Electron Dance

Dead in the Water

This is the sixteeneth part of The Ouroboros Sequence, a series on puzzle games.

The Steam store page for genius puzzle game Recursed (Portponky, 2016) highlights two reviews.

One is my very own, published on Rock Paper Shotgun. I wrote “Recursed’s brilliance is how it spawns complexity from a few simple constraints.”

The other is from the one and only Jonathan Blow who worked on a puzzle game you might have heard about. “I played for a while,” wrote Blow, “but it seemed really slow / simple.”

I’m only going to warn you once. Strap yourselves in.

Imagine you are at school and your teacher, Mrs. Blow, is explaining to the class how to factorise a quadratic expression. Mrs. Blow goes on for, say, fifteen minutes. Then she hands out the practice sheets and you are a little surprised because there is only one question on it. Everybody struggles with the problem because it’s the first factorisation they’ve ever done. Nonetheless, ten minutes later, everyone has it figured out! You are very proud of yourself.

Mrs. Blow congratulates the class. Then she announces that as everyone has cracked factorisation, the class will now apply it to solving quadratic equations.

Breaking news: there is no Mrs. Blow. No teacher would expect everyone in their class to have digested and mastered a new concept with just one example. The kids would struggle with the next few lessons as they hadn't grasped the basics. Under this teaching regime, how many of them would wind up dropping out?

Yet this is the ideal that puzzle designers aspire to. The tightest sequence of challenges which keeps things fresh and allows the player to grow. No two puzzles are structurally the same. “Think you’re smart, eh? Well, take this,” the designer says as they hurl a new curveball at the player.

Now this bit of game design is proper hard. The sequence is like a trail of breadcrumbs which leads players deeper into the forest of challenge. Insights are not so much fed to but bled from the player. Occasionally they are forced to question their assumptions and have a winning “Oh My Christ! I didn’t realise you could do that!” moment which is of course the best videogame moment (hey, have you played The Witness).

Creating this sequence is a hard problem because the developer is the all-knowing god and it is not possible to see with mortal eyes. To test whether the sequence gets it right is laborious. Once a player has been through the game, they too are cursed with knowing, thus not as useful to review a new version of the sequence. To get the best feedback, you need fresh victims, fresh meat for the puzzle grinder.

I am sure you have heard of the “state of flow” that occurs when you are completely absorbed by a task. Mihaly Csikszentmihalyi, who developed the theory of flow, proposed that flow only happens when there’s a perfect match between an individual’s skills and the challenge before them. This zone is known as the flow channel. If skill and challenge are mismatched, players end up either bored or frustrated. Here’s a really simple graph that demonstrates the relationship.

Designers are pursuing the Holy Grail - a game which ushers players down the flow channel. They have full control of the challenge variable - they designed the levels, after all. But designers have tenuous control over skill and here be the problem. Not all players are created equally. Some have played block-pushers for years while others are just losing their puzzle virginity and think a monster pushing snowballs around sounds kinda fun.

All of these people are put through the same sequence, a one-size fits all approach to puzzle education. Some games offer players a choose-your-own-adventure approach instead, such as Cosmic Express (Hazelden & Davis, 2017) and Stephen’s Sausage Roll (Increpare Games, 2016) where players are not forced to tread a linear sequence but instead dine on an exciting buffet of levels. But, in our classroom metaphor, the adventure map is not offering multiple examples of one subject, but multiple subjects: are you going to attempt the question on solving quadratic equations, the one on vector products or try your hand at Lebesgue integration?

Cosmic Express level map

I am being unkind. There is often overlap and sometimes follow-up levels are attempting to shine a light on the current limits of a player’s thinking. Designers will always try to work with as many players of varying ability, but inevitability there are limits. Here's Alan Hazelden, designer of Cosmic Express amongst others, interviewed in an Electron Dance essay earlier in the year, The Developers Who Won’t Hold Your Hand:

I remember one playtest where the player took five minutes to realise in level 2 that the train can only carry one passenger at a time. That felt bad, but I think I stuck to my guns and went ‘nope, this is not unfair, you just have to pay attention to what's happening’. Since it happened to one person I'm sure it's happened to others - I just have to hope that if someone stops playing there they probably wouldn't have enjoyed the rest of the game anyway.

I’m not arguing this is bad or evil. It’s not a dereliction of game design duty. It is fact. You cannot design for everyone. However, designers are biased towards pushing players into the frustration space where skill < challenge rather than the boredom space where skill > challenge. This is Blow’s comment on Recursed: aim for the flow channel or higher, not below you doofus. Stephen’s Sausage Roll could be viewed as this methodology taken to the extreme. Each puzzle is unique; there is no repetition and almost certainly no holding of hands. There is no notion of practice with Stephen’s Sausage Roll. Do. There is no try. You sausage. Or die.

This philosophy is so heavily ingrained in contemporary puzzle design that when presented with the alternate view, it is casually accepted as failure. Jonathan Blow’s non-review tweet of Recursed suggests he has interpreted the bias towards the boredom side of the flow channel as a signature of this-one-ain't-really-worth-it.

This dismissal of “slow” progression reminds me strongly of the arguments against making games “easier” and “more accessible”. And Recursed isn’t unique. A lot of fresh talent starts out in the floor of the flow channel, dipping into boredom space. Tricone Lab (Partickhill Games, 2017) is another one where the puzzles are a bloody breeze for seven aeons but it does, eventually, push back against the player. The question is whether all that is necessary. Tricone Lab dazzles you with lots of tiny monochromatic flying bits - would a tighter sequencing lose more players? The grey interface could probably do with an improvement, but the Tricone Lab player doesn’t have much prior experience to rely on. It’s not a block pusher or a Match-3 game.

Tricone Lab

Let’s take a gander at a 2018 paper titled “Model Matching Theory: A Framework for Examining the Alignment between Game Mechanics and Mental Models” by Rory Gloin, Joe Wasserman and Andy Boyan. The authors propose that learning in games can be better explained through the interactions of a player’s mental models. Players don’t need to be taught about gravity, so they can understand a game in which “stuff falls down”. But experience of how the WASD keys move an avatar around a first-person shooter is specialised knowledge. If there is a high match between existing mental models and a game, the faster a player will pick things up.

There is one paragraph in particular that stood out to me. The authors posit that games are particularly good teachers because of:

  • Repetition. Repetition presents players with the opportunity to review and revise mental models.
  • Progression. Games gradually increase difficulty allowing previous inferences to be retested within new contexts.
  • Spectating. It is possible to learn from the experience of other players.

In the world of puzzle games, designers are focused heavily on the progression aspect. As puzzle games are generally solitary experiences, spectating is not a big feature and, indeed, most puzzle players want to find a solution themselves than copy what Brian Eggbrain did in his Let’s Play. But what of repetition?

You can argue that players continue to “fight” a particular puzzle, repeatedly, until they defeat it. Failure is a learning experience. But remember the key takeaway from The Monte Carlo Player. Sometimes players solve a level and have no idea how. This is the worst-case scenario for a designer, who hopes that none of the levels are vulnerable to brute force solution. Random solution is camouflaged failure.

A game designed around “frustration-bias” is a high momentum approach, pushing players to look forward not back. Puzzle games never ask a player explicitly whether they really understood the puzzle but canny designers may pose a follow-up that tests player understanding. If that understanding isn’t there, the level gates progress and the player, it is hoped, will reflect on what they know. Or, more precisely, what they do not know.

Consider the first puzzle panel track you encounter in The Witness which is there to explain the meaning of the dots on a puzzle. When you see black dots and white dots, you have to draw lines to keep them separate.

The first puzzle practically solves itself - drag right for the win. But there are other possible interpretations of this single puzzle. The next few puzzles are designed to nail it.

Further puzzles explore whether all the dots have to be collected together. However, the puzzles shift focus towards heuristics, such as learning to trace around the edge to avoid partitioning.

And because this tutorial has, in the main, encouraged the player to see the puzzles in terms of white dots, one of them challenges that assumption. It's easier to think of the following panel in terms of black dots.

Great design, but I need to confess something.

The Witness attempts to teach its rudimentary rules like this again and again. And one of them did not work for me: the star puzzles in the treehouse zone. I’m not going to spoil them if you haven’t played before but after breezing through several of the star puzzles using random experimentation, I realised I hadn’t a fucking clue what was going on. After some frustration, I buggered off to another zone to get some respite.

It’s time for the point, reader. Even the most amazing puzzle design will lose somebody. Even people who are really into your design. What saved the star puzzles was being able to move back through the completed puzzles and experiment with them.

And this is why, at the end of The Laboratory of Logic, I said “this will prove to be important later”. As I explained, the UI of many puzzle games are not very good at encouraging you to review past challenges. Something like Cosmic Express is not bad, although it’s difficult to see exactly what puzzles you’ve travelled through. The Witness is brilliant because each puzzle is located somewhere on an island and that lends your puzzle memory some geography.

I’ve pointed out that laser reflection games are, ironically, not very good at reflection as they have a tendency to “solve accidentally” and distract you with WELL DONE, LET’S GO TO THE NEXT LEVEL, PLAYER ONE. And while Sokobanlike Stephen’s Sausage Roll does offer replay, the functionality is buried under sixteen tonnes of UI rock and practically useless.

Machine at the Heart of the World (Evidently Cube, 2018) is a game about decrypting instructions. Each puzzle has a unique solution and you literally have to guess the solutions of the simple puzzles. These correct guesses arm you with knowledge. Only then can you solve the complex puzzles that are largely immune to brute force. Machine is built around experimentation but review is key: tapping the left arrow key will shift you back to previous solved puzzles instantaneously. It becomes a reflex to do this. Tap, tap, tap, observe, think, try again. I don’t know of another puzzle game that acknowledges so explicitly your solution history... but I’m sure a reader does. Answers in the comments, people.

Machine at the Heart of the World

Valve were a pioneer of using clever tricks to lead players into predestined drama that also convinced the player they had agency. One of my favourite tricks is the positioning of a fence to nudge the player into looking directly at the first appearance of a Hunter in Half-Life 2: Episode 2 (Valve, 2007).

This intense stage management is now a hallmark of the AAA FPS. It is also where puzzle design is today. Many logic puzzles aspire to giving players a sense of drama, driving them up the flow channel, feeling like they’ve just stolen victory from the jaws of defeat every single time. Wow, YOU solved it with nothing other than your wits! Yet the designer has been shaping your experience, carefully plotting a linear puzzle sequence that nudges you ever onwards... or presenting an adventure map with judicious gating.

Gating can be done subtly. We've already mentioned levels that require players to break assumptions to pass. We could also deploy noise, using unnecessary puzzle clutter that a player can only solve if they have a clear mental model that allows them to cut through the crap. Such clutter is out of vogue - I am not a fan of obfuscation myself.

These are incredibly sophisticated efforts but they do not always succeed. As an example of where tight design can go awry, I offer touchscreen puzzler Invert (Glitchnap, 2017). Invert is a type of Lights Out puzzle; you want all the tiles to be the same colour, and pressing switches will flip an associated group of tiles. Generally, I find Lights Out puzzles dull - although recent Illiteracy (Lucas Le Slo, 2018) was a pleasant diversion - so a game built around them has to mix things up if it’s going to go the distance. Invert is aware of this danger and, after eight levels, offers an entirely different tile formation with its own nuances.


That’s the whole game. Eight levels per tileset. Now those eight levels are not even eight levels because the first few are inevitably thin tutorials to get you to play with the switches. After those tutorials, I was using trial and error to muddle my way through. It was only after the eight levels I would finally start to become familiar with a tileset but it was at this point Invert said bored now: let’s try something else. I was whisked off to another tileset where I had to start learning all over again. Thus I spent most of my game guessing and was robbed of consolidating what I had learnt at the last minute. Defeat stolen from the jaws of victory.

When the player fails to keep pace with the game, enlightened puzzlers allow players to skip levels or offer hints. But look, once again, to the first-person shooter - where are the “low-difficulty” modes for puzzle games? Imagine inserting more levels for the troubled player to give them a better chance to learn. A recent paper by Chris Martens et al, “Generating Puzzle Progressions to Study Mental Model Matching” attempts to procedurally-generate the sort of progression that contemporary designers aim for - a procgen puzzle engine is a far more promising angle for producing a sequence that adapts to the player’s skill level. What learning curve do you want? And the engine spits out five puzzles where one might have done. Remember, as I explained in the previous Ouroboros essay, I didn’t develop a mental model of Dissembler (Ian MacLarty, 2018) over the course of its handcrafted campaign but through its procedurally-generated content:

I had become an honest-to-God Dissembler expert with a head full of heuristics that I could not express in words. I had a deep understanding of how colours could be migrated across the board and the limits of migration. I was able to conjure alternative strategies when the more obvious ones would fail. Dissembler did not teach this through individual examples but through daily practice.

The alternative to designing for different skill levels is to pull the handbrake and encourage review. The Witness was all over this but its design is somewhat unique. It's time to admit that rather than describing puzzle games as a year in Mrs. Blow’s classroom, a better analogy would be Hazelden University, where you get a couple of lectures a week and it’s totally down to you to transport the knowledge into your head through practice and review. At least at Hazelden Uni, we know what is expected of us. But the average puzzle game shrugs its shoulders as if players should have known all this already. What are you, it snarks, a four-year-old?

Now and then, I catch a glint of arrogance in the puzzle genre. It does not come from the developers whose intentions are entirely honorable. But it is reflected in the surface of their games, with a focus on efficient sequencing and the withdrawal of explicit instructions in favour of a design where the player has to figure out the basics. There’s a whiff of a feedback loop here in which design chases the hardcore, as the first-person shooter did, a snake eating its own tail.

I love efficient design. I love discoverable systems. But are player practices evolving with modern puzzle design... or is design leaving some players behind? I can’t help but wonder how many players abandon puzzles because they have not been catered for.

And I wonder how many of those dropouts feel like it’s their fault.

Download my FREE eBook on the collapse of indie game prices an accessible and comprehensive explanation of what has happened to the market.

Sign up for the monthly Electron Dance Newsletter and follow on Twitter!

Electron Dance Highlights

Comments (30) Trackbacks (0)
  1. I’ve really been enjoying the Ouroboros Sequence posts and I feel they’ve been building together to make some incredible points. I was amused to read the Mrs. Blow metaphor, because in fact there are way too many Mrs. Blows out there (I’m a public school teacher in the USA). Have you heard of Vygotsky’s Zone of Proximal Development? It’s essentially the same thing as Csikszentmihalyi’s description of flow, except about a century older.

    Your articles do a great job of describing theory of the mind to people who might not normally be into those topics. I appreciate your work.

  2. Great article! I don’t have much time play anything these days, but I will try to read and contribute some way time to time, maybe after some cold beer, haha.

    Let’s suppose a procedurally generated game, inverted, in which the game is generated by the gamer, inputting data and thus generating the game. The game would be for example, an artist simulator, like a painter, the gamer would throughout the play learn the necessary skills and in each stage be challenged making a piece of art. The parameters would vary, make something beautiful, something intense, something provocative, something experimental and so on. What would be the “engine” that would judge this something? If the data exists before the player’s input, then it would be subject to a frozen and therefore broken parameter, if these parameters are variables, wouldn’t be “natural”, as they would use only a quantified perspective, but absent of quality. Following this argument, the next question would be, how to define the “victory” of these challenges, without making them false? And would this be better fit as a game or as a tool, and if one, why not being the two? So, as I can see, the biggest problem of games, which puzzles, adventures, interactive text, simulation, strategies, and so on, stumble, is the fact that you can’t as a designer imitate the flowing “design” of nature, that is, you cant truly make something new from something old, the limits are from the previous construction, the designer. You can think in all the problems of the world, you will still be locked for the limitations of these problems, there is a finite number of problems you can put, and a limited numbers of ways you can present them. Of course, we can use philosophy in this, but right now I have no strength to put any philosophy in it, as I am a perfectionist and would like to make some research before saying something. So I am just spelling some words and seeing what happens.

  3. It will be interesting to see what you make of my next game! Which is currently still a long way off, unfortunately.

    (Yes, this comment is really just a tease, sorry)

  4. «What saved the star puzzles was being able to move back through the completed puzzles and experiment with them.»

    Well, that’s to me the main “message” of the game: witness the world, but even more importantly, witness your own learning process. The re-visitability of the puzzles is not an accidental feature: it invites (sometimes unsuccessfully, sadly) to experiment and challenge assumptions or de-randomize solutions, like a toy version of a scientific methodology.

  5. So about the level select in Snakebird….

  6. Hi everybody, it’s been *quite* a day. My wife’s father passed away suddenly this morning and we’ve been debating whether the entire family should fly out to Japan. Finally decided my wife would fly out alone and I’ll be single dad for the next week, probably doing half days from home. Not sure how present I’ll be on the site over the next week, we will see.


    Thanks, John! I hadn’t heard of Vygotsky’s ZPD but I’ve looked it up now! I can see what you’re driving at. I’m not sure the ZPD concept aligns as neatly for the ideas in this article – but maybe that’s just because I haven’t had to try :)

    The original conception for this article was pretty very lean: maybe the designing for the hardcore is a problem. It was only recently I started writing it and all this other interesting information just dropped onto my lap. I’m happy it’s worked out so well!


    Hello again. Sorry you don’t have much time for games. Sounds like all you have to do is buy Red Dead Redemption 2 and you won’t need another game for the next four years.

    There are two aspects of your comments I want to tackle. The first is defining the victory condition. Effectively, we’re describing the game of the “game designer”. The interesting thing about this game is that what makes a good game is always moving, dependent on context. When 3D came in big in the 90s, every game had to have 3D otherwise it would be considered inferior. That’s the thing in this essay: the idea of the perfect puzzler has been moving towards what a hardcore puzzle nut wants, but do the less switched-on puzzle players consider this a worthy goal? Answers on a postcard.

    Your point about the “finiteness” of the randomness is something that troubles a lot of procgen content. That the human brain is pretty good at deducing patterns thus your procgen must be really wild to keep surprising us humans. A lot of No Man’s Sky looks familiar after a few planets (oh look it’s another one of those fan-shaped rocks!) Perhaps you just need enough variables, as many as the lines of code in Space Invaders, maybe. Enough variables to represent a program….


    Oh no, it’s set in Draknek University, isn’t it?!?


    That’s interesting. I’d been so focused on what the design of The Witness was doing for the player that I had ignored how it played into the larger themes.


    Oh Matt. It is a bit shit, isn’t it? Functionally useful but not much beyond that.

  7. Condolences! Hope you’re doing okay.

    The goal for this game is to be Draknek pre-school through post-grad, and anyone can drop out when they feel like they’ve had enough formal education thank you very much. We’ll see how effective we are at that goal…

  8. Condolences, Joel, and especially to your wife.

    The level select in Snakebird is meant to be quite on topic but I’ll leave that for later.

  9. @Joel
    “Sounds like all you have to do is buy Red Dead Redemption 2 and you won’t need another game for the next four years.” – That was funny, who knows, maybe, who never joined in the forbidden pleasures of a triple A open game?

    “Sorry you don’t have much time for games. ” – I overreacted my lack of free time, I guess this year I used my free time sleeping, reading (when not reading for studying), going out, travelling or resolving personal problems, and of course debating about extreme-right in the world, like in my country, I need spare some moment to understand for example what’s going on with Brexit, it’s very confusing for someone outside.

    “That’s the thing in this essay: the idea of the perfect puzzler has been moving towards what a hardcore puzzle nut wants, but do the less switched-on puzzle players consider this a worthy goal?” – I’m starting to understand the contradictions you are trying to unveil. You can have both types, the hardcore and the casual. But in the same time, the difficult still is making a game for both audiences, even if you chose one audience, the hardcore, you have to make something playable, if you chose the casual but decided that casual is the same as stupidity, your puzzle it’s not a puzzle, just a passive experience.


    Sorry for your loss, for you, Joel and your family. Hope everything goes fine in this moment of grief, wish all the best!

  10. 1. If I have a natural bias, it’s against games being ‘dumbed down’. I think it’s that the phrasing itself is clearly loaded, but I’m also protective of my achievements in having beaten hard games, or at least games I found hard, as silly as that may sound. But this post encouraged me to think of times I experienced outright frustration, and my thoughts immediately went to my hopeless attempts to get into the ‘Gaidenetta’ 3rd-person action game genre.

    Constantly struggling to master the camera and basic combos, nevermind the elaborate and bewilderingly-named additional extra mechanics stretched over the framework as the genre proliferated, I bumped along the very bottom of games like Bayonetta and the Wonderful 101, constantly being awarded the worst scores in their regular micro-assessments of my floundering but never quite being bad enough to fail outright.

    The games knew I was bad at them – they were routinely telling me so – but acted as though I was good, always ushering me on to the next area, where I would fare no better but would now encounter harder enemies, character limitations or environmental hazards. Cranking up the challenge so that I was constantly well behind the curve. Contrast this to, say, Dark Souls, which certainly wasn’t shy about setting me challenges, but patiently waited, impassive, until I proved I was could meet them before moving on to something tougher or more advanced.

    2. Lemmings had four difficulties – not modes, but level packs, clearly demarcated categories ranging from beginner to insane. I don’t want to cause anyone conniptions by suggesting Lemmings was a puzzle game, but it was a game with puzzles *in* it. The difficulty gaps were huge between categories, but rather gradual between levels, and there was plenty of time for repetition and refinement, as perhaps expected for a game aimed at the mass market rather than the brooding and intellectual campus of Puzzle Gamer U.

    I beat all of Beginner and Fun, most of Taxing and precisely one level of Insane. What’s interesting is that completing one distinct subset of levels tricks my brain into thinking I didn’t give up on it, even though I never came close to seeing the credits. Whereas if the levels had been presented as one long death march, the point I turned my back on the game would be the point at which I failed to beat it. Instead it feels like I sampled as much as I cared to, and even came away with a sense that I *did* beat it.. in part.

    3. Hoshi Saga! There’s no point here, they just make me smile.

  11. Alan

    Everything is fine here but usually my other half is responsible for organizing shopping and meals and the children’s activities – but for just over one week, it’s all me. Hoping I don’t screw up because the children will never want me in charge again.

    You know, I did have this fanciful idea at one point (not when writing this) that you could have games that get progressively harder but it would be fine to bow out when it surpassed your skill potential. This is the way I approach all sorts of action games that I like – THOTH for example, I loved the campaign, but the procgen additions were definitely beyond me. But there’s also a problem of satisfaction; are players happy if they’re just supposed to abandon at a certain point? The answer to that question I think depends on the game.


    Thanks Matt. She is taking it well. She left for Japan just a few hours ago.

    I remember you did bring up the Snakebird level select on another article…?


    Sleeping, reading, going out are highly recommended activities and I wish I could do some of these myself :)

    Back to puzzles. I guess my issue is that you can’t just divide up between “casual” and “hardcore” because those titles hide a whole spectrum of players. And continuing to court what you see as hardcore will result in games that are inevitably chasing smaller and smaller audiences. If you think commercially, that’s going to be a problem. However, neither do I want to disuade anyone from making the next Stephen’s Sausage Roll. There have to be some gritty works out there which are clearly not for a general audience. We need that colour.

    I am also aware creating puzzles which are “review-friendly” and/or have adaptive difficulty are not necessarily easy to do. This is more a theoretical essay than about pragmatism or real-world solutions.


    Hello CA! I guess we’ll run through your points separately.

    1. The “dumbed down” thing is really complex to unpack. I mean, I get it too. On one hand, gah stop telling me where to go in the missions, let me find my own way, let me suffer terrible deaths to give the play meaning etc. etc. But then you have This Game Is Clearly Not For Me. This series, after all, is building up to my terrible experience of Snakebird.

    We all want games to lock into our own flow channel, so they feel special and dedicated to us. I was playing the free Fire of Kala today and I had nothing but death, death, death and most of it felt unfair. But read through the comments, some people love it. I don’t have the skill set the game requires… but the game is interesting. I’ve been excluded from it due to a fixed difficulty setting above my head.

    2. I think I could go with calling Lemmings a puzzle game because most of the levels did require a fair bit of the cerebral to figure out how to meet the goals for each level.

    As I mentioned in the above response to Alan, the only danger in a design that expects/facilities players to quit without 100% completion is making sure players are good with it. If I play GTA: San Andreas and decide to give up after finishing the first city Los Santos, I’m going to feel I didn’t get my “money’s worth”. Whereas I can happily give on collectibles, stunt jumps, side missions and so on. With Lemmings, it sounds like you were fine with it :)

    3. I did briefly cover that kind of “puzzle game” in The Box Impossible. For the purposes of Ouroboros, I discounted games like The Room and Hoshi Saga because they’re more like playing with a toy and seeing what works (GNOG, Vignettes, Vectorpark’s works) than solving something with rules of logic. Bart Bonte’s Yellow/Red/Black series is another game that follows the Hoshi Saga template if that’s the kind of thing you like. (I loved the Hoshi Saga games when I first came across them.)

  12. Oh, nice to see Machine at the Heart of the World. I was hoping to see it mentioned ever since I suggested it to you and I admit I was hoping for more text about it but still, it makes me happy that the design has spoken loud and clear that revisiting previous puzzles is encouraged.
    I am really proud of how much “intentionality” I managed to fit into Machine despite it being just a game jam production. I can’t not ask – did you enjoy the game?

    Ever since I played Celeste and read some commentary from the authors explaining the inclusion of assistance modes I had a huge shift in my approach to accessibility in games. And I am not talking about making games available for people with disabilities (which I am all for) but rather catering to the fact that different people like different games and like to experience them differently. And often with little effort you can create something that can be enjoyed by a very-wide public without interfering with your vision.

    What I just realized is that I should’ve had this epiphany slightly sooner – some time before playing Celeste, together with another DROD community member we made and released a hold specifically for another forumite’s 5-year old daughter. I know you expressed dislike/indifference towards DROD in the past, but if I caught your attention with this, you can check it out by yourself: http://forum.caravelgames.com/viewtopic.php?TopicID=42270

  13. @Joel

    “Sleeping, reading, going out are highly recommended activities and I wish I could do some of these myself :)”

    – Let’s not forget to eat proper and fresh food! Like I see in a meme somewhere, a joke that we should get water, take light, in the end saying we are complicated emotional plants. Talking about the Sun, it reminds me of that game Kojima produced for GBA, Boktai: The Sun Is in Your Hand, where equipped with a light sensor, you should play in the light for recharge your weapon, very innovating, but I don’t imagine me playing for hours toasting in the Sun.

    “(…) If you think commercially, that’s going to be a problem. (…) There have to be some gritty works out there which are clearly not for a general audience. We need that colour. (…) I am also aware creating puzzles which are “review-friendly” and/or have adaptive difficulty are not necessarily easy to do. (…)”

    – In a way I agree whit you, don’t know if we are in the same page, but in my mind discussing games always falls in the same “industrial”, “commercial”, “consumer” discourse. Since the very beginnings of game reviews, we tend to talk about games in terms of market or industry, since is comprehensibly, because video games begins in the field of electronic toys to the entertainment industry and the first magazines like “Nintendo Power” ware basically shop guides, with reviews of their catalogue, previews of new games and propaganda of the major ones. The issue I see in the community of video games and in the “pop” in general, there is, comic books, blockbuster movies, RPGs, TV Shows, pop music, music video (I now realised that “videoclipe” it’s not a term used by english speakers, of course…), etc, it’s made by an audience that not only enjoy to talk about what they enjoy, but try to make a kind of critic sense of the object they are talking and in the same time imagine their own ideal object. We see many people saying they want be writers, game developers, movie directors, not sure if this is related to fact the things they like and have pleasure with it, since based on fiction and so worked on imagination, gives a sense of freedom and inspiration to make your own movie, comic book, game and so on, because supposedly everybody have imagination for free, therefore being possible to everybody. Another thing is this culture of consumption, embracing the neoliberal idea that our reality is perfectly modelled by the invisible hand of the market, that competition is the base of progress, pure capitalism for freedom and where is more consumption, there is more development.

    So is OK to think that the entertainment industry as the fountain of our pleasures, must be incentivised to grown, because if grows, more interesting new products will comes out. It’s strange to see independent developers, talking about their own developing in terms of industry, if is independent it can’t be industrial, if it is industrial, then is not independent, the whole idea of independence is to be independent from the roles of market and his big industry, that is why a independent works are not make for profit, it is a endeavour to cast a project that no one would be willing to finance, because of the lack of capital return. The point is, thinking about design in the same time as thinking in the market, will inevitable kills originality.

    This is why we can have Star Wars episode 100, Jurassic World in Jurassic Universe in Jurassic Parallel Reality, Terminator: The Infinite Final Judgment, Super Hyper Ultra Mega Street Fighter, Avengers: The Recall and not having any single original interesting work. Even in technology, now we have Iphone XYZ 99999, that looks all the same, the tendency of something going right is to maintain that thing, if Kodak could, we would still be revealing photographs in 24 hours. Now Netflix makes original series, but as the company grows, the risks increases, Daredevil first season was very original, second and third just repeat the formula. On the other hand they will make new shows based on search field of big data of their users and on the expectations of the pop culture to create a product that will sell and be talked about in social media. There is nothing to do which the making something creative and good for their audience.

    I think Kickstarter is a reflex of this, it begins with projects of games everybody wanted to play, but the market simply ignored, not because they have their own vision of what would sell most, but to maintain the authority to make the exact same things that people will be obliged to buy. Now that Kickstarter becomes a market by himself, it reproduces a commercial formula to “sell” empty projects with no costs to people willing to pay for the next hype. I’m not saying that no one should use or share these projects, by the contrary, there is very good games that come out in these platforms. Not only for games, other things also, like many people using to support their studies or social projects. I think is a great idea, it should be used, not for big companies, but for independent people, who wishes to make something interesting, in the same time trying to survive and achieve a professional or educational success and can’t find their way in the normal market.

    To conclude, it seems to me the path to change the concept of games, is changing their ideas, be critical. It has a great potential, we all agree it is a platform very intuitive to pedagogic and educational purposes, it sends a message in way other medias can’t. It’s disturbing to finds out that a great portion of Europe doesn’t know what was the holocaust, in USA every week a different person in a mass shooting, ignorance in very corner. I don’t know the reason games can’t be used in way more intelligent, not saying nothing about action games for example, we could have an action game based on second world war that truly explains the social and political implications of that time, the terrors of war, without fetishizing the nazifacism or taking any sides besides the side of human being. I remember an article of yours that you complains of the lack of sand box games that can be played for children too, GTA it’s a great world, very alive, but with mechanics centralized in idiotic violence (not that I don’t play that either), so Minecraft becomes special to you since you could play with your sons and in the same time you too enjoy exploring the world and his kind of simulation”. That’s the thing, we can have products (not in the sense of consumer product, but in the sense of creativity work product) that are both fun, well done and educational, for children and for the adult, making a content that are interesting and critical with 18 + stuff everybody likes. And if the developers are luck, will get enough money, like Minecrat, Cuphead, Shovel Knight, Gone Home, Amnesia, etc, it’s a win win situation!

  14. Interesting to read your thoughts Pedro, especially about Kickstarter as a ‘parallel market’ of sorts. The most notable thing about it I guess is that it’s relatively smaller; no developer received ‘true’ funding for a project such that it was comparable to the development budget of even a medium sized modern-day game, and so we never got to see it acting as a true alternative route to market for under-served genres, even though as you say many positive things have come from it. Well, I guess there is one exception, Star Citizen.. but that’s a whole other rabbit hole.

    Joel, about this idea of allowing to disengage from a game prior to 100% completion, while still feeling a sense of having beaten the game. It seems to me mainstream (ie non-puzzle) games have a few ways of doing this.

    1. Hub-style games, with access to the ending gated behind X% completion where X < 100. Think Mario 64: You can confront Bowser with anything between 70-120 stars and see the ending. This opens up the possibility of beating the game to a larger swathe of players, while still giving the hardcore an opportunity for a sense of true mastery. Many developers have over time undermined this laudable concession to accessibility by providing a True Ending or other significant bonus for 100% completion, leaving players with a nagging sensation that by ending the game early they'll miss out on a vital part of the experience.

    To tie this back to puzzle games, I know for example that I feel like I beat Braid. I mean, sort of. After all, I saw the ending. It was a major relief: playing this game, more than any other, was a constant struggle against the anxiety that *this* was the level I would come up short, would be found intellectually wanting in front of Mrs. Blow and the rest of the class. And yet later I discovered that I needed to do some frankly bullshit nonsense to trigger the true ending to the game which would reveal that it was Really About nuclear war or something.

    In this sense the game begins to feel like an elaborate joke played on both sides: actively wasting the time of the completionists, mocking their attempt to achieve total mastery with large tracts of opaque, frustrating or null gameplay, but also mocking the middlebrow normies patting themselves on the back for beating An Intellectual Game without ever getting close to the True Secrets which radically transform its meaning.

    2. Post-game content. This is different to 1. in that the player doesn't choose when they're done with the game, but is rather provided an 'out' at the developer's discretion in the form of the (what may only be the first) ending and credit sequence. There's more content to come, but in this way it's signposted as optional. You're still negotiating the tension of what constitutes true completion based on what incentives you provide the player for beating the post game content (the true ending rears its head again).

    Sorry. Getting off topic here.

    3. New game plus. Varying implementations of new game plus modes have cleaved a clear philosophical divide between 'the player wants to experience the content again, so let's make it easy by letting them keep their power progression' and 'the player wants to experience fresh challenges, so let's rebalance the difficulty for the dedicated experts'.

    It's the latter that I guess is more relevant to this discussion, although I'm not sure I've ever seen a puzzle game offer a NG+ mode. How would that work? Puzzles, unlike combat encounters, have less variables to play with to revise the difficulty standard. Perhaps you demand the player be more efficient with time or move restrictions. Maybe you just straight up offer new puzzles. In any case it could theoretically be a way to 'hide' some of your game from all but the most dedicated, providing license for the less than hardcore to step away without feeling they were abandoning it.

  15. @CA

    “Interesting to read your thoughts Pedro, especially about Kickstarter as a ‘parallel market’ of sorts. The most notable thing about it I guess is that it’s relatively smaller; no developer received ‘true’ funding for a project such that it was comparable to the development budget of even a medium sized modern-day game, and so we never got to see it acting as a true alternative route to market for under-served genres, even though as you say many positive things have come from it. Well, I guess there is one exception, Star Citizen.. but that’s a whole other rabbit hole.”

    -Glad to bring some reflection. Trying not be so harsh, I was very optimistic, indeed. To truly and sincere independent projects backed by Kickstarter and similar, not only it’s not sufficient, if it becomes a market by itself, it loses the purpose to a platform of collective financing, so you can’t seek profit from it. Well not for all the projects from people who uses it, but the Kickstarter itself has a business plan that involves that very profit’s seek, which firstly can bee having x % the funding and his excess, secondly can be making certain projects more visible than others (more like Steam does) and third making direct business with people or firms that wants to be funded by the platform, but in the same time having all the resources and consulting of the Kickstater. For mere mortals, who doesn’t pass from any hype or visibility, doesn’t have any backup, will use platform for example to getting funds enough to make that project comes true. Not every thing it’s make from scratch, many just need a financing to print the copies of a book, or pay a server or pay the actors, if is a theater’s play. There is others objectives too, Kickstater can be used for marketing purposes, or just to be the “kick” that will get you “started”. A light in the end to tunnel or the begnning of a tunnel. So, for an independent developer it could be naive to think you will get everything you need from a “collective financing” or “crowdfunding” platform. I guess my “optimism” it’s coming from the Kickstater, but from the possibility that people can get what they want with a collective thought, so if there is for example, a student needing to funds to go travel and make a University in another country, people can be sensitive and help. But this “optimism” it’s not a truly optimism, since what happens is that people who doesn’t need the money getting the money, using for example schemes. So someone can make a project so interesting, like a game, that will attract all the hype and all heat, will get 1 million in cash, promising, but not delivering, actually, disappearing. Going even further, people using it to fraud an social action, like the group who was faking taking an ex soldier from the streets. Therefore, this is the “parallel” market, in sense that as demand increases, will create a “corruption” that will move all his forces to make more and more profit, I don’t use market, in this case, in sense of good and positive, my questioning is the notion that “market”, “industry”, “company”, “competition”, “pure capital” is synonymous of “progress”, “innovation”, “good”, “development”. So this distortion may be harmful for people that is making “indie” content, not only by comparing himself as part of an industry, killing the purpose of independence, trying to seek be part of the big market, but believing that as in the market, will, like a golden ticket get things “done”, like an invisible hand that takes care of everything. Thinking about it, trying just to be rich and successful it’s not a winner formula and it shouldn’t be anyway.

  16. I thought Recursed’s earlier levels were pretty well-paced even with the repetition. And yet I’ve seen quite a few people give up on it early on for being too hard.

    I wonder if there’s a good way to solve this problem of “making sure the game caters to everyone”. I’ve been thinking about difficulty settings for emergent-mechanics puzzle games for a while (e.g. Recursed, The Witness, Portal, Braid), since I also agree that they’d be good to have in the same way that many other games have difficulty settings or assist modes.

    I think Celeste (Matt Makes Games) managed their Assist Mode well. The player is encouraged to use it if they need to, but also encouraged to give the game a good go without it first. And the game certainly doesn’t shame you for using it either. Something in the same spirit could easily work for puzzle games. However, non-puzzle games, tend to simply tweak a few parameters in their games, like enemy accuracy, enemy count, or player health; something that isn’t so easily done in puzzle games. Here are two ideas I’ve come up with (completely untested of course, so take this with a shaker of salt):

    Idea #1: Construct alternate versions of the same level, based around the same sort of concept. Upon a player completing a level, allow the player to go onto the next level or play a previously-beaten level. Playing a previously-beaten level will actually give the player an alternate version of that level, so that the player has more practice familiarizing themselves with some idea.

    Of course, you’d have to communicate to the player that replaying a previous level gives a different level on the same idea, and that it is for practice purposes alone; that game completion and achievements depend only on the “levels” beaten and not the alternate versions, so that experienced puzzlers can breeze through the game without ever feeling guilty about not having completed all levels.

    From what you say, Invert seems to be something like that, except for the facts that Invert’s levels for each tile formation differ greatly in difficulty and that it seems like you need to beat all the levels in a tile formation to proceed.

    With Marcos Donnantuoni’s Dis Pontibus, which uses autogeneration and autocuration of levels, it seems like he could pull this sort of thing off without too much of a sweat, as he could generate variant puzzles of similar difficulty and pop them all in the same puzzle. (I think hand generation might well be needed to make sure that the levels are in fact similar in idea.)

    Idea #2: Give the player hints that are explicitly about the idea in the game. Further, have a bunch of difficulty modes (that can be changed in-game) that give different degrees of hinting.

    Let me use Reach from Recursed as an example:

    * At the lowest difficulty, the hint could say “In this level, the chest is green, and it contains the room you’re in. If you exit the room, the chest will stay where you put it. Further, you’ll pop out of the chest no matter where it is. Therefore, if you enter the chest, and throw it up to that ledge, then exit, you will pop out up on that ledge. You can use this to get to the exit.”
    * At the next highest difficulty, the hint might say “In this level, the chest is green, and it contains the room you’re in. If you move the chest and then exit it, you’ll pop out where the chest now is. You can use this to get to the exit.”
    * At the next difficulty, the hint might say “Try entering the chest, moving it, then exiting it.”
    * At the next difficulty, the hint might say “Experiment with the chest.”
    * At the final difficulty, the hint might say “You are on the highest difficulty. Hints are disabled.”

    At the start of the game, when the player selects the difficulty mode, show them examples of the various hints so that they can judge for themselves how hinty they want the game to be.

    This would have the drawback of potentially also fragmenting multi-concept levels into multiple single-concept levels. For instance, Chests in Recursed is a multi-concept level: you can enter chests, chests take you to new rooms, you can exit chests, you can move items out of chests, and you can move chests. Talking about all five of these concepts in a single hint would be difficult without spamming the player with a wall of text (which the lowest difficulty hint above is already close to). You could split Chests into two different levels on the lower difficulties, but this probably raises other issues design-wise; what levels should be presented if a player switches difficulties midway through a puzzle?

    Although, if we’re going to go that far, perhaps we should simply have different levels in different difficulties. For instance, CrossCode (Radical Fish Games) hints strongly at its solutions to dungeon puzzles by putting markings on the floor basically telling you in which direction to shoot. Of course, it’s possible to miss this entirely, so perhaps on higher difficulty levels they could have simply done away completely with the markings, and on lower difficulty levels they could have made the markings more prominent. Maybe hints in the level design (e.g. signs and surface placements in Portal) could be omitted or made more prominent in puzzle games’ difficulties too. (Then again, am I just suggesting to cripple a game’s level design to make it cater to hardcore fans? I don’t know, I’m tired at the moment.)

  17. Yeah, I deliberately made the very first levels of Tricone slow-paced. This was because I had (perhaps foolishly) chosen to make the gameplay environment totally abstract (i.e. without relation to real-world concepts) and rather unlike well-known game genres. So the player brings almost nothing of use to the game in terms of prior knowledge. So every aspect of Tricone’s mechanics had to be presented clearly, and then quickly and repeatedly reinforced so the player doesn’t forget it.

    The first “true puzzle” of the game is the 9th level, “BreakSlice”:


    It appears after 8 simple tutorial levels. By this point the player knows exactly what to do, and how things work. The median time to solve is about 2 minutes but I’ve seen many players bash their heads against the puzzle for quite some time. One YouTube reviewer actually failed to solve it and gave up the game completely at this point!

    The game gets tougher later on. I have seen one or two players get bored during the initial sequence and ditch the game, presumably assuming that the game will never present a challenge. But mostly, adept players will just slice through the tutorial stages quickly without wasting too much time before they get to the tougher core of the game. On the other hand less confident players will actually struggle a little bit with these earlier stages, which is good because it means they’re learning what they need to and getting a bit of confidence-building and satisfaction.

  18. Hey CA! The whole thing about completion vs. post-game stuff is a big part of what’s behind my obsession with level-selects. Did you see my earlier rant about Closure’s level select and the placement of its credit sequence? One of the big aspects of that was that the level select was fighting against all the cues about when the game had ended. The level select drives you from one level to the next without regard for whether you’ve done the secret things that aren’t required to unlock the next level, at the end you find out that you really have to do all those secret things, and then the level select puts a lot of friction in the process of going back and collecting them all.

    About Braid: I figured that the process of unlocking the secret ending was so laborious that it couldn’t possibly be considered necessary. It’s a pretty cool insight to realize that if you stand on that one cloud for forever something cool might happen but who has the damn time for it? A consolation is that the reward seems to be more of Blow’s Prose And Ideas About Things which is something I am actively interested in avoiding.

    I hear someone has made a video about The Witness and its ideas about 100%ing games but I’ve never watched it on the offchance I might play The Witness. (In related news, I saw TW was 75% off and was like ooh and then I realized that that’s the same price as Into The Breach at 33% off. Which means, I think, that my next game purchase will not be The Witness.)

    Joel–so I thought of Snakebird’s level select specifically for this one article, because its particular way of nonlinear unlocking does suggest a solution to the dilemma in the post. If there’s a mechanic that you get, you can jump ahead to other things–if you’ve solved it but haven’t internalized, you can do more things with the same puzzle. I won’t claim I get any mechanic in this way, but stumbled into unlocking levels 4-> 21-> 43, which was the first level with a moveable piece and was a right bastard at the time with like three impossible-seeming things to do. (It has a U-shaped gate with fruit inside it–you have to figure out how the snakebird that eats the fruit can possibly get out, then you’ll have another situation where a snakebird looks irrevocably stuck, then getting to the exit is a challenge.)

    Whereas if you proceed in a more orderly fashion, the first level with a moveable piece is 22 (unlocked by 12, not 21), which has one snakebird and a square that you move from landmark to landmark in a fairly intuitive way–that is, it’s not unchallenging to figure how to get it from place to place, but I didn’t have any “this is impossible moments.” A much gentler introduction to the mechanic, and if you start there you’ll do more levels which are just about manipulating gates.

    In educational parlance it’s more like the adaptive classroom, where the teacher is supposed to present everyone with challenges that are suitable to their learning level. To really work this way the levels would have to have a bit more repetitiveness of ideas in them, so there would be some that were mostly about confirming mastery that you might skip with a clearer conscience.–And then there’s the question of what the person who skips ahead does with the uncompleted levels. Does the game expect you to clean them up in some sort of flow state, or is it OK if you leave them skipped? Or come back to them after you’ve seen the ending?

    My experience with Swapper was a bit like this–most of the areas can be unlocked without finishing all the previous puzzles, and I think I tended to skip them because I wanted to see what was next rather than because I was stuck on one. But then you need to have solved everything to see the end. I thought that was an unfortunate design choice, though the way the game works it wouldn’t be possible to let you go back and clean up the unsolved puzzles after you’ve seen the end.

  19. Hi Matt,

    I’m afraid I didn’t see it, but I’m not surprised to see that someone has covered it before on here. Very interesting to read your thoughts. I think we’re on exactly the same page with this stuff. While some players need to complete 100% of what a game has to offer, and some players won’t care at all, there’s a third category, which might either be the majority or perhaps just you and me, who want something in the middle, the Goldilocks experience if you like, and rely on the game itself to confirm which of its parts constitute that through its interface and design.

    Where you put your credit roll is an important decision for designers to think about, and I think the credit roll is a monolithic enough thing that developers will indeed put a lot of thought into it. But there are also other design elements that are important for signposting to the player the structure and (the other kind of) flow of your game, as you’ve discussed. I agree that here game designers sometimes seem to maybe not think through all the implications of the decisions they make quite so thoroughly.

    When it comes to this question, developers have a… I don’t want to say ‘responsibility’, but it’s something like that. They’re the ones designing these games after all, assembling their constituent parts. They could be displaying greater intentionality, is what I’m saying. There are games that do it well and games that do it poorly. Often games will offer menus and progress bars to keep track of the optional collectibles scattered throughout their worlds: the designers recognise that it causes anxiety in some players not to track and acknowledge their progress. But they’ve been less cognisant of anxiety regarding game completion in a more general sense. As we’ve been discussing, there’s no general convention in place in games for allowing a player to disengage from a game when they’ve had their fill: indeed, many games are designed to maximise player engagement, to keep their hooks in the brain and discs in the tray, with explicit mechanics designed to tempt/guilt us into to coming back at least once a day.

    Of course, assuming developers were turning their attention and intent to this question, the solution they provided wouldn’t necessarily work for everyone. As with difficulty and the flow channel, it isn’t easy – perhaps it’s impossible – to please everyone in the diverse and stratified group that may come to play your game. Could we find some design solution that allowed players to decide for themselves when to cut ties with a game, while still giving them a sense of closure? Would players welcome that?

    The open world genre seems to be at an advantage here. Take Breath of the Wild, for example. The game is famous for being open to completion from almost minute one: tutorial area over, you can beat the game in as little time as it takes to reach and defeat the final boss. At all times the player is in control of how much of the game they want to experience before they go ahead and finish up.

    But there’s more to it than just that. The game doesn’t tell you (as best as I could tell) how many shrines there are in the game. Nor does it tell you the percentage of the total korok seeds you’ve collected. In not disclosing these numbers, the game subtly signals that these are peripheral activities. So the game discourages an anxious, completionist mindset (‘you’re ending the game? Why, but it says you’ve only beaten 72/120 shrines.. are you sure?’).

    But at the same time it might also be bewildering and anxiety-provoking to leave everything entirely in the player’s hands. This is a big world after all. So the game provides and tracks some progress tent poles: aside from beating the final boss (the overall objective), you have five major quests: rescue the four ancient Hylian beasts and recover the master sword. You have a secondary screen for sub-quests you receive out in the world. You also have the map, which starts out dark but is fleshed out once you climb the control tower in each sector. Finally, you have twelve memory points to track down in the world which provide a large chunk of what passes for the game’s main story.

    Taken together, these constitute a proposed structure for the player’s experience with the game: you don’t have to find everything, but here are some things you might want to do. The player can take or leave them: they can go straight to the final boss, or they can turn off the UI and go climb the nearest mountain to see what’s on the other side. The game supports free-form, user-directed play. But it also curates a hierarchy of checklists from which the player can assess their progress through the parts of the game the designers have decided are significant (but not essential) markers of completion: have you done the main quest tent poles, have you at least visited every region, have you seen all our cut scenes? Yes? Great! Well, whenever you’re ready…

    This is all presented through the game’s UI. And to reiterate the most important bit, the activities which rank in the hundreds (have you collected all our seeds? have you beaten all the shrines?) are NOT represented in terms of ‘you have done X of Y’. The game isn’t designed to encourage the player associate completionism with completeness, if that makes any sense. Players are directed towards discrete objectives; the more sprawling, collectathon stuff is left to the player to prioritise for themselves. This is quite different from say, the Ubisoft approach, of just vomiting a carpet of icons across the whole map, with the implication that the player won’t really be done with the game until they’ve scrubbed it clean.

    It’s not as though this approach would work for every game, of course. I’m not sure designers of linear games would find much to usefully apply. But I appreciate that Nintendo appear to have put quite a bit of thought into how they went about it here.

  20. I just realised that “you can beat the game in as little time as it takes to reach and defeat the final boss” applies to, er, almost every game ever? But in this case the boss is waiting for you, clearly signposted, in the castle at the middle of the map. There are almost no preconditions to challenging it other than physically reaching it, which isn’t true of many linear games.

  21. Just replying to a few island comments with this one. Hope to respond to the rest later. The week has been brutal as expected.


    This piece naturally brought out Machine at the Heart of the World but I’m not done with talking about it yet. I don’t know if it will feature in Ouroboros again but I will sing its praises a bit more in January. I really loved it, finished it over a week ago — without hints I might add. I can understand why some might have found it unreadable; were the occasional narrations added after complaints or always there from the start?

    I have a playlist in the newsletters and it got a little heart beside it which highlights games I’ve *really* got something out of. And that ending! That’s all I’m going to say here.

    On “accessibility”, yes, this is the point of view I have slowly shifted towards through my puzzle tribulations. There’s a response to this article over on Reddit making the argument that casual puzzle players have hidden object games and things like that, whereas it’s hard to find something challenging enough for the niche hardcore. This is straight out of the “hardcore gamer” textbook though: we’ve seen this kind of talk bubble up around AAA fare for years. It also oversimplifies the problem that there’s “casuals / hidden object game people” vs “professionals / sausage rollers”. That neat partitioning is far from the truth and that’s the problem.

    If you keep chasing a smaller and smaller hardcore niche eventually you’ve got games that no one can make a living from.

    It would be tough but this type of accessibility would be important. Puzzle design still feels like one-size-fits-all approach with a few weak supporting mechanisms to aid players who aren’t quite getting it. I’m addressing some of this with two Ouroboros episodes to come.

    Lovely that you’ve tried so hard with that DROD hold. I haven’t looked at it (I did read through the forum page) but I think it’s a great idea.


    Hi again Josh. Yes, my impression was you were being careful with the pacing, that Tricone has a LOT of different types of craziness and the UI helpful enough to keep it all in order (although I did use the help function sometimes!). This goes back to that paper on mental models, where the presentation doesn’t do a great job in attempting to bridge the abstract ideas of the game to existing mental models. I memorised that the letter “Z” meant breaker. That isn’t actually a letter “Z” but that’s how I was able to remember it.

    The interesting thing is how the level select just sort of explodes, more levels keep being added but even if you get stuck there always seems to be other levels to have a go at. Of course, that process won’t go on forever, but it felt very generous.(If there’s another thing I’d improve it would be the level select because it feels a bit like a giant list!)

    But I personally appreciated the slow pacing because later it seemed I needed that grounding. It was almost a surprise to think, wait, hang on, why can’t I see a solution?

  22. Okay so this is a general response to the conversation between Pedro, CA, edderiofer and Matt. We’re all armchair puzzle designers today.

    We start out discussing the age-old problem of creativity under capitalism. That money inevitably shapes product and financial risk skews commerical art toward replication and imitation, towards safety. I guess I would approach it like this – if you’re trying to make money, I think the puzzle accessibility may become as commonplace as undo down the line, we’ll wonder how we ever did without it. It’s basically a usability feature. The reason I spiked the commerical aspect is because – why would anyone go to all the extra effort in the first place?

    Those scratching ideas into the halls of puzzlescript and raw indie projects – you’re unlikely to find adaptive difficulty/accesibility improvements there. Some of these people are only just about able to get the idea out at all. The idea is what we’re interested in and that will contribute to puzzle design lineage.

    I think adaptive difficulty/accessibility is more likely to emerge from a commercial angle. It’s all very well designers building for the hardcore crowd but, in time, it’s building for a smaller and smaller audience. The commercial makers need sales to keep going. (For reference, here’s the Reddit comment rejecting the premise of the essay.) And as I’ve tried to clarify, there are so much more aspects than some people get things this faster than others; bringing experience of Sokobanlikes to the table, for example, means you have a head start on someone who hasn’t really played these with any dedication. Some of these games are like books with RTFM written on their exterior, because the designers are aiming for puzzle jockeys like themselves.

    The interjection regarding Kickstarter… I am not so hopeful on that. As CA alludes, Kickstarter does not really cover costs any more and there is an argument that nostalgia/vapourware tend to attract the most funding and, further to that, stretch goals are more likely to kill a project than help it. Kickstarter as Pedro later discusses is just another type of market, a beauty pageant trying to attract dollars any way they can. It’s the same as the regular market, but different.

    Moving on to ideas to solve these puzzle accessibility issues…

    CA discusses the hub idea. In my “secret” Witness video, I was critical of the decision to make extra content gated behind 100% completion while Jon Blow was out there telling colourblind people it’s no problem, you don’t need to solve every puzzle to finish the game. That is total bullshit and exposes the design problem with the hub approach. Completionists tend to demand special rewards for 100% because that process in many titles is some masochistic; The Witness is no exception.

    I quite like my experiences with GTA III, though. And CA later calls out Breath of the Wild as a similar example. GTA III has a main campaign with a single ending. But there are all these sidemissions and alternative targets which have no impact on the main story. Many of these do produce rewards – like 100%ing the firefighter missions makes you fireproof in GTA III – but they do not change the nature of the core campaign. I note that Cosmic Express has this kind of structure – you can get to the end of the game without finishing every single level. PLUS there is no additional reward for 100%ing the whole thing with every secret challenge. I was okay with that because the satisfaction of 100% was enough for me… but I am not everybody.

    CA also floats the idea of a New Game+ approach to puzzle games. I think this only works if you have procgen puzzles; you can’t make players run through the same challenges again unless you do a Cosmic Express style twist where the same levels have to be done a different way. And if you’re doing that, then you get into to the suggestion that edderiofer makes…

    …alternative level designs. This is the Thief approach which I absolutely loved but the problem is it ramps up development costs invisibly. You may be creating double or triple the number of levels, but will need the same amount of testing for the game to work. Players may only count the single content they see. This is why I think proc gen is the only way forward for adaptive difficulty. Dissembler demonstrates that it can be done; Dissembler produces puzzles for six different difficulties each day. More complexity is obviously going to need serious proc gen work. (Ian MacLarty says that procedurally generating puzzles with the new twists in Dissembler is not straightforward, so you’ve only got handcrafted content for the new mechanics.)

    The other suggestion from edderiofer is hints. I’ll expand on my reservations on hints in the last few Ouroboros episodes in January but as with all these ideas they impact players in unknown ways. The most basic problem is taking an optional hint can feel like failure; possibly masking the hint mechanism as a tutorial can work but if we let the player choose the level of “tutorialising” then this feels a little awkward. The other issue with hints is that they can inhibit learning and induce a cascade of hint dependency. A tertiary concern is that hinting mechanisms more explicit that arrows pointing on the board (see Dissembler!) can feel a bit cheap. “TRY MOVING THE BOULDER AROUND THE PIT USING THE LATERAL CROSS LASER” flashed up on the screen. Something we can get into later. Dead in the Water was pretty scary to write because it opens up a lot of areas of discussion :)

    Matt, I also get the point about Snakebird’s level select but personally I found it was unhelpful. I started jumping to new levels when I got stuck, but that just propelled me into new, unfamiliar shit. On my second run, I forced myself to go through in order. I don’t know if Snakebird really does “unnecessary levels”, so the skipping seemed to be there more for players who were stuck, opening up options rather than adaptive difficulty.

    To round up, I think we can all see there are possible options but all of this is expensive. Our armchair design thoughts are going to need some puzzle designers to bite the bullet and try these and other solutions to see what is doable. And context will be king: if you’re trying to solve the handcrafted content + adaptive difficulty problem is more painful than that for procgen levels.

  23. Thanks for your comprehensive response Joel. It’s surprising how such a seemingly innocuous question quickly billows out in size when you start to discuss it. If all design challenges have such inflationary qualities it’s a wonder games get made at all!

    Regarding the Witness, I didn’t hear about the controversy at the time. What was it you found bullshit? That Blow didn’t add accessibility options for the puzzles affecting the colour-blind, or that he thought that the hub structure excused him from having to do so?

  24. @Joel
    > […] were the occasional narrations added after complaints or always there from the start?
    To answer your question – no, the only person who really tested the game before it was released for the jam was me (I had Cage, the graphic artist also play it a bit but just to make the art he had to know the rules), which is also why the level design is spotty. From the get go my goal was to introduce each interaction in a separate stage and, for the most part that worked, but when I replayed it over and over again and talked with Cage I decided that the game needs something that would push players in the right direction (again, especially in later stages).
    What I initially wanted to do was to have tiny inscriptions visually attempting to explain the interaction but with this being a jam game we knew there won’t be enough time for it.
    So what I ended up doing was researching what kind of symbols can be used to explain the mechanic and later adding narration.

    In short, it wasn’t there from the start nor added after complaints – it was introduced later during the development :).

    Also thank you for the praise. It really rekindled my spirit for making games and made me actually start working on something again (kind of)!

  25. CA

    That’s hitting the nail in a nutshell, alright. If you’re “serious” about design, oh my god, you can fall into some deep rabbit holes. It is no wonder game jams are very popular: you just get out of the way and let the game happen.

    There was and wasn’t controversy. There were complaints that if you were colourblind, it was impossible to solve the colour puzzles. Jon Blow replied that it had been hard to solve this accessibility problem, so they solved it by allowing you to get into the mountain without 100% of puzzles. Yet, to fully explore the game you still need to 100% it. You can’t reach the challenge without 100%. Without the challenge, you can’t obtain the final film. And while you can technically still reach the other ending without 100%, you don’t get that vital additional hint if you haven’t 100%ed.


    That’s good to hear. The reason for the question – I was worried that I was playing an “easier” version of the game that got lots of other people stumped. Not true, I can still feel smug! I really liked it. Planning to stream a little bit of it in Feb. (This is how my schedule works…)

  26. Maurycy:

    I just played through Machine, and I’m really looking forward to further discussion (hopefully Joel and you will agree to spoiler-ly discuss what you were going for with the ending – I don’t think I wasn’t paying attention or that your environmental storytelling was weak, but I am still a little flummoxed). It was very good!

    Regarding the difficulty and learning, I hit the snags you mentioned in your post-mortem, but they were pretty surmountable with how easy you made it to look back at previous puzzles to figure out the nuances of the rules. It helped that you had some yin and yang ambiguity early on to prime the idea of looking back to refine hypotheses. It all felt pretty fair and learn-able, though the screen-filling puzzle was a bit of a jump. I also liked looking at the “diary” after I was done. If you ever turned this into a “more than game-jam” game, building that into the game would be a nice springboard for putting in Celeste-style the accessibility options.

    Also, contra-Joel’s expectations, I really can’t think of any games I’ve played that made it so easy and painless to review past puzzles and work on theories about how things work. That was super great. The only tiny quibble I’d have, is that it’d be nice for the game to remind you to check past solutions after a number of wrong attempts (maybe it does but I didn’t reach the limit because I’m soooo clever and only messed up 10x per tough one?). The keyboard overlay disappeared before I was 100% sure I’d committed everything to memory, and I didn’t see a way to get it back. I just already knew from the write-up here, that reviewing past puzzles would be key. But I can see a 100% unspoiled player not thinking to go back and getting frustrated.

    So yeah – neat game! It definitely belongs in a group of “good discoverability games” with Illiteracy. Not sure of what else would go in there..

  27. Dan, drat it, I was sure someone would be able to dredge a game up that offered immediate review! Good to hear you got through Machine! The big puzzles were definitely quite sudden but I think of them as boss monsters. There were a couple of puzzles I guessed through and intended to go back to understand them, but understanding wasn’t required :O regarding the end SPOILER SPOILER it was just unexpected that on the verge of getting to the heart of the mystery you were stopped by the unseen perpetrator, no resolution more like a short story with a grim, haunting end

  28. @Dan
    Some of the things I’ll say here will probably be duplicating what I put in the post-mortem, but to address some of your remarks:
    > though the screen-filling puzzle was a bit of a jump
    The short answer here is reiteration of Joel’s interpretation, these are boss stages. Or exams, that’s what they were called in my head when I worked on them, their task was to ensure that you’ve internalized the rules so far.

    > I also liked looking at the “diary” after I was done
    So Joel has asked me earlier whether the narration was added due to complaints, and that’s where the diary came in – I saw people ask me the same questions (+ post-release I noticed some level design issues) which prompted me to create the diary as a way to eat the cake and keep the cake. The game stays true to the initial vision, but at the same there is one canonical hint source for players who can’t get past a given section or don’t care to trial-and-error their way through. I guess it’s a kind of equivalent to the “Accessibility” option in Celeste.

    > If you ever turned this into a “more than game-jam” game
    If I ever do anything else with this idea (and I kind of want to) I don’t think the resulting game would use the same “perspective”, so to speak. One huge problem in this game is that the measure of success is binary, unlike The Witness where you get told exactly which rules you broke. I touched on this in the post-mortem. Non-jam version would most likely have things switched around – you’d be the one providing the input (placing glyphs) requiring you to get the expected output. This way at least the measure of success would be non-binary, you’d see the output you created and the expected output and would be able to work from that.
    The downside of this is that it makes it kinda like the numerous programming games, except you work with a very alien machine.
    But I am stuck trying to figure out how to frame the game.

    > hopefully Joel and you will agree to spoiler-ly discuss what you were going for with the ending
    Joel got it right again :). But to elaborate on this, in the intro as the protagonist enters the cave he encounters the caretaker “a dried up body of a small man, his left temple horribly broken” which heavily implies the caretaker was murdered, but also he mentions “as I laid my eyes on the enormous machine I knew it was the root of the problem”. These should clue you that what what you are doing is fixing the machine to make the world work properly again.
    With these two data points it should be an obvious[1] that the protagonist gets killed by the person who stopped the machine in the first place, they probably want things to stay the way they are.
    For some reason *insert ominous music here*

    And Joel’s comparison to a short story is excellent. I wouldn’t dare make an ending like that in a bigger game, at least not without some foreshadowing, because it would be a very uninteresting twist in my opinion. But it works great in a tiny game like that I think :).

    I am really happy you enjoyed the game, thanks!

    > Planning to stream a little bit of it in Feb. (This is how my schedule works…)
    Is there a way I can know when to tune in?

    Btw, speaking of The Witness, a lot of people mention sections which are color-blind unfriendly (and in my case bad-color-vision unfriendly too, I had to do a bit of guessing in that one place), but not a lot of people say anything about quite a bunch of puzzles not only requiring hearing but also good hearing. I have this problem with my hearing where if I hear more than one sound source at the time I have huge trouble discerning between them, and the forest was a huge pain to me.

    Also I hope you don’t mind me sort of directing the whole comment section for this post towards my game.

    [1]: I say “should be obvious” but in reality I know first hand that the unspoken-but-obvious things can be not-so-obvious if you’re not the author.

  29. Maurycy! You made Flash DROD! I’ve been playing that a lot especially since my laptop broke and I’ve been using the obsolete laptop! Thank you!

  30. Maurycy,

    As I played over multiple sessions (three, I think) I completely forgot the dead body in the opening. I guess that’s a problem with a puzzle story that only has two parts, beginning and end. This works for a much shorter puzzle but not so much here. But I still *got* the ending, so not a huge problem or anything.

    I’m also unsure if this could “go large”. There’s definitely scope for a larger game but I share your uncertainty about whether it would prove to be eventually frustrating and get bad scores on, say, Steam, even though some of us with find this nutrituous brain food. Here you apply heavy gating to control the delivery of concepts but I’m sure with some effort it could be turned into “multiple tracks” where if you got stuck with one you could switch to another.

    “Also I hope you don’t mind me sort of directing the whole comment section for this post towards my game.” Oh, don’t worry about that, the comments get derailed all the time.

Leave a comment

Notify me of followup comments via e-mail.

No trackbacks yet.