Every time you advance, the game naturally gets harder and more chaotic. You’ve got to defeat more targets in the same short timeframe. And you have to dodge the bullets of all your previous selves. It becomes a manic warzone of your own creation as you dance through a hailstorm of bullets for which, to be honest, you only have yourself to blame.

Quick shout-out to the clever animation style. Snapping between distinct poses makes things very economical for a hurried game jam participant. But it also means that every time the game freeze-frames on the last kill of the round… it looks utterly badass. Everyone’s going to be jacking this style soon, I can feel it.

Okay, this is Draftula - which deserves a nod for its name alone. On each loop, you draft a new mansion for Dracula by plonking down rooms on this Tetris-style board. Each room has benefits and drawbacks - and must be placed strategically to get the most out of them. For instance, one room adds more coins every time you draft another room on the same row. Another room blocks you from drafting anything directly above it. Once you run out of money, a valiant knight knocks on the door, knocks down your gaff, and you have to start the whole thing again.

But there’s also a meta-level mystery to solve, as you must draft specific rooms next to each other to unlock more stuff - like putting two sigils next to each other to unlock infinitely high buildings. There’s a, ahem, lot of room to build on top of this idea, if the developer fancies it.

Okay, next up is Songrunner. This is a series of single-screen platforming stages where you build the level yourself. How? By playing notes - triggered through the keys 1 through 8 - when the timeline sweeps through the stage. So it’s like Lumines, but as a platformer. And that’s pretty neat. Best of all, your stage-making shenanigans create a unique track as you play. And when you finish the stage, all the other instruments kick in to turn your level layout into a song - which sounds surprisingly good for a track made up of awkwardly placed platforms. This one is simple but smart - just the way I like ’em.

While 1 couldn’t be much more basic. Use your mouse to draw a loop around the entire screen. And yeah, it’s so straightforward that you can look at the screen and intuit what to do in seconds. But that’s an impressive feat. Also impressive: how the developer used music, visual effects, and a singular sense of style to make something so basic into something thrilling and dangerous. An easy recommendation.

Another interesting take on the theme: how repeating the same level over and over again allows you to gain mastery and new understanding. That’s clearly shown in Loophole Pie. The premise is simple: a little boy wants to sneak into his neighbour’s garden and pinch a cooling pie. Easy. Only once you do that, the loop repeats with a new rule that blocks you from doing what you did in the last go. So: no more gate. Maybe sneak across the grass. Okay, next time - no gate, no grass. Then: no gate, no grass, no crates, no remote-controlled lawnmower, no rocks.

It turns the game into a sort of mix of Untitled Goose Game and Hitman, as you must use your mastery of this simple level to find more and more outlandish ways to reach the pie. It’s a great game design lesson: the more restrictions you put on the player, the more devious they must become.

And then finally - there were a lot of games in the jam about drawing loops around enemies to kill them. That became a common refrain. But no one did it better than the developers behind Bootloop. This is a rapid-fire knockabout arcade gem about ensnaring computer viruses in a deadly electrical trap. Here’s how it works: walk into a peg to start a loop. Then walk into another to make it bigger. Then hit Z to close the circuit and make a zapping zone in the middle.

Simple enough, but things escalate. For one, every peg you use must cool down, so using too many at once will leave you vulnerable after you attack. New enemy types appear, including a nasty scissor enemy that can cut your wire before you have a chance to complete the circuit. And then, between rounds, we get the classic three-upgrade screen as we must choose whether to boost health, the circuit, or a pack of last-ditch bombs.

It’s got fantastic pixel art and music, but this thing is not all style, no substance. It’s also just a really sharp time sink that I could see myself going back to again and again. Bravo, Team Booti.

And so there we have it! By thinking about a game being on a loop, we got games about building knowledge and mastery, about considering your past actions, about feeling the beat, and about manipulating space and time.

And now it’s, ahem, time for some honourable mentions!

Ninja Step has you programming the looping choreography of your player character to get through baddies and obstacles. Bend-Jo has you bending the entire playspace to solve mind-melting puzzles. Love Letters has you circling letters to get your jumbled words out to a lady who has you tongue-tied. And Loop/Death has you looping back to the starting point to solve tricky conundrums.

I could go on - did I mention there were over 9,000 games submitted? As always, I want to give a humongous thank you to everyone who took the time to make a game. There’s something amazing about a bunch of people coming together to make something just for the thrill of it. To be part of a community, to impress their friends, to better their skills, or for the sheer joy of making games.And all by hand - no AI slop in sight. It’s absolutely inspirational.

Thanks also to my fabulous mod team who make the jam run smoothly, the staff at itch who are always on hand to help, and my patrons who pay for the jam and mean we never need to bother with corporate sponsors. GMTK Game Jam will be back in 2026 for its 10th birthday! 10! How time flies! You can register your interest now, by clicking here. Thanks, see you soon.

The Nintendo Game Boy came out in 1989, the year I was born, and it was the first games console I ever owned.

It started a lifelong obsession with video games, and with Nintendo.

And so I'm thrilled to announce that my game, Mind Over Magnet, is available on the two latest Nintendo consoles: the Switch and the Switch 2.

Oh yeah, it's also available on the PS4, the Xbox One, the PS5, and the Xbox Series, and of course it's still available on Steam.

But there's something kind of poignant about this full loop moment of the game being available on what is, essentially, the great great grandchild of the console that sparked my love my gaming.

If you told my childhood self about this, I don't think he'd believe me.

So I want to give a thank you to Alchemy Games in Brazil, who ported the game to all six consoles. They handled the port, the certification process, the relationship with Nintendo and Sony and Microsoft. I barely had to lift a finger.

So if you want to know how to port a game to consoles, I'm afraid this is not that post!

Instead, I just wanted to mark this moment, be grateful for it, and give thanks to everyone who helped me get here.

Cheers. That's all. See you soon.

New releases

Donkey Kong Bananza - Played

Weirdly, Nintendo waited until mere days before launch to reveal that, yes, this is the new game from the core Mario team. The expert designers behind Super Mario Odyssey.

But you can tell as soon as you start playing. Snappy animations, charming graphics, a focus on exploration, and more collectibles than you can count. This time it's all about finding gigantic bananas, as well as fossils, coins, and gold. So much gold. Gold in the thousands. Gold with decimal places.

Why's it not a Mario game? Because it leans on the brawn of Donkey Kong. The game is all about smashing, crashing, and bashing stuff. The entire world is a destructible playground of rocks and dirt, and you can carve through it all by hammering on the punch button. And challenges revolve around breaking stuff, over jumping. Destroy this house under the time limit. Throw this rock at that bomb. That sort of thing.

So far (I'm about 3 worlds in), it's really fun, charming, and imaginative. But it's also a bit too manic at times. You tear through these levels so quickly that it's hard to truly know them. I have no idea where's up, where's down, where's the start and where's the finish. It's all just rocks and concrete slabs that explodes with a flurry of button presses. It's a long way from the intricate level layouts of a typical Mario game.

But, hey, let them cook. I'll be happy to play more as I slowly pick up bananas and work my way to 100% completion.

Read more

Hi, I'm Mark Brown from Game Maker's Toolkit, and I am chuffed to bits to let you know that my new game, Word Play, is out now on Steam.

It's a game where you spell words to score points, so you can unlock powerful perks that let you score even more points. It's a game about spelling, and strategy, and synergies, and special powers. It's got a classy art style, a boppin' soundtrack, it's got accessibility settings, and difficulty modes, and loads more. I hope you'll check it out.

Now for my first game, Mind Over Magnet, I shared the entire process of making it across 15 or so episodes of my YouTube series, Developing.

But for Word Play?

Well, there just hasn't been time. The game has come together so quickly that I've barely had a chance to tell you how I made it! But better late than never, right? So here it is. Here's how I made Word Play in just seven months.

December 2024

Okay, so the idea for Word Play - which was essentially to combine the poker roguelike Balatro with a spelling game like Scrabble or Bookworm Adventures - the idea has been rattling around my head ever since I played Balatro at the start of 2024.

But in December, I finally had a reason to try and make it for our secret, Patreon-only GMTK Game Jam.

So I bashed together a game about spelling words, scoring points, and unlocking rule-bending modifiers.

It was called Wordy, and it was actually pretty dang fun. The response was great, I had a really good time working on it, it clearly had a lot of potential, and if I had any intention of turning this into a real game, well I had better get on with it before someone else has the same idea.

And so, shortly after the jam was completed, I opened up a fresh Unity project and started working on a more fleshed out version. Now I decided that the best thing to do at this early stage was to not worry about things like perks, or rounds, or rogue-like elements - but to just build a really good spelling game. Something where it was nice and juicy to spell words.

Something that would work seamlessly across mouse and controller, and keyboards, and touchscreens.

January 2025

At the beginning of January, I continued to build that basic word game - adding in features like a shuffle button to mix up your tiles, and a letter bag. But then, with the spelling game mostly done, I started adding on the rogue-like features like the shop, and the perks.

I added the basic architecture for special tiles, and started making modifiers which are your passive perks - think jokers in Balatro, or boons in Hades.

By this point the game had 11 perks, though they were mostly just there to test different functions like having a modifier that can boost a tile's score, or the whole word's score.

February 2025

In February, work started to slow down - mostly because I was just focusing a lot of time on making videos for GMTK. I added a few features, like the ability to sell modifiers, but that was it.

Then midway into February, disaster struck.

I was scrolling through my YouTube subscriptions and saw a video on Second Wind with Yahtzee playing a game called Wordatro. Which I didn't even need to click on it to know what that one was all about. It was another game that mixed Balatro with Scrabble.

Damn.

Now, I briefly considered cancelling the game, and I definitely regretted not showing this thing earlier, but ultimately I decided that I can't control what anyone else makes. The only thing I can control is how good my game is. So I decided to keep working on it, and just make the game as awesome as I could make it.

Oh, and then later that same month, I listened to a completely random podcast about screenwriting, and one of the hosts said...

“We have a game for you to try out right now. So it's called Birdigo. It is a roguelike deck builder where you're trying to make words.”

Which you might think would give me a heart attack. Another one!

But actually, this made me feel better. Because between Birdigo and Wordatro, and over the next few months, many more word-based roguelikes all having the same basic idea of Balatro plus words but going in different directions.... well it feels less like I'm competing with and, in some people's eyes, copying a single game, and more like I'm part of a growing genre.

So crisis averted, and I even ended up chatting with the devs of both Wordatro and Birdigo and, well, I'll have more to share on that down the line. Suffice to say, we're all good.

March 2025

In March, I finally gave the game a proper name. Here's how that happened.

You see, the game has, essentially, lives. But you use a life every time you spell a word. So it's a bit weird to lose a life when you do something correctly. That just feels weird. So I asked my Discord users for alternative words to "lives", and user Fly suggested "plays". Which sparked the idea for Word Play.

Now, when you're making a word game, as it turns out, every single possible name has already been taken by some other game or toy. But Word Play? Well, Word Play doesn't exist on Steam. So you know what? I'll take it. I'll snap up Word Play.

Anyway, I started March by continuing to add more roguelike elements. So I added upgrades. They're your active perks which can change, improve, duplicate, or discard tiles. I added in all the special tiles - golden, diamond, emerald, potion, and dot tiles - which all have unique gameplay. I added in the special rounds, which are, well, special rounds with unique requirements like "your word must have six tiles". I added modifiers that can multiply the score. And I added reroll to the shop.

So now that I had made the spelling game, and coded in all the roguelike stuff, I could tie all of those elements together into a full gameplay loop - so that it could be played as an actual game. Which was especially important with the Game Developers Conference coming up at the end of the month.

So now I could start getting proper playtesting feedback on this new version of the game. I distributed it on the GMTK Discord, and took it on an iPad when I did talks at schools and colleges. That led to lots of changes based on user feedback and requests. I made it so you could rearrange the tiles on the board. And I made it so you get extra bonus points if you spell especially long words - because otherwise you barely get any points for dropping a single mega word.

And then I took it to GDC, where I spent the entire week playing it with other developers and got fantastic feedback from the devs behind games like Patrick's Parabox, Spell Tower, and Alto's Adventure. Which led to even more changes.

Also I rode in a driverless taxi, and had my breakfast delivered by a robot. San Francisco!

Also in March, I started reaching out to my collaborators on the project. I emailed Zach, who did the music for Mind Over Magnet, and offered him first dibs for the music on this game. And I DMed Jay on Discord. He made a spec video of the game with different sound effects, and they were so good I hired him to do the whole game's sound design.

I also added in Unity Analytics to help with future balancing. I created the Steam store page, though kept it private. And I brought the total number of perks up to 52.

Basically, I went into hyperdrive. Looking back on my notes, I can't believe how much I accomplished in one month. I was racing through this game. Development was coming along so well.

And then I broke my arm.

April 2025

Yeah. April was a bit slow.

I had to stop development for several weeks as I waited for my elbow to heal. Jay finished the sound effects, but it would be a while before I could put them into the game.

But I was able to slowly get back to work. I added in a new special round where you have to play a mandatory tile, or else lose two plays. I added compound tiles, letting you play the letters I N G with a single tile. And I added in the first modifier that was based around the specifics of the word you spelled.

It's this one, which gives you a boost if the first and last letters are the same. These would prove to be very important in the game's overall sense of strategy and balance. But I think I'll talk more in depth about that in a future video.

Oh, and this also brought the total number of perks up to 74.

May 2025

May started out with disappointment. I managed to get in touch with the chap in charge of Apple Arcade, which is the iPhone maker's Game Pass-like subscription service. And so at his request, I sent over a build of Word Play on TestFlight.

Ten agonising days later, Apple Arcade wrote back to pass on the game. Which was a huge shame, but hey, it was cool for them to play it at the very least.

But okay, with that out of the way, my goal throughout the rest of May was to get a demo ready for Steam Next Fest in June. Which meant fixing lots of bugs, adding lots of polish, and chucking in a lot more cool features - like the glass tile, which can be used to have a one-time copy of any other letter.

And so towards the end of May, I was finally ready to reveal the game to GMTK viewers like you - with a demo, an announcement video, and a let's play of the game.

Which was amazing. The response was overwhelming. 20,000 players in a week! The game got picked up by various YouTubers and streamers, including Yahtzee on Second Wind, him again, and the world's biggest Balatro streamer, Northernlion.

I also got an email from… uh, Apple. Yeah, not their Apple Arcade team this time, but a different group at the company, who reached out to offer some help, listing it on the main App Store.

More on that in the future, because right now I'm focused entirely on today's Steam release.

Plus, all of those demo players sent me so much great feedback and ideas, which led to all sorts of cool stuff - like being able to submit words over 10 letters, and being able to petition for more words to be added to the dictionary.

And so at the end of May, the game was in a very good state. And the game now has 90 perks.

June 2025

By June, the basic underlying game was essentially done. It was absolutely a minimum viable product, or MVP. It had a fully functioning game loop, it had saving and loading, options, and difficulty modes.

Still, I added a few more quality of life adjustments, like the ability to peek into your bag or at the board during the perk screen. Some accessibility features, like alternative controls, dyslexia-friendly fonts, and spelling suggestions. And I even added the groundwork for people to mod in custom dictionaries and letter bags if they want to play with non-English words.

But, with all of that done, and with Steam Next Fest taken care of, my job now was to simply make stuff.

So I added new tiles. The mirror tile. The exclamation mark tile. The locked tile. I added new special rounds, like one where the maximum word length starts at four, and increases with each word you submit. And I added perks. So many perks.

I started by bringing the game up to 100, but then I thought of some more and took it to 120. Then I had a few more ideas, and it was like, why stop at 143? I might as well take it to 150, that's a nice round number. And over the rest of the month, I slowly added a few more until I really, really had to stop at a whopping 160 perks.

And look, I've got even more ideas, but let's save those for future updates.

But at this point it was time to start doing the final quality assurance, or QA.

I brought in a group of helpers - a mix of GMTK patrons and people who gave particularly insightful feedback about the demo, and had them run through the game over and over and over again, looking for problems. They found bugs. They found typos. They found perks that were overpowered, and perks that were pointlessly weak. They helped me balance the difficulty curve, suggested Steam achievements, and pushed me to add in features that I had been way too lazy to make previously.

To be honest, this lot were the real MVPs - the most valuable players. Without their help and guidance, the game that just launched would be buggy, unbalanced, and just significantly less fun. So a huge, heartfelt thanks to all of them. Cheers.

July 2025

And then finally, it was July. It is July! And the game is out now.

And so that is the story of how I made Word Play in seven months. And if you compare that to the development of my first game, Mind Over Magnet, which took me well over three years to produce, well obviously things went a lot more quickly.

Now, of course, they're very different games. And I was also a very different developer back then. I literally had to learn how to use Unity while I made the game.

But it's not just that.

I also thought carefully about how to develop Word Play while I was making it. To make sure the process was streamlined, and to ensure that this game didn't devour years of my life in the way that Mind Over Magnet did.

Now I'll talk more about this in future episodes of Game Dev 101, which I'm looking forward to getting back to soon. But as the Cliffs Notes version, it really came down to two things.

Number one. I was ruthless about the game's scope. I would routinely make decisions through the lens of which choice is going to take the least amount of time.

For instance, I could have made unique art for all 160 perks, like the jokers in Balatro. And I would totally accept the criticism that the game is worse off for not having that. But ultimately, using a more simple set of icons saved weeks of time. And it was the difference between the game coming out while I still had some fuel in the tank, versus causing the development to drag on for many more months.

And number two. I would develop the game in a kind of layered approach, where I would focus on one part at a time.

For instance, in December, I focused exclusively on just making a good spelling game. Then when that was done, I focused on getting the roguelike game loop complete. Then I just made perks all day, every day. And then I just did bug fixing and balancing.

Now of course, none of that is really true - I was still adding in features from previous layers as development went along. But for the most part, I did just focus on making the bones of the game before turning my attention to the content. Which made development faster and more agile, as I was always building on the really strong foundation of the previous layer.

And so ultimately, while I'm really proud of this game - and I think it is genuinely a lot of fun and I hope you dig it - I think I'm even more proud of the fact that I made a video game without losing my mind.

By taking everything I learned from Mind Over Magnet, and really planning out the development instead of just making stuff willy nilly, I was able to take this game from idea, to prototype, to playtesting, to QA, and to release in just seven months.

So I'm going to leave it there, so you can actually go and play the dang game. It's out now on Steam for PC, Mac, and Steam Deck. There's a 10% discount for the first week, and there's still a free demo if you haven't tried it yet. And now, well, give me a sec to fix all of the inevitable bugs, and then I'll see you later this month for GMTK Game Jam 2025.

I will not be taking part. Do not let me. I cannot be trusted.

Bye bye!

Hello GMTK supporters!

Okay, I'm gonna level with you. My game is releasing in two weeks. I've been spending every spare moment fixing bugs, tweaking the balance, making trailers, and all the other stuff you have to do before a game comes out.

So this ain't gonna be the best GMTK Digest I've ever done. But I hope you dig it nonetheless! I'll hopefully be back on form for the next one!

New releases

Played: Mario Kart World

Oh yeah, we got a new video game console in June! Hello Switch 2! The big release is, of course, the latest Mario Kart game. This one is just as fun, silly and polished as the rest of the games. And it's packed with nostalgic nods (including a truly enormous soundtrack that references almost every Mario game you can name).

But it's also got an open world that is, for my money, kinda bad. Free roaming around this lifeless universe feels pointless. And the GP mode now features these long stretches of liminal, interstitial nothingness between the "actual" tracks. From my experience this adds nothing to the game... except for the fun knockout mode where you have to survive while racing across the entire island. That, I'll admit, is pretty fun.

Read more