make-believe.org

Vale Digitalo

I just found out that my old employers, Digitalo Studios, shut up shop forever in November. That’s really quite sad. Digitalo was a vibrant, independent, and exciting game developer with a real emphasis on exploring the less-worn paths of gaming experiences. They gave me a task and attached few restrictions on my execution of it, and consequently my term working on the artificial intelligence for Devastation was some of the most creatively rewarding time I’ve spent pushing buttons on keyboards. Devastation didn’t succeed, but not for the reasons games usually don’t succeed — it was a plethora rather than a paucity of ideas that kept it from capturing the gaming public’s imagination.

Unfortunately I couldn’t work on the game to its completion, and as a result the artificial intelligence did not get any further along the trajectory I had envisaged for it, because time and cost constraints were ultimately prohibitive. The AI was probably about 50% complete when I left Digitalo (because they ran out of money for a while, and I wanted to go back to school), and when the game was reworked as a single-player experience, no-one else was really in a position to retrain and complete my bots.

Of course, that really hurt the game in reviews. The easiest way to attack a game in a review is to target the AI, because firstly few if any users really understand what it does, and there is a standard (namely, the human one) which no AI designed for open and endlessly flexible gameplay can hope to match. Now you’re not supposed to know how the AI works; that is precisely whence its magic derives. But like people, AI needs (and generally deserves) some time before it is judged. Like people it seems at first one-dimensional, but with familiarity gains more dimensionality. Some of the best AI I have ever seen in computer games has been mauled by ignorant reviewers who have not donated the time or attention to perceive its complexity on the level that most dedicated gamers do. And because they don’t donate the time or attention, they apply hypocritical standards. Human players win because their strategies are not momentary, they are long-term, and they are rarely wholly logical; certainly infrequently consistent anyway. Bots are held to those human standards of variability and success, but are also held to the robotic standards of logicality, consistency, and above all, short term planning. Bots are asked to be comprehensible in a way that humans never are, which leads to a no-win situation (oh that’s a bad pun). Furthermore, people confuse unsuccessful AI, or rather not-always-successful AI, with bad AI; which seems reasonable until you play against always-successful AI, which is usually “cheating” AI, and realise that there is no fun in that. That “happy medium” of bots that are both good at their job and fun to play against because they are fallible (and not just because of a single exploitable flaw, which is an outdated way of approaching the problem), is incredibly hard to attain.

Having said that though, the AI in Devastation was hopelessly unfinished — at times it slowed the game down to a crawl, which is just inexcusable (and largely my fault) — and in the end the bots weren’t tailored to the primary gametype. They just didn’t know how to play the finished game. That’s a damn shame, because their architecture was a fascinating one, and in terms of basic concepts, one I suspect will increasingly become the trend in game AI. (Remind me at some stage to go into that, okay?)

It is clear playing Devastation that it is in some crucial way unrealised, and that is bittersweet: both inspiring, and frustrating. One day I hope it finds an afterlife in an abode like the Home of the Underdogs, where it might receive its proper dues.

As well as developing Devastation in entirity, Digitalo Studios also provided creative direction to the PC game interpretation of Harry Potter and the Sorceror’s Stone. There was also a kooky little walkthru of the Notre Dame cathedral before my time.

Digitalo pushed the boundaries of game development by collaborating entirely on the internet. I worked on the game without leaving Melbourne; the lead programmer Bob was in Japan and the other main programmer Ray hammered away at the user interface from his pad in New Mexico. Creative direction was provided by Vic in Florida. Artists and designers were scattered across the continents (though we didn’t have anyone in Antartica). I never actually met any of my colleagues in person, in fact. That we produced Devastation in such circumstances, across in particular such disparate timezones, is I think an impressive feat. You have to judge the game on its merits, of course, and not the circumstances of its creation, but those circumstances are a reason I’m still very proud of it.

Bob, Vic, Barry, Ben, Mike and co, I hope you stay in the industry and stay together as a team somehow. Good luck for the future, which with your collective talents and collaborative dynamic is still surely bright.

Joseph | 26 Dec 2003

Sorry, comments are not available on this post.

stuff & nonsense

  • Topographic viewTopographic view
     shows elements on a webpage according to how deeply nested they are. It's a bookmarklet for web development.
  • The qualifierThe qualifier
     renders controversial statements on this page harmless. Reinstate the slings and barbs by refreshing. Also a bookmarklet.

  • jjmap
    American Diary

    Two weeks with the apple and the lone star (illustrated).

all posts, ordered by month in reverse-chronological order:

In Words

In Other Words