Learning From Failure

Learning From Failure
The Game That Ate the Earth

Jonas Kyratzes | 21 Dec 2010 12:35
Learning From Failure - RSS 2.0

The beginning was glorious. The levels were huge, giving a very real sense of the vast spaces of the ruined Earth. The graphics were exactly what I wanted, poised between the bizarre and the just barely recognizable, flashes of familiarity in an alien environment. I could see the world grow right in front of me, like a living being ... or a whirlpool.


Already I was making mistakes. They didn't feel like mistakes at the time, and it's hard to call them that now, but somewhere deep down I knew I wasn't being careful enough. Making levels was so much fun, and the results were so hypnotically atmospheric, that I was neglecting to think about the overall design. The first version had a detailed research system, different types of atmospheres, radiation levels, and many other features. Some of these I'd already mentally deleted, especially those that hadn't worked very well in the original. But what about the rest? Part of me assumed I could just carry them forward unchanged.

I knew that I wanted the player to gain various abilities that would allow them to explore further, yet I kept pushing the implementation of those abilities further away, concentrating on building the world. But how can you build the world if you don't know exactly what abilities the player will have? Other questions were popping up and being ignored: I'd carried forward the notion of collecting four types of resources, but how much of each resource would be needed for what? Oh, I'd figure that out later.


Phenomenon 32 kept spreading through my life. I wrote more dialogue, got together more actors. I worked very hard on a beautiful and chilling intro that used various voices, from fictional characters to Carl Sagan to Malcolm X, to establish that this was a story about humanity. I worked and I worked and I worked, and the game was nowhere close to being finished. But the work was some of the best I'd done, layered and complex and powerful, and I knew that this game signified something special for me.

Hypnotized by the world I was building and caught in the belief that this game was it, this one would be my breakthrough, I ignored another opening whirlpool.

Construct had a memory leak. It was big, rapidly approaching humongous. And now the crashes began. But it was just the development environment, not the game, so everything was all right. Right? Right.

How many times did I post to my website that the game was almost done? Yes, it would be done next week. In two weeks. Next month. I believed it, too. Then it crashed again. By the end, it was crashing after every four actions I took.

At this point, a rational person would have stopped. I was running out of time, money and endurance. But I was lost in the ruins of Earth. Not finishing the game now seemed like a death sentence. It had to be done. It had to be finished.

I was going mad. "This can't be happening!" I would scream as I hit another bug. "This isn't programming, it's magic! How can the same code work here and not work there?" Sometimes I'd break into tears, other times I'd start laughing maniacally. About 30% of the time I was simply being an idiot. The rest of the time, much more frustratingly, I'd managed to find another exotic bug in Construct.

Comments on