[some discussion on this and especially questions around what happens when the user's hands differ - e.g. if they are missing a part of their finger...]

Alice: what are the extra joints for... how does it work for missing joings, missing fingers?

Yves: each hand is one imput source...

Alice: medical names is kind of a no-no from our naming principles....

Sangwhan: it's very medical.

Alice: that is kind of the main point of feedback - asking people to memorize the bone names...

Hadley: but if we need a vocab, that's a good one to use...

Sangwhan: could use index numbers...

Alice: i suggested that... referencing the unity API... easier to understand... The response made some sense. Different platforms expose different amounts of information - if you're using a numbered system you still want to be able to see which parts of the hand are included in that number system... It's a subset of possible bones.

Sangwhan: if you extend it - you have this weird situation where you have 1,2,3,4,.... when you add something in it could become 25.

Alice: the last commetn said it's changing the constants to enums... which avoids this problem.

Alice: nobody has come up with a plain english set of names for every bone in the hand...

Hadley: there are lots of resources out thee [to help developers know these names]

Hadley: I can imagine a future need to name other bones in the body and we've already used this special thing for hands...

Alice: it's about detecting a gesture... based on image recognition... If there were some other vocab to use we should use that but there isn't.

... I gave some feedback to ask if they could consider a richer data structure...

Dan: performance issue... something something uncanny valley

Sangwhan: isn't this API just exposing a network of vertices in a semantic OO form...

Alice: I would like to see the updated examples... They talked about it being iteratable.. Don't understand how that helps...

Alice: I think we are actually done with this one. I should respond to the last comment that says I have been spending effort imaginging a more inclusive system. it's very hardcoded that everyone hasa got two hands and five fingers.. well any number of hands. Any given hand has 5 fingers, any given finger has the expected joints. Which obviously doesn't bear up to reality. We talked about this is the baseline API until someone ships an API that takes that into account, but they've got to do the classic thing, it's hardware, so the superset of what all the underlying things do. I was posing the question of whose responsibility is it to come up with the more inclusive version, so the answers are quite good, they have ideas, have been working with some of the WGs including device manufacturers, but they don't hae time yet. That seems pretty good. The one outstanding issue that another commenter raised right up in the second comment about privacy, the fact that it returns real metrics about peoples finger lengths etc and this is the one where the explainer starts talking about hand phrenology. It's really wild. It might be worth reiterating that point. They ask for elaboration on anonymity - we don't have a clear idea of this.

Dan: it says implementations are required to employ strategies to mitigate, reduce precision of sampling data, adding noise

Alice: I guess that's fine. I suggested could it be a richer data structure and a bunch of enums and they said no not right now. They did change it to enums instead of constants but other than that. It looks pretty good. I'm happy to comment and propose closing.

Dan: I think we should do that, we've gone around and around and they've been responsive to our feedback. I support that.