On Thu, 30 Jan 2020 at 17:03, James Cook <jc...@cs.berkeley.edu> wrote: > Here's a somewhat different way we could do it: > > * An announcement resolving a decision doesn't need to specify > anything other than the decision --- not even the outcome. That causes > the decision to resolve to the (platonically) correct outcome, and it > is self-ratifying that that occurred. > > * The resolver SHALL include all that extra stuff in their resolution > message (and maybe SHALL respond to CoEs). > > Is there anything wrong with that? I feel with the current system, > even when we eventually figure out which proposals are adopted, > there's some disturbing temporary uncertainty about when exactly they > were adopted, which doesn't seem better than the temporary uncertainty > this version would introduce about what the outcome was.
As I often do, I sent this just a little too soon and should have thought more. An obvious flaw with what I wrote is that we may never know for sure what exactly self-ratified, whereas the current system explicitly makes the outcome ratify. Anyway, I like G.'s proposal, but why even require a reasonably accurate tally for it to be self-ratifying? Just require decision+outcome, and make the rest SHALL. - Falsifian
