Starseed Pilgrim

Reviews of Starseed Pilgrim were fixated on its sense of mystery rather than its challenging core. It became an indie title notorious for inscrutability and dividing players into two camps, The Ah-Ha-I-Geddits and The Emperor-Has-No-Clothesies. Starseed Pilgrim was intended as a B-side to another game, Probability 0, so becoming the latest indie game talking point in 2012 was a surprise to its developer, Alexander ‘droqen’ Martin. But its design was no accident.

“There was an article marvelling at the way Half-Life 2 guides the player through its first level through design that makes the ‘right path’ apparent. The wrong paths were, of course, dead ends, and the seemingly open level is actually very linear,” Martin tells me. “Anyway, I thought that was gross and stupid and decided that was the antithesis of everything I wanted to accomplish.”

“I didn’t add instruction because I believed in what I can only describe now as ‘discoverable systems’. Systems, rulesets, that are interesting to discover yourself and which you wouldn’t want to have spoiled by instruction.”

While the AAA industry gravitated towards telling the player exactly what to do sometimes to the point of alienating them (Dead Space’s ‘cut off the limbs’ anyone?), indie games have been consciously exploring what happens when the player is left to their own devices. What are the benefits for players – and the risks for developers?

Martin’s ‘discoverable systems’ are more ubiquitous than they seem. Cart Life, deceptively subtitled ‘a retail simulation’, is awash with unexplained systems, one of the reasons it swept the IGF awards in 2013. Roguelikes and roguelites hand players enough information to get started but often leave plenty to be discovered; consider all the mechanics awaiting new players in Spelunky, Duskers, Teleglitch and early access title Beacon which I discussed in last month’s stream.

In Future Unfolding, the player is stranded in a procedurally-generated forest with no indication of what they are supposed to achieve or how. Andreas Zecher from German developer Spaces of Play explains, “We wanted to create a game that was truly about exploration. Not only about exploration of the game world, but exploration of its possibility space.”

This desire to adapt mechanics as a medium of exploration also inspired Santa Ragione when they designed Mirrormoon EP, which is part space exploration, part procedurally-generated puzzler. It’s notable for putting the player in front of a spaceship control console stripped of text and icons. How do you even turn the thing on?

Mirrormoon EP

“We wanted to communicate this concept of exploring, of getting to know a new thing, through experience, through unguided interaction,” they say. “From the ship’s interface to the planet’s puzzles all is an expression of a single theme that is almost entirely conveyed without text or verbal content. During the development we often used the metaphor of playing with a broken radio – you’d know it was broken, and yet all those knobs, buttons and levers breed an untold story, hold a misery; even though interacting with the broken radio generates no output, you can imagine yourself operating an alien device you don’t understand and be rewarded by that pure analog interactive pleasure.”

Discoverable systems reframe a game as a laboratory in which the player must conduct experiments but this can be as much a practical decision as an aesthetic one. Ellipsis is a self-styled ‘avoid ‘em up’ action game, where each tense level typically lasts only seconds. But there are no instructions to be found. Developer Yacine Salmi says, “I suck at writing good text or making fun intro videos. I thought if we could stick with zero in-game text it would make localization a lot easier.”


Some indie developers have been using light touch tutorials for years, where highly constrained level design is used to teach mechanics rather than direct instruction. They are the simplest form of a discoverable system. In Alan Hazelden’s puzzle game Cosmic Express (Hazelden & Davis, 2017) all the mechanics are learnt through play – from the basics of drawing a rail to the slimy alien passengers that ‘soil’ a carriage. “It fits my design philosophy,” Hazelden tells me, “and in some ways it’s actually less work than writing text and hoping people read it.”

Hazelden treads carefully to ensure the tutorial element does not dominate the design. “I ease you in, but I try to avoid completely trivial puzzles – each one should feel distinct from what’s come before. Some games would make sure you’ve got it by giving you ten repetitive easy puzzles, which I absolutely hate. Obviously the ‘throw you in at the deep end’ approach can work, but I think it’s less accessible and more confrontational than it needs to be.”

Cosmic Express

Explicit tutorials are maligned as getting in the way of the real game and it is well-known that no one reads manuals no matter how many times designers might yell the techie pejorative RTFM. Don Norman’s 1988 book The Design of Everyday Things explored how objects communicate their utility and he termed how an object can be used an ‘affordance’. The magic of good design is to make affordances obvious to the user and Randy Smith, designer on the original Thief series and Spider games, gave a great presentation on affordances for puzzle design back in 2009 which you can watch on the GDC Vault.

Exposing affordance is superior to a tutorial because it makes potential interactions feel natural rather than a tussle with arcane rules. The problem with tutorials is that they are painful to revisit, particularly when integrated into the levels of a game. Modern titles assume once they have taught you something you will never need to be taught it again. If you take an extended break from a game, you might have to start again if you need a refresher for the controls.

I recently took a break from Megaton Rainfall, a beautiful but flawed first-person superhero simulator. On completing a level, Megaton Rainfall bestows a new superpower on the player which means after several levels I was struggling to remember all of the controls. After my break, I had forgotten everything. Hundreds of people were dying because I couldn’t get my superhero protagonist to “drill” into the ground and take out a subterranean enemy. I tried guessing but, eventually, I just quit the game for good.

This example highlights the limits of affordance-led design when it comes to videogames because there are two layers of interaction between a player and a game: between controller and avatar and between avatar and environment. The controller, such as a keyboard, a mouse, a gamepad or motion controller, has fixed affordances and its design exposes them, such as clicking or pushing. This tells the player nothing about what these generic inputs do inside a game. Hazelden puts it this way: “I think you need to teach controls with tutorials, but you are generally better off teaching mechanics with in-game examples.”

Cultist Simulator

Now the dilemma for the architect of a discoverable system is how clear to make these in-game affordances. If feedback is clear and juicy… then how much mystery is there? Cultist Simulator, Weather Factory’s recent card-based Lovecraftian horror game, is light on explanation, leaving it to the player to resolve the purpose of each card. It delivers a constant stream of delicious a-ha moments but occasionally feedback is too vague, trapping unlucky players in what appears to be a progress cul-de-sac.

Last year at Rezzed, I spent some time with work-in-progress A Light in Chorus. The experience was similar to Mirrormoon EP; I was unsure of what I was doing yet somehow kept moving forward. It came across like a bold, self-confident design, but Eliott Johnson, designer and artist for the project, explained the team were still uncertain about its eventual direction. “We’re still working it out – trying to leave enough room in the game for both affordance and discovery. But right now we are working on making the critical path rely more on affordance and sometimes more direct – but still non-verbal – instruction, than experiment.

A Light in Chorus

“It’s a decision that comes partly out of a worry about alienating our audience too much, and partly from a growing feeling that we can amplify the feeling of discovery if it occurs as an act of rebellion or ignorance of instruction. There’s a weight of extra intentionality about exploration if you’ve consciously made the choice to veer off the beaten path. I really enjoy pure exploration games, but when I think back to the moments of exploration that have given me the most delight it’s been in games where it wasn’t the biggest focus.”

But there is no silver bullet because all players are different. One person’s obvious is another person’s bafflement. “Some players did not enjoy the lack of guidance, and started to feel bored after a while,” says Zecher on Future Unfolding. “One problem we saw is that some people would solve a puzzle, but did not understand how they had solved it, so it felt random to them. This could happen with some of the puzzles that had more of a surreal dream logic to them.”

Last year, Smoke Some Frogs released Intra-System: Trust Issues a small branching narrative game that offers more questions than answers. It presents a bespoke interface which is married to the game’s underlying mystery. Despite the apparent simplicity of its branching narrative model, designer Angela Reinert explained even they had a confused player. “One of our testers told us that he actually had problems figuring out what to do in the beginning. Since it was the only report we got regarding this we considered it rather unproblematic. Nevertheless, we decided to add a small hint near the end of development: if you stay idle in the first screen for a minute or so the system will tell you to use the play button to start the first voice message.”

Intra-System: Trust Issues

Because some players are going to walk away from a game frustrated, discoverable systems can test a developer’s resolve. On A Light in Chorus, Johnson says, “The most common problem we’ve seen is players feeling so overwhelmed by new stimulus they’re unsure of what to do – another reason we’re moving towards showing goals more clearly.” However, Johnson admits that most players, once past the initial panic, persist to the end of the demo. “In the absence of information people will form all kinds of theories about what they are supposed to be doing, or what role they are playing. Pattern recognition usually wins out.”

“For Cosmic Express,” Hazelden says, “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. This is basically the rationalisation for why Stephen’s Sausage Roll is the way it is, I think.”

Stephen’s Sausage Roll

Stephen’s Sausage Roll from Stephen Lavelle is a sausage-grilling puzzle game which, at first glance, appears to follow the light touch tutorial model: mechanics are introduced through constrained level design. But something is awry in this picture. Reviewers made a conscious decision not to discuss anything beyond the basics, just like they did with Starseed Pilgrim. That’s because SSR is much closer to a game about discoverable systems; in one notorious level the player needs to utilise multiple new mechanics… without any indication that they exist other than the apparent impossibility of the puzzle.

However, context is critical. Lavelle has made a name for himself with a diverse line-up of offbeat titles, most of which eschew instruction completely. This prepares many players for what lies within… or rather, what does not. Even the simple fact of being a PC game contextualises as Ellipsis’ Salmi admits, “I think on PC you can get away with being more ‘hardcore’. The bar is lower on mobile.”

Yet context enabled the emergence of the discoverable system game. Designers may have started out with the humble goal of submerging the tutorial in gameplay but they kept pushing. At some point, the tutorial disappears completely and the player is left with mystery and exploration. A title like Stephen’s Sausage Roll demonstrates how fine a line that is.

I wondered whether any developers considered their discoverable systems to be unsuccessful. Naturally, Johnson was the most anxious as A Light in Chorus is still in development. “I’ve no doubt the AAA penchant for tutorials has been focus tested to provide a frictionless experience to the largest audience possible. It’s so easy right now to duck out of something if you feel your time is being wasted or there is the slightest bit of resistance. There are so many good games just a click away, especially if you’re using something like Steam. It’s a terrifying proposition to hold someone’s attention let alone grab it in the first place.”

Santa Ragione were happy with the results of Mirrormoon EP. “We would certainly change things now and improve many aspects of it, but the overall idea worked.” They stress, however, the difference between reducing tutorial elements as good game design and the intent to conjure mystery. “It is not about a world of games with no instructions. If exploration or similar themes are the main goal, playing around with the unguided exploration of games’ rules might make sense. Otherwise, it’d be more about making the experience enjoyable and limiting the amount of arbitrary rules a player needs to learn in order to enjoy the game.”

Alexander Martin is defiant. “I don’t like making decisions for the sake of hypothetical potential players. I don’t like making decisions for cranking up the number of potential players I’m gonna get. I think the thing I can’t abide by is sanding down a rough edge without giving proper attention to the character of that edge, to what someone might have loved about its roughness.”

Zecher is of the same mind. “Personally, I can’t stand to play many AAA games that I would potentially be interested in. I play the demo and the game loses my interest immediately in the non-skippable tutorial. I’d argue bad tutorials make a higher percentage of players stop playing than games that let you figure things out yourself.”

Future Unfolding

But his touchstone is ye olde gaming days. “We all grew up with games from the Amiga era which didn’t have tutorials like games do today. You started up a game and tried to figure it out and mastered it. For us, the magic from that time got lost a little bit. Many people who played Future Unfolding told us they find the absence of any hand-holding refreshing and relaxing.”

Three decades ago, game developers learnt not to squander scant technical resources so, in most cases, relegated instructions to little pamphlets or cassette inlays. In addition, the field of human-computer interaction was in its infancy. These nascent years were driven by the first wave of indies and most of them were winging it. Despite this, players picked up games without consulting manuals. And not just because the average player found reading less exciting than playing but because that pirate copy a friend passed you at school was a blank cassette tape sans instructions. Okay, maybe the friend had scribbled on it, “Boot without BASIC”. Cheers.

The transition from the bedroom programmer age to the console dominant era led to the videogame industry taking UI seriously, with the aim of expanding audiences and improving accessibility. They built tutorials. They took on QA teams. The days of a joystick with a single button are long gone; it would be nonsense to expect the player to guess what each of the sixteen inputs on a regular XBox controller mean for a game. So don’t expect AAA to do away with tutorials nor casual games to dispense with cute characters encouraging you to swipe here to buy gems from the in-game store for just $5.99.

While it might seem game design is coming full circle, note that unlike the days of 8-bit yore when games were accidental laboratories, today’s designers are intentionally crafting them. History tells us players can deal with this and the appetite for them has been established: last month’s Cultist Simulator has done much better than expected for developer Weather Factory and is entirely about discoverable systems.

I do have one question, though. Will this be a long-term trend or just some brief fad? Maybe this one you should go figure out for yourself.

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!

6 thoughts on “The Developers Who Won’t Hold Your Hand

  1. This reminds me of a discoverability based game I made 2 years ago called plant pit
    its got its flaws like understanding the puzzle of how you have to achieve the goal and if you had found everything to find but I think it engages the player well with the discoveries that they make playing it.

    I believe discoverability and leaving players to experiment and explore can valuable bringing engagement and respect to players over tutorials and other teaching elements if done right but if done wrong like if a small important element is not set up properly for them to understand it can be really frustrating
    for the player. I think it’s something most games should work towards but it can be quite hard to pull off well.

  2. Hi Rosden! I have to give that a go when I get home tonight. (I did try The Art of Cloning after a recent comment recommendation and I have not solved it!)

    But, yeah, pulling off effective discoverable systems is quite a feat. I’m sympathetic for Johnson’s doubt over how far to push it. A lot of people got completely stuck on Starseed Pilgrim and gave up (some reviewers too).

    I think Lottie Bevan of Cultist Simulator said recently that because they were in the Steam front page for awhile, they had an influx of players who were completely unsuited for the game causing a higher percentage of bad reviews than expected. Loss of context.

  3. I’ve sunk a few hours into Mirrormoon EP and I can’t figure out how to visit new planets. It’s very tempting to look at guides, especially since I’m generally too impatient to figure things out. However, the game is too beautiful to ruin it that way, so I’m taking a break from it until Winter (that’s the best season to play it, IMO).

    I really value games like this as long as I know the game is using this approach. There’s a huge difference between the challenge of N++ and puzzle games, and I’ve got to be in the right mood.

    When it comes to these non-handholdy games, I think what keeps me going the most is being able to play around until I figure it out, where failure is fun. Trial-and-error isn’t for everyone, but there’s something to be said about Limbo being so popular, right?

    Anyways, I’ll continue playing these types of games when I feel like it and avoid them if I don’t.

  4. Rosden

    I see what you mean about Plant Pit. I enjoyed messing around with it and liked the fact that I had more types of seeds turning up than I knew what to do with. (I’ll admit I put it down because I just couldn’t throw seeds into that box with any accuracy.)

    Hi Daniel

    I actually never finished Mirrormoon EP but that’s mainly because, well, it’s supposed to be a multiplayer collaborative game – i.e. there are way too many planets to explore as a single player and you’re supposed to work with others to find what’s out there. I won’t give you hints about how visit other planets 🙂

    I see “discoverable system” games as a small segment and it’s not desirable nor likely to become a significant proportion of all games. It’s definitely a riskier play if you hope to make money from a game but this article is really a celebration of those developers who have conviction to take that design risk: they illuminate the way for others.

    And you’re right, that’s an astute point – the bridge between the traditional title and the “discoverable system” is the embrace of failure. There are a lot of games now that expect players to fail a lot – you can probably level that one at the door of From Software – and discoverable systems are just a different kind of game where failure is expected.

  5. “Modern titles assume once they have taught you something you will never need to be taught it again.”

    This always surprises me when it happens–I thought Half-Life (1) was notorious for teaching you crouch jump in the tutorial and then you never needing it until halfway through the game. (In fact it’s been so long that now I can’t even find anything about this online, closest I found was — where the respondent literally goes “but duh, they taught you to crouch jump in the tutorial!!!”

  6. Yeah, my son is working through Half-Life and it’s interesting to observe where it’s design is not quite right. The crouch jump – now there’s a concept that never survived into another game. It often catches my son out and I have to remind him to do a crouch jump. It just isn’t that natural. I remember MYSELF playing first time getting stuck trying to jump into a vent, when the vent was the obvious exit, and only remembering after minutes of frustration that there was that crouch jump thingy.

    You can, of course, revisit Half-Life’s tutorial “Hazard Course”, but it’s pretty long and doesn’t offer the kind of quite refresher you’re probably looking for.

Comments are closed.