Bug#727708: Call for votes on init system resolution
Adrian Bunk <bunk@stusta.de> writes:
> On Fri, Feb 07, 2014 at 07:22:10AM +1000, Anthony Towns wrote:
>> Presuming everyone votes, where you put F only has an impact in either
>> case only if at least three other ctte members will also vote FD above
>> T or DT (given UT is irrelevant); which based on the already expressed
>> preferences and votes, isn't actually going to happen.
> Why not?
> I am not sure whether Colin is aware that it currently depends on him
> whether or not DT can win - and whether that might make him consider
> changing his vote.
> If Ian convinces Colin to change his vote to move DT from 5. to 7. on
> his ballot, then DT cannot pass FD and is dead.
Which is just a concrete way of pointing out that Debian's standard
resolution method fails later-no-harm.
https://en.wikipedia.org/wiki/Later-no-harm_criterion
This is a known weakness in Condorcet, which I suspect we have made worse
with the way that Debian treats FD.
This is one of the major reasons why I'm voting GR second. I see Bdale's
point that we shouldn't abdicate our responsibility to make the best
decision that we can, and I followed that maxim by voting my preference
first. But the reality is that, regardless of whether Condorcet is
capable of spitting out some technical compromise, we are deadlocked, and
sufficiently so that we're running into edge case failure modes of the
standard resolution method.
In that situation, the TC recusing itself is not the worst thing that
could happen.
--
Russ Allbery (rra@debian.org) <http://www.eyrie.org/~eagle/>
Reply to: