The idea here is to give a simple multiple choice question for how likely you are to attend a thing. Each option should be understandable to non-nerds but also include a range of probabilities. For example:

(a) I'm very probably coming! (>90% chance)
(b) I'm maybe probably coming (>50% chance)
(c) Unlikely this time but still hoping I can (<50% chance)
(d) Sad trombone — I probably can't come! (<10% chance)

And there could be an "other" option that makes you enter a specific probability. Nerds can use that if they want something in between the given options — like "quite likely" which might be, say, 80%. And non-nerds could use it to say things like "definitely can't make it", ie, 0%, or "wouldn't miss it for the world", ie, 100%. Or 50% for "right on the fence".

(Or maybe it's better to minimize the cognitive load by only allowing a fixed handful of choices. Roughly: yes, probably, probably not, and no.)

Anyway, you end up with a list of people each with min and max probabilties. (Anyone who picked "other" will have min=max, which is fine.) So then you can get a range on the expected number of people who will show up. Just add up the min probabilities for the lower bound and add up the max probabilities for the upper bound.

That's not actually implemented yet so you can just type anything you want below but go ahead and use it for a genuine RSVP for the August SSC meetup. [UPDATE: I'm not maintaining this at the moment. Don't actually RSVP here now!] Or if you think it would be fun to help build this thing in Glitch, which is super fun btw, go ahead and say so here:

Responses, if you please...

    Attendance estimate: 00

    Past case studies...

    For the Slate Star Codex meetup on 2017-05-07: The RSVProbably algorithm, as we could maybe call it, predicted 9.05 to 12.7 people. Actual was 13 people at the peak but one person arrived not too much before the first person left. If we count everyone's attendance fractionally based on arrival and departure time then the total attendance was 10.25 people.

    For the Pomodoro Poker Hack Night at the Beehive on 2017-05-09: The algorithm predicted 3.2 to 3.8 people. (Though people's RSVPs changed as the event approached and it was a bit arbitrary where to draw the line between altered predictions and realizations of outcomes. Like if someone who was probable says an hour beforehand they're probably not coming after all, that should probably count as a wrong prediction rather than an updated prediction that turns out to be right when they don't in fact show up.) Anyway, actual attendance was 3 people.

    For the Slate Star Codex meetup on 2017-06-10: Predicted: 6-7.1 people. Actual: 9.5 people.

    Slate Star Codex 2017-07-09: Predicted: 1.4-1.8. Actual: 4 people.