I forgot that we needed a code of conduct for NumFOCUS. This feels like NumFOCUS overstepping their bounds, though with good intentions of course. If they put too many constraints on us, we may have to break off and form our own foundation.
I don’t actually think Columbia has a role in running Stan other than indirectly because they pay a bunch of us and hold the copyright to a bunch of the code. In that way, Columbia’s no different than Metrum, Novartis, University of Michigan, etc.
NumFOCUS does have some practical role because we use it for general funds outside of Columbia, so it’s generally like community money to spend. Most of it’s pretty tightly earmarked for either programmers or conferences or equipment (I think we have less than $30K that’s not earmarked).
I would very much like centralized control rather than splitting things between the Stan community and the NumFOCUS (I want to do an Andrew and respell their name as “Numfocus”, but it’s just not in my nature).
There’s no limit to the size of the NumFOCUS leadership body, only a constraint that a majority of members are not from the same institution. So there’s no reason we couldn’t in principle route everything through the NumFOCUS leadership body. The question’s then just who to put on and how unwieldy will it be when we need majority votes to pay invoices?
Well put. They’re necessary but not sufficient conditions. We constantly get into “debates” about what consitutes complaince with project standards, what constitues complete doc and how extensive unit tests should be. I think adding use-case and anti-use-case recommendations is a great idea.
As far as diversity, we have exactly one woman developer and I don’t think we have any “underrepresented minorities” by NSF’s measure. Our NumFOCUS board is more diverse, so maybe we could tap that.
@sakrejda: while it’d be nice to have all of the bits (lpdf, lcdf, lccdf, rng) for every distribution we add, I don’t think we should make it a requirement (just my opinion of course, and it contradicts what Michael suggested in the meeting). Ideally, we’d like all those things if they’re possible.