Cullen Jennings <fluffy@xxxxxx> wrote:
> I just want to point out one other issue in with this type of
> algorithm. When you do this for a given WG, you often get a time that
> is after 8 am pacific time and before early evening in Europe as that
> is simply the distribution of participants in many WGs. If, for every
> meeting, you always use this algorithm, you always get a time that is
> really bad for Asia. That does not seem fair and even if you don’t care
> about fair, it does not seem like an optimal outcome for the
> internet. I don’t think that anyone wants an algorithm that selects
> meetings that are always convenient for silicon valley at the cost of
> never being convenient for Asia.
I agree that this happens.
> It’s really hard to decide what is it we are trying to optimize when
> choosing meetings. For a bunch of the WebRTC meeting we tried to place
> them in the location or times where the frequency of how often the
> meeting was convenient for users in region X was roughly proportional
> to the number of participants from region X.
Part of the challenge is that there is effectively nobody (except Brian
Carpenter!) living in the 8 hours/timezones between China and San Francisco,
so "night" almost always winds up in the pacific. So, if we could just move
a couple of extra million people (including a thousand IETFs) to New Zealand...
Or Google's Syndey office.
So we need a variation of the doodle poll, where we simply ask what your time
zone is. Maybe we could actually just pull that out of mailing list
headers. Hmm. There is some bias there, but it might be a good bias (post
more, and you'll get it at a better time?), or it might wind up as a
self-selecting bias (since the time always sucks, you never get involved, and
thus never post).
But, I agree with making the pain shared.
--
Michael Richardson <mcr+IETF@xxxxxxxxxxxx>, Sandelman Software Works
-= IPv6 IoT consulting =-
Attachment:
signature.asc
Description: PGP signature
