In Defense of Giving a Shit

An IGF judge argues that the festival needs an explicit policy on generative AI. Its current approach lets each judge decide whether to consider games using AI, but entrants may not know that roughly a third of the judging pool will abstain. The debate intensified after Big Walk disclosed that LLMs were used to diagnose and fix errors in less than 0.1% of its code late in development. The author rejects the claim that rules are pointless because they are hard to enforce: policies communicate what a community values, give members grounds to respond to violations, and raise the cost of concealing them. Whatever the policy’s position on AI, leaving it undefined risks alienating judges and discouraging submissions.

For this year, the author says, the IGF must apply its existing practice of individual judgment to Big Walk’s award eligibility. The game’s quality and House House’s reputation should not determine whether its AI use matters; disqualification, if warranted, would be a limited consequence, not a public punishment. The author also connects even small uses of AI coding tools to pressure on game-industry workers, including freelancers seeking scarce contracts, and questions why a successful studio would not hire a specialist to resolve the issue. Defenders who distinguish AI-generated code from other AI use, the author argues, overlook how such choices contribute to broader labor concerns. House House’s delayed public explanation also makes the disclosure harder to dismiss, since the studio appears to have understood the controversy’s significance.