I'm sure counting votes was easy once you got the data, but I'm curious why you didn't use AI.
Because I wouldn't have trusted whatever answer it gave me.
Each time I've played around with ChatGPT (and other LLM-based stuff), I've been disappointed by the correctness-level of its responses compared to the confidence-level with which it speaks. So, I would've had to "check" its work anyway, effectively by repeating it independently; may as well then just do the work myself and not rely on its "assistance" in the first place. Especially with technical stuff, it falls apart quickly. I suppose, if you don't care whether the answers you get from it are
actually right or not, then ChatGPT must seem like this incredible tool from the future, but to me, it's so often wrong that I've stopped wasting my time even experimenting with it.
I keep hitting your Merit limit of 50 per month. Maybe it's time to patch that

I mean, you kid, but it's occurred to me more than once that it would be nice to be able to leave "latent" merit on a post. (That is, if you can't merit a post because of a lack of sMerit or because you've hit the limit for that user, then it would be nice if you could still leave what amounts to a "merit promise" that will get automatically fulfilled when possible. There'd still have to be some kind of limit, so that people don't create unrealistically deep backlogs, and I think unresolved merit transactions probably shouldn't be visible to anyone but the sender, otherwise you risk creating an incentive for people to sometimes send merit only because they're trying to "unlock" the latent merit that they've been sent. But, those two niggles aside, I think it would be a handy feature.)