Wot I Think: Quadrilateral Cowboy

I’ve lined up my suitcase rifle, sawed open my escape route, and written two lines of code to control a small robot when I blink. Blink once and a set of lasers turn off allowing me to enter through a space station maintenance tunnel without triggering any alarms; blink twice and a second set turn off, allowing me to exit cleanly. It’s only a few moments later, as I stand in the vacuum of space, that I realise I left my deck – the computer by which I write scripts to control suitcase rifles, small robots, lasers and more – back in the maintenance corridor.

Quadrilateral Cowboy is a game about breaking into buildings to raid vaults, steal safes and hack coma patients. It’s a stylish, retro-futurist love letter to computing, engineering and ’90s videogame level design. It also feels like the prelude to a better game.

It starts well. Technically, you are not carrying out heists, but planning them. You and your two partners-in-crime have built detailed 3D worlds of your target buildings, which you then enter to complete your task virtually as practice for the eventual mission, which happens off-screen. This justifies a great many things, including simple greybox level design that removes the visual clutter between you and your goal; patch notes floating in the air containing subtle hints among the changelogs; and the ability to press F1 at any point to switch into ‘noclip’ mode in order to explore a level fully before you set your plan in motion.

The main affordance of this fictional trick is your deck, the tool by which you interact with the world. Set down this computer on any surface and you can then control its simple DOS-style interface with your keyboard, typing short commands like “door2.open(3)” to open door 2 for 3 seconds. Eventually you start stringing together those short commands like “jump; datajack; datajack 0” in order to, all at once, tell a small robotic dog to jump forward, plug itself into a ‘datajack’ socket, and perform that datajack’s first action – which is, just as often, opening a door for three seconds.

Don’t fret if coding isn’t your thing. The commands are never more complicated than those above and the game introduces you to new commands at a leisurely pace. In fact, almost every level of the game is designed to introduce you to something new.

After becoming familiar with using the deck to directly interact with the environment, you’re introduced to that small robotic dog I keep mentioning. Her size can access areas you can’t, so you can direct her using your deck to control machinery from greater distances. After becoming familiar with the dog for a mission or two, she’s forgotten. You’re introduced to the suitcase rifle, which you aim and fire with the deck. Once you feel comfortable with it, using it to fire bullets into buttons not people, its usefulness diminishes. Instead, here comes the Launcher, which is like a Quake 3 jump-pad where you can control the pitch of where it tosses you.

Next you’re given the ability to enter heists as different people with different abilities, and to put those abilities to use simultaneously. You’ll enter a level as one person, place a launcher, hop across a large gap, and use a tool to jack open a door for 60 seconds; then you’ll hop out and re-enter the level from the beginning as the greaser, who can jump high and squeeze through small spaces. You steer them to follow the ghost of your first character, enter that now open door and clear the next set of obstacles. This feels like the game’s ultimate form, where your complete toolset allows for the clockwork precision of an Ocean’s Eleven caper.

Which makes it strange that the game then abandons even this concept after two missions, piling all those abilities into a single character. After two further heists that make use of all your tools simultaneously and which, finally, seem to allow for some creativity, the game is over.

It’s not that the game feels too short. It’s that it feels like a succession of tutorials leading up to a game that then lasts for just a few missions. None of the tools you’re introduced to feel as if they’re used to their full potential. You’re given a chance to get used to them and then the game moves on. This may be because those tools don’t have a great deal of potential: whether writing scripts on your deck, steering your robot towards datajacks, or deploying doorjaws, powersaws, or your rifle, all you’re ever doing is opening a door.

Unlike other reach-the-door puzzlers such as Portal, there’s no moment where the level design combines with your earned understanding of the tools to enable thrilling ingenuity; unlike other similar heist-’em-ups such as Gunpoint, there’s no comic consequences to experimentation. My experience was of muddling through the game’s four-hour length and never being wholly satisfied by my solutions, but while others may find it compelling to polish performances in order to beat friend’s completion times, I don’t find that motivating enough to want to play missions a second time.

Yet despite my criticisms, I do think it’s worth muddling through those missions at least once, simply to bask in its style. Quadrilateral Cowboy is created by Blendo Games, whose previous games include Gravity Bone and Thirty Flights of Loving. Many of those games’ flourishes return here, from the full-screen title cards introducing each mission, to between-heist vignettes which wordlessly advance the story via smash cuts and rooftop badminton matches.

There’s also an affection for old technology that I find wholly charming. Where background details might be drawn with simple, flat-textured blocks, machinery is often rendered with multiple moving parts. Your first task at your headquarters is to build the computer with which you’re going to simulate each heist, individually slotting in first the CPU, then the heatsink, then the memory and so on. At any point during each mission you can put some music on by popping a vinyl record in your portable player. You hack those coma patient’s brains with a device that sounds like an old dial-up modem. The books at your base have titles which each convey parts of the compile process for an old-fashioned first-person shooter level, from BSP to VIS to RAD.

Few games are ever truly made by a single person, and there are other names in the credits, but Quadrilateral Cowboy feels like a singular vision. It’s what happens when someone is allowed to develop a personal voice, from Quake 2 mods through to (Doom 3 engine powered!) commercial releases. It is delightful enough that I can forgive the hollowness of its core challenge.

Less easy to forgive are its bugs. I experienced six or seven crashes to desktop while playing, and multiple occasions during which my view, controls or deck became stuck and required a mission restart. I rarely lost much more than a few minutes progress, but when that progress involved fiddly setup and scripting, that was a few minutes too many.

In 2011, when describing why Introversion had abandoned Subversion, their long-in-development heist game, designer Chris Delay described a problem with the genre: “In the end, after all that development and years of work, you still completed the bank heist by walking up to the first door, cracking it with a pin cracker tool, then walking into the vault and stealing the money. There was no other way to complete that level. And this would be the essential method by which you would complete every level after that.”

It feels as if Quadrilateral Cowboy never finds a solution to this problem, but it moves through different ideas quickly enough, and does enough with its cool, colourful world and story of silent friendship, that I enjoyed my time with it.

Quadrilateral Cowboy is out now on Windows via Steam and Humble for £15/$20/€20.


  1. Hideous says:

    I completed the game last night as well, and I was getting pretty excited about the missions. Finally, they were becoming challenging!

    … And then suddenly, the ending appeared.

    I do really enjoy the gameplay in this, so I’m hoping the mod workshop can fill the void. Maybe we’ll get some fun heists out of that?

    • Eddy9000 says:

      I thought this about Giants: Citizen Kabuto. You spend the first levels jetting about on your own, get introduced to these really neat strategic combat tools, use them for one level and then that act of the game ends. Giants was essentially a number of different genre mini games rolled together but it was disappointing none of them lasted that long.

  2. TΛPETRVE says:

    Typical Blendo problem, I guess. Pretty much all of their games, with the notable exception of Atom Zombie Smasher, feel like unfinished proofs of concept. Charming, entertaining, but disappointingly hollow in terms of content. Not to mention, criminally overpriced for anyone who doesn’t consider themselves an enthusiast.

    • HothMonster says:

      Meh, Gravity Bone was free and 30 Flights is $5 which seems a pretty fair price for an hour of fun.

      I’m a big fan of the minimalist story and aesthetics they do so well though. So it is rather up my alley. I was really hoping this one would be a bit more of a full experience though.

    • Jeroen D Stout says:

      Flotilla, Gravity Bone and 30 Flights of Loving all seemed really finished to me. You have to appreciate a sort of strange understated style, but they didn’t strik me as unfinished

  3. plsgodontvisitheforums_ says:

    This is why we can’t have nice things.

  4. golem09 says:

    Played this last night for a while, then pulled the plug and got a steam refund. I like coding, and I thought this was a game about coding. Instead coding was just used as glorified lever puzzles. Story was absent as well, leaving me with not much to enjoy. I bought TIS-100 after that instead.

    • preshrunk_cyberpunk says:

      TIS-100 is great.

      Have you played Screeps?

      • MortyDice says:

        Or Human Resource Machine?

        • preshrunk_cyberpunk says:

          Human Resource Machine’s art theme (and Little Inferno)give me the creeps.

          But I suppose its intentional.

  5. FroshKiller says:

    In the end, after all that development and years of work, you still completed the bank heist by walking up to the first door, cracking it with a pin cracker tool, then walking into the vault and stealing the money. There was no other way to complete that level. And this would be the essential method by which you would complete every level after that.

    Allow me to solve this problem: The heist starts after the safe has been opened. The contents are surprisingly different from what you expected. You have to quickly create and execute a plan to make off with the loot using the materials you brought for what you expected to get and what’s available on site.

    • Gravy100 says:

      Yeah, this sounds like it a route you could go, it worked for Thief right? Why not this style but with hacking? Happy to hear why this is a silly idea since I haven’t played QC.

    • horsemedic says:

      Invisible Inc. does this—but it uses the evac to complement the infiltration phase, rather than replace it. You choose how far to press into each level, gaining more loot the more rooms you crack, but also increasing the number of guards you’ll have to deal with when you break for the exit.

  6. Turkey says:

    What a shame. I watched a Quick Look earlier, and it looked super charming.

  7. Wisq says:

    I adore Introversion Games, and have bought most of their stuff twice over, and I completely understand if they’re not really “feeling” a game concept and need to cancel it or put it on hold.

    That said, I do think their “problem with the genre” is way overblown. There have been plenty of good heist games — Monaco, The Masterplan (mixed reviews but I liked it), The Marvelous Miss Take, Crookz, and now QuadCow, each with their own take on how heists work. Even Satellite Reign basically qualifies, even if it’s more varied than just stealthy break-ins.

    The difference is, all of those relied on hand-crafted levels rather than procedurally-generated stuff, which is how I understand they wanted to do Subversion. It’s a lot harder to strike a “this mission is hard and exciting, but possible” balance when it’s a random algorithm generating it — typically, you either put too few rules in and get impossible levels, or too many rules in and get predictable ones.

    If that was indeed their problem, I wonder if they couldn’t learn something from QuadCow’s “casing” approach. Let the player do a virtual playthrough first — no risk, no reward, but let them probe the building’s security and figure out breaching methods. Let them noclip, use any agent’s abilities, etc. Finally, have a “real” run, where rules apply, mistakes matter, and results count.

    Another option, independently or on top of the above, would be to worry less about whether it’s truly possible to complete a breach without detection, or maybe even to complete it at all. Leave that decision up to the player, like in real life. Does it look doable? Is there a silent route, or will you have to set off some alarms and work around them, or finish the mission under time pressure? Maybe toss in a dash of Satellite Reign / Incognita / Payday 2 — if/when security shows up, do we have the resources to deal with them? Etc.

    All of which might be beyond the resources of a couple of self-described bedroom programmers — but not, I think, indication of any real flaw in the genre. It just requires some thinking outside the box, like some of the heists themselves.

    • Xocrates says:

      The impression I got watching various Subversion videos and presentations, was that the game was massively over-engineered, to the point where you had a lot of really neat and complex toys, but the game itself didn’t properly support them because more straightforward ways were available.

      The “problem with the genre” isn’t a problem at all. Merely a design glitch to be fixed.

  8. ROMhack2 says:

    I’ve been playing it a lot yesterday/today and I agree with Graham on this one – it feels a lot like a prologue.

    I’m not disappointed with it because the aesthetics and world-building are superb but I did expect the developer to build on the narrative strength shown in his 30 Flights ‘o’ Loving and Grav Bone.

    Minimal but charming is the best way to sum it up, I guess. The mechanics are decent too, which should be mentioned because I imagine they could have easily been messed up.

    I’ve seen people on Steam mention the film-influences. I’m not seeing that as much with this one but I guess the characters sorta remind me a bit of the precocious people found in a Wes Anderson movie. Maybe even something like the main character in Kumiko: Treasure Hunter.

  9. Hyena Grin says:

    I believe I’ve played most of Blendo’s games up to now. Atom Zombie Smasher, 30 Flights of Lovin, Gravity Bone, Flotilla.

    They are all extremely pristine visions of a game with all of the fat removed. There’s no padding – like at all – and I feel like this is almost part of their mission statement. They present a concise, well-executed idea, and that is it.

    I sincerely admire this about them. You get a great experience within the confines (game mechanics) that the game establishes, and more often than not, that leaves me wanting for more to do with those mechanics. Which is both good, and bad, in the sense that you have to ask yourself sometimes whether the game feels complete, or content-deprived.

    But then I have to remind myself that they are selling their games for a fraction of the price of bigger, more content-laden titles. And I have to accept that, probably, for the price, I got more than what I paid for in return.

    And then I wonder if the developers would be better or worse off (financially) if they spent more time making a game that cost more.

    Hard to say.

    • ROMhack2 says:

      Good point about purity. That’s pretty much Blendo Games’ ethos.

  10. GWOP says:

    A game that introduces a mechanic, uses it for one level, and then forgets about it?

    Quadrilateral Cowboy is the Call of Duty of indie games.