The title of this post is the subtitle of an article in Game Developer (March 2010) by Matthew S. Burns about the production methods used by various leading game developers. (I have no idea why I’ve started receiving copies of this magazine for software engineers in the video game industry, which I’m enjoying despite — because — it’s over my head.) According to the article, Valve — the source of some of the greatest games ever, including Half-life, Portal, and Left4Dead — “works in a cooperative, adaptable way that is difficult to explain to people who are used to the top-down, hierarchical management at most other large game developers.” Valve trusts its employees to make good decisions, but it is not a free-for-all. Decisions are made in consultation with others (“relevant parties”) because, as Erik Johnson says, “…we know you’re going to be wrong more often than if you made decisions together.” In addition, what Matthew calls “a kind of decision market” develops because people who design a system also build it, so you “‘vote’ by spending time on the features most important” to you. Vote with your code.

Valve also believes in making incremental decisions. Week by week. But what does that do to long-term planning? Robin Walker says that one of the ways she (he?) judges how far they are from shipping by “how may pages of notes I’m taking from each session.” That means Valve “can’t plan more than three months out,” but planning out further than that increases the chances of being wrong.

Interesting approach. Interesting article. Great games.