Preliminary results for the AI coding poll

Hi all! I've been running an unofficial poll on the Ludum Dare rules. (If you haven't filled it out, the poll is still open here). Here's an update on the responses so far:

The main part: Run-off voting

I set up the poll using what Wikipedia calls Instant Run-off Voting, because it allows us to pick between these four options more fairly compared to other ways of voting.

The way it works is that everyone orders the choices from their first choice to last choice. We start with counting the votes based on everyone's first choice. If your first choice is not popular, instead of your vote ending there, you get to vote for your second preference as a fallback. The process repeats until one option has more than half the votes.

Here's the simulation of how the votes were divided up, with votes colored based on the first preference:

liked.gif

After handling second & third preferences, the most popular choice is ❌❌ No AI coding allowed for Jam or Compo. Though it's pretty split, and there's still a significant group that would prefer it be available in some form.

Also interesting that people are so split when you only consider the first preference.

What option might stop people from being involved?

One argument in favor or against a choice is that it might turn away people. So I asked two questions on that:

If all AI coding tools were explicitly allowed, would you continue to participate in Ludum Dare?

Screenshot 2026-10-12 at 20.28.19.png

  • 1 = I would stop participating completely
  • 3 = Neutral, I would participate at the same amount that I currently do
  • 5 = I would be more likely to participate

If all AI coding tools were explicitly banned, would you continue to participate in Ludum Dare?

Screenshot 2026-10-12 at 20.28.28.png

  • 1 = I would stop participating completely
  • 3 = Neutral, I would participate at the same amount that I currently do
  • 5 = I would be more likely to participate

Allowing AI has the biggest impact here with 46% of people saying they'd be less likely to do Ludum Dare. That's a lot! Disallowing AI would discourage a smaller group, 15%. I'd personally like to not exclude that 15%, but if it's a strict trade-off between these two extremes, banning AI seems the better choice based on this data.

Where people seem to agree:

Most people seemed to agree on these questions:

Should you have to disclose AI tools?

Screenshot 2026-10-12 at 20.29.06.png

This seems to be basically unanimous to me.

Should AI autocomplete be allowed?

Screenshot 2026-10-12 at 20.29.12.png

This surprised me actually given the way the other questions went, but many people are ok with AI autocomplete.

Free form responses

There's a range of responses in the free-form section. Honestly there's a lot of meaningful stuff there and I don't think I can really do justice in summarizing it but here's some that give the range of responses:

  • AI, to me, is against the spirit of a game jam [...]
  • Don't overthink, allow AI, slop games will get bad ratings [...]
  • I dont care if its made with AI [...] for me the end product is important.
  • I deal with this shit enough day to day, I finally want to feel like I am doing something for fun.

Poll weaknesses

Because the site has been down I've only been able to share it with people that are more involved with Ludum Dare in general, e.g. people that hang around on Discord or people that were on the site months before the event. So it's going to be a biased sampling in that regard.

Raw Data

The current data is available as a spreadsheet if you want to dig into things and analyse things more. (Unfortunately Google forms doesn't let me share the summary, as far as I can tell).

What to do with this?

Well... right now with Mike in charge of Ludum Dare, the best we can do is advocate that some type of rule change seems to be strongly needed for the sake of Ludum Dare. At a bare minimum, disclosing AI usage seems like a good rule to add, and I feel like disallowing AI for compo seems like a pretty uncontroversial choice (even though that's not the most popular option, seems like a reasonable start).

If we ever switch to a truly community run Ludum Dare, then this data might be a good starting point for a rules change, although it would be nice to get everyone to vote somehow if we are going to use this in a binding way.