This is the tenth part of The Ouroboros Sequence, a series on puzzle games.
In this article, effectively the final part of a trilogy on puzzle innovation, I want to head away from well-worn genres and talk about designs which feel more fresh.
Algorithms on Rails
I’ll admit I haven’t seriously played any Zachtronics titles but they all share a common trait. They are programming metaphors, asking the player to construct an algorithm that solves a given problem. We need not play recent Opus Magnum (Zachtronics, 2017) to delve into the world of algorithm puzzles, although we might not get such delectable GIFs. Rail-based puzzles such as Trainyard (Matt Rix, 2010) and Cosmic Express (Hazelden & Davis, 2017) are also algorithm games.
In Trainyard, the player must draw a rail to connect train depots to a destination. Each train has a colour and must arrive at a correspondingly coloured station. Once you’ve drawn the rail, you start the trains and watch what happens. There are twists, of course: if trains “collide” their colours will combine and they will change colour if they travel through “painting squares”. Trainyard was not for me because I did not enjoy predicting where multiple trains would be at particular times by counting squares or running experiments. However, Trainyard enjoyed much success, being one of the early indie breakouts in the nascent years of the App Store. You know you’ve made it when someone writes up the latest “how to make a smash hit” on your game.
Cosmic Express could not be more different. It has a single train with limited space for passengers. This time, the player’s route must pick up all alien passengers, drop them at their destinations then leave the level. There are variations such as a crossroads piece and portals. Perhaps most challenging are the little green aliens that “soil” a carriage – other races will refuse to board a soiled carriage which tends to soil your perfect solution. From these humble constraints, a dizzying variety of puzzles emerge.
I thoroughly enjoyed Cosmic Express last year which focuses the player on sequence and efficient use of space.
There are a lot of Picross games, none of which I have played. Numbers down the edge of a grid indicate the number of “coloured” cells in each row and column. From this, you have to figure out the shape contained in the grid. Picross games often function like Minesweeper: only if you solve the whole picture will the game declare you a winner. You spend your time wringing out deductions from a scarcity of information. Now, Minesweeper has a little rogue in its machinery, make one mistake and it’s game over. In contrast, Picross tends to be more zen; you just keep filling in the board until you’ve got it.
Hexcells (Matthew Brown, 2013) is an interesting Picross/Minesweeper hybird. Find the blue hexes using local Minesweeper information such as the number of blue neighbours and, later in the game, aggregate Picross information such as number of hexes in a row. I was able to complete the whole set of levels in under two hours although the followup Hexcells Plus (Matthew Brown, 2014) requires a lot more chinstroking.
In contrast, I found the abstract RYB (FLEB, 2016) more invigorating. Hexcells is a binary pursuit, hexes are either on or off. RYB assigns shapes a colour: initially, they can be one of three colours (red, yellow or blue) but this later expands to six. In principle, RYB tells you the number of differently coloured neighbours each shape has, and the players job is to figure out the colours. Similarly to Hexcells, colouring a shape will reveal information about that shape’s neighbours.
But there are so many interesting twists in the RYB approach. Hexcells chases agoraphobia and enlarges the grid to conjure complexity but RYB rarely needs to explode the board in such a way because the spatial structure of each board is entirely different. There is no grid. Note this means there’s no row/column aggregate mechanic like Hexcells, although some RYB levels offer a “total count of colours” like Hexcells which can be vital for resolving the last few shapes.
Some Hexcells levels tend to offer multiple routes to success but the tighteness of an RYB level can feel linear in contrast – as if you’re sweeping out the one true path the developer has laid open for you. This feeling is heightened when you discover RYB is very careful what new information a coloured shape reveals; sometimes it will not tell you about all of its neighbours.
Perhaps the cleverest feature is how RYB goes next level with the additional three colours, green, orange and purple. On early levels, if RYB says you have two red neighbours, it does what it says on the tin – yes, the shape has two red neighbours. On later ones, this can mean colours which contain red: red, orange or purple. I had to laugh at that reveal because it just seemed to multiply the complexity. Yet what seems impossible actually isn’t. RYB is always solvable.
However, as discussed in my recent stream that covered RYB it shares a fundamental problem with Hexcells. If you make a mistake in Minesweeper, you’re dead. If you make a mistake in Picross, the game won’t declare you a winner until you’ve fixed the picture. If you make a mistake in Hexcells or RYB, it’ll buzz you. That’s it. Hexcells counts the mistakes. RYB gives you three lives then resets the puzzle. But that mistake reveals information which can spoil the puzzle. I’m sure some players will fall back on this to get through a challenging puzzle. Although if you’re like me, I ignore mistakes and try to figure out the correct deduction method for the next step although “correct mistakes” – suppose you clicked the wrong area – become permanent which is maddening; this sort of revelation ruins solution discovery.
I’m not sure how you’d fix this as it seems an unavoidable consequence of a design that reveals information as you proceed… unless you go rogue and do procgen levels with permadeath. You know, like Minesweeper.
I’ve tried but I can’t quite think of another game Tricone Lab (Partickhill Games, 2017) reminds me of. A labour of love built over several years by its developer, Tricone Lab doesn’t quite have the looks or sounds to pull in a crowd, but it’s got a lot under the hood.
In each level you must assemble “tricone” molecules from three, uh, cones. These components must be in the same space to combine them using a tricone catalyst. If the cones and catalyst lie in different regions, you must either break open those regions or transport the pieces across regions. And the pieces needed to do that may be in different regions themselves or disassembled.
Now be warned, the progression is very gentle and a thick wad of easy Tricone Lab levels will lead you to ponder whether there is any challenge here. Indeed there is but only once the game has thrown its kitchen sink at you. Some levels have anticatalysts that steal resources. Some have charged regions that inhibit assembly… but can also be used to your advantage.
There are definitely criticisms to be levelled at Tricone Lab, the primary one being readability of the UI. It can sometimes get confusing remembering what different components do – and they are a lot of them – and building transports for transport components can be confusing. Nonetheless, I don’t think there’s anything out there quite like it. Tricone Lab packs in a lot and only occasionally treads in agoraphobic territory.
I covered Tricone Lab in this stream last year:
Now The Witness is a funny one. Where does the puzzle begin and the puzzle end?
At its most basic, The Witness is about panels containing line puzzles. You must draw a line from a starting point to an ending point and the line cannot cross itself. Each puzzle, however, features constraints that your solution must obey. The first real constraint you’ll encounter are the black and white blobs that need to be kept separate. But wander through the village and you’ll be shown all sorts of strange symbols that appear incomprehensible.
The panel puzzles themselves are very good and there are two varieties of panels – ones which follow strict logic and ones which… well, let’s just say, you need to be observant.
There’s also another layer in the game typically referred to as the [REDACTED] “puzzles” but while discovering them for the first time is one of those My God It’s Full of Stars moments, most of them are not really puzzles and do not require any logical thinking. Initially, they’re fun but eventually collapse into frustrating collectibles. I suspect the frustrating part is deliebrate.
And the game is more than the sum of its parts. There are questions about what it all means and it has little secrets that go beyond the simple “solve the panels” and [REDACTED]. But much of the game is about contemplation. My interpretation can be found on YouTube where I spoil the game to shit and explain why it’s called The Witness in the first place:
The Witness is one of the few puzzle games that succesfully embed puzzles in a fully realised environment; I am a big fan of the game’s semi-openness. Portal is another (and it led to many games copying the “all the puzzles are actually tests” template) and The Talos Principle also makes a go of it, although the Talos environments have far less to say than Portal or The Witness.
I haven’t finished Antichamber (Alexander Bruce, 2013) yet but it seems like it was expensive to make. Unlike everything else listed on this page, it isn’t a game driven by a puzzle template, although there are a lot of “block manipulation” puzzles later. Some puzzles play with spatial connectivity but no two puzzles are really alike. A lot of Antichamber is really about confounding expectations about how things are supposed to work. It’s the kind of game which tells you “here’s the exit, aha not really”.
I’m on the verge of arguing it isn’t quite a puzzle game as it doesn’t feel like solving logical puzzles but seeking out what logic cannot answer. Still, I’ve generally enjoyed it so far although the little life philosophy posters on the wall associated with each puzzle are a bit too much like those motivational posters that big corporations love.
Dis Pontibus (2017) is a small procedurally generated puzzle game by Electron Dance reader Marcos Donnantuoni which is remarkably simple. But I am genuinely rubbish at it.
The world is made of tiny islands and to make progress to other islands, you need to build a bridges using the pieces available at each crossing point – rotators, sliders and snakes. For example, a rotator could be turned to face the opposite island and then a slider piece might be moved up to bridge the final gap. The problem is that you can never cause these pieces to disconnect from the original island. As soon as there’s clear blue water between your piece and any others connected to your original island – down they go into the depths.
Now because your range of moves are limited and there are heavy contraints on what moves are safe, most solutions involve a much longer sequence of moves than you anticipate and sometimes require moves that seem uninituitive at first. There are no agoraphobic issues here – the puzzles are always small but generally maddening. Kind of game you’re pleased to discover is kitted out with undo and restart. I can’t think of anywhere else I’ve seen this type of puzzle but let me know if it reminds you of anything.
Also covered in a stream this year:
Don’t fall for those pretenders to the throne such as 1024 or 2048. You want the real deal because this thing has been polished within an inch of its life.
I tend to think of Threes (Sirvo, 2014) as a survival roguelike. You are the manager of a 4×4 board in which one card appears each turn. A turn is a “slide” of the board which is difficult to explain unless you see it. All the cards slide in the direction you swipe or push and if they come into contact with a card of the same number they will merge and double instead – the only exception to this is 1 and 2 which will only merge with one another. If the board ever fills up, you’re dead.
I think of Threes as a roguelike because although it has an endpoint I doubt I will ever reach it. I played Threes hard for a while which is as good a recommendation as any. Wonderful game, requires a lot of attention from the start if you intend to go the distance. Unlike 1024 or 2048 which you can just idle through in a zen haze for an hour.
There’s a free version you can play online.
In Twofold Inc (grapefrukt, 2016), each level sets you up with a set of coloured recipes and you must satisfy them by tracing out continuous paths is each colour on the grid. Each claimed square space doubles the line’s score: 2, 4, 8 and so on. To score 128 for a colour, you’d need to trace out a line 7 spaces long. You can also shift rows and columns to create longer chains.
So what’s the puzzle aspect? Well, each recipe has a turn counter within which it must be fulfilled and after that you start losing lives. Like Threes, Twofold gets hard fast. It has a number of subtleties which I liked although I used to lament on Twitter that the gold never turned up enough.
I felt much more at the whim of randomness than I liked and eventually gave up out of frustration. Although, while researching for this post, I discovered there was an extra move I could use that the game didn’t explain. OMGZ.
I love this game so much and I wrote about it for Rock Paper Shotgun. The point of Recursed (Portponky, 2016) is to find the exit in each level and escape. That’s it. To get there, you might need keys to unlock doors, blocks to climb on and… well, chests.
Like Tricone Lab, Recursed takes its sweet time getting around to challenging the player. There was clearly some anxiety that players wouldn’t get the basic idea. The idea that jumping into a chest, can take you to another room and you can carry these chests about. Now this might sound very Portal-ish. You have a set of “teleports” to different rooms which you can move around. But that would completely miss the point. Oh my God, you have no idea.
There are two brainfucking things that change everything. Leaving a chest means the room within gets reset, so you could keep taking the same item out of a chest’s interior again and again. You could make infinite copies. The other brainfucking thing is you can also do this to chests themselves. So you can have multiple copies of the same room. Oh yes, you can most certainly carry a copy of a chest inside itself. It gets even worse when you meet green “static” items that persist across all copies of the same room and survive resets.
There are some levels in there which are literally one or two rooms and will blow your mind trying to find the solution. Beautiful stuff. Some of them do edge on agoraphobia simply because it’s not fun juggling too many copies in your head. But there are also paradoxes, jars and the extra stuff in the free expansions – The Oobleck Conundrum and The Last Tapestry.
It is not perfect – but by God, it’s clever. I also covered it in the same stream as Tricone Lab above.
Whoa, stop the press, we’ve got a late entrant. Free game Illiteracy (Le Slo, 2018) looks something like this…
No instructions. No title screen. Which means we’re definitely in the “discoverable systems” space. I guess it took me something like half an hour or more to solve? I enjoyed the gentle prodding and poking and figuring out how things work although once you’re through that stage it decomposes into a puzzle type I usually hate – it hasn’t been mentioned in Ouroboros so far so no clues there – but happy to get into it in the comments if you’re interested.
Is this really fresh? If anything, it feels like the kind of puzzle you might find in a Mystlike. Before you, a stone wall covered in weird squiggles and you have to find some combination to open the hidden doorway. But your experimentation in Illiteracy must be very precise and that’s what makes this a fun half hour.
I’m pretty sure I haven’t played every puzzle in the world so if there are any other cool puzzleworks which aren’t sokobanlikes, match-3, laser reflection and actually involve puzzle solving rather than hotspot scanning, please drop us a note in the comments.