Hmm, excellent question... I would think (I'm also not a connoisseur per se, just someone who really likes them) that a reasonable metric would depend on the gameplay - the more closely something resembles a narrative text, the more closely your original 1k minimum can apply without much tweaking. The more its mechanics vary and tend towards a computer game, the more you'd want to consider time spent playing. IF isn't intended to be prioritising gameplay over narration, though. So if you split it 50/50 game / text, I'd think 10 minutes to play the game from start to finish would make sense.

For example, if it's a work that has a puzzle in it (let's say, a casefic where you solve the case), then the complexity of the puzzle should give for that 10 minutes play time, which would include the time to read the background, clues, etc. Assuming average reading time is of order 200 words/minute, and that it's split 50/50 reading vs solving puzzle, then 5 minutes reading gives you about 1000 words already (TIL: I am a very slow reader) and then ballpark 5 minutes to solve the puzzle. 10 minutes might be too short, but it could be better to err on a minimum and have people surprise themselves. It's your call, though.

For most types of narrative IF where the focus is really on more on the text than on a gameplay feature (like choose your own adventure type stuff - mostly the document is text with a few choices) - you could apply something like 1000 words / (number of possible paths in the story from start to an end state, ignoring loops if there are any). Therefore, if there are no choices, there's only one path, and it recovers your previous word minimum. That might be too simplistic.

For parser games/text-adventure type games, I think aiming for 1000 words as a minimum is already doable, and you'd probably make this just from the background text alone.

I don't know what kind of generators people intend to use, but speaking from my own experience in Twine, you could also consider setting a minimum based on number of passages. I think at least 15 total passages would be pretty reasonable. While that may not result in 1000 words in an average run of the game from start to end state, accounting for slightly different paths in different episodes would. It'd presume that the recip would be interested in replaying the game and making different choices. (Also, passages in Twine have a tendency to, uh, proliferate. 15 is a pretty easy minimum to make.)

I doubt people who are offering IF (in fact I... may be the only one) will code a parser game or a puzzle game, so this would more likely be a choose your own adventure type with a few paths. It should be easy to calculate a minimum off that using the guidelines above, but AO3 won't reflect that in the word count (especially if the game is linked to from another website).

Anyway! Just throwing some thoughts out there. It might be easier all things considered to a) tell people who are interested in IF to just wait for an unconventional fanworks exchange (that's valid, I'm still keen on signing up without IF) or b) tell those recipients who would be interested in potentially receiving IF to leave a comment on an opt-in post, like Yuletide did this year. And maybe IF can wait for a future year when we have better minimums formulated for it. If that's your decision, I would totally accept this too.

That could work but solely for the IF that is the choose your own adventure type. Also without further note about what choice point entails it's hard to gauge the complexity of choices that land you in different game states vs dramatically changing the outcome of a story; and whether the choice is binary (much easier to program) or anything higher.

I think to work off of minimum choice points would bias you to a particular IF structure. Unless the point is to restrict to solely that structure, which would be a choice you could make. To go back to your original minimum requirements for Other, "equivalent effort" in the IF should probably relate to some judgement on the blend of: a) the complexity of game structure, and b) the length of total text in the fiction from start state to finish, such that games that are pretty convoluted to program but don't have a lot of text are judged on the same relative level as a game that has relatively few choice points but a path from beginning to end is guaranteed 1k words. But it's up to you how precisely you want to define what "equivalent effort" is and whether people will take that in good faith. And this all might be annoyingly and unnecessarily complex to code into a rule anyway. Personally I (and everyone I know who has received IF) have yet to be disappointed with an IF piece - as in there was never a feeling like IF gifts were somehow lacking, in exchanges where there was nothing more precise than "text must be 1k" - but others might have had different experiences.

A vote sounds great!