Quadrilateral Birthday

Quadrilateral Cowboy turns 2 years old today! Happy birthday to this video game.

To celebrate, I thought I’d do a little writeup about its early development.


The first thing I did was get the deck up and running.

It worked! It felt great to type on a little portable computer. A lot of the project depended on this interface, so I was glad it worked well.

Behind the deck you can see some doorways and railings. Those are look-and-feel tests – I think size metrics are important for good level design, so I try to make these decisions as early as I can.


I wanted Quadrilateral Cowboy to focus on low-tech analog equipment. So originally, the intention was for the portable deck computer to always have to be plugged into a wall.

Basically, this meant cable management. Here is a 2012 video of an early attempt at simulating cables:

As you can see, the cable behaves like a wild animal. Near the end of the video, the cable rudely slams me back into a wall.

Quadrilateral Cowboy’s game engine, idTech4 (the Doom 3 engine), has a physics component, but this was my first time digging deep into its guts. I wasn’t having much luck but I felt the payoff would be worth it. Here was another attempt:

The cable was, yet again, an unruly beast.

I really wanted to get cables working. So, I kept at it, this time trying out Doom 3’s chains system…

…which I was unable to wrangle into something useable. Alas! The cables were not meant to be.

Deciding whether to cut your losses and move on is often a murky and wiggly choice. I personally feel low-tech solutions can get you very far – so if I’m having a lot of trouble with a complicated bit, I document it and then try a simpler solution.

In this case, the simpler solution was to have devices be wireless and… not use cables. And the game worked just fine!

However, months (years?) later, I… got cables working. Here is the piece of equipment called the Launcher:

Would Quadrilateral Cowboy be better if the cables had worked earlier? I don’t know. And honestly, I kind of don’t care? I think projects need room to breathe and grow and mess up and find their own path. And this was the path this project chose.


Look at this adorable biped.

One of my design goals was to not have any hostile AI characters in Quadrilateral Cowboy. There’s something compelling about you versus a building, and I wanted to explore that.

The Elk was intended to be a robot that you hijack and control. The idea was that it had some kind of security clearance, so you’d use it to infiltrate a secured area.

For various reasons, it slowly morphed into the tiny bite-sized Weevil: It was more interesting to sneak the Weevil into ventilation shafts and narrow cracks, whereas the Elk was more or less a full-sized human. So, goodbye Elk, hello Weevil.


For some time, I explored the idea of text-to-speech. The idea was to give players another tool in their toolset and give the world some extra character. But mostly, robotic text-to-speech voices always make me laugh, so I wanted to include it.

Here is the test run, in which I make the text-to-speech say “baboo” many times:

There ended up being a framerate hitch whenever the text-to-speech was called, and I couldn’t figure out how to resolve it. So, this was scrapped.


I really like the analog clock on the older version of the map:

But, the dang thing took up so much room and covered up so much of the screen. So, it was later converted into a more compact digital clock. Alas!


Here is a run of a very early test area:

The pieces felt good, and more importantly, it felt like there was potential. And so, Quadrilateral Cowboy grew from there.

Happy birthday, Quadrilateral Cowboy. Thanks for reading!