On 3/13/21 3:35 PM, Brian E Carpenter wrote:
Do an experiment. Run a meeting every 6 days (so in a few weeks we have an indication of the most popular days of the week) and run it one hour later each time (so that in a few weeks we have an indication of the most popular hours of the day).
As above, but treat showing up for a meeting on a particular day of the
week at a particular time slot as if it were a vote indicating
preference for a meeting day (of the week) and time. Imagine that
everyone's individual vote (as expressed by attending a meeting) is
written on a piece of paper and placed in a (virtual) hat. Then for
subsequent meetings, pull one of those pieces of paper out of that
virtual hat and hold the meeting on that day and time. Discard votes
after N weeks of their being cast so that the votes continue to reflect
(nearly) current preferences. That makes it "fair" in the sense that
everyone's preference is equally represented, though people with less
common preferences will get meetings in their favorite days/time zones
less often.
(bikeshedding is fun!)
Alternatively, compute statistics on the times of day and days of the week (in UTC) when the most messages are sent on IETF lists. We already have the data in the mail archives, and that can tell us objectively when people are most available for IETF business.
Being available to send an email is very different than being available
to chat, though.
Keith
