[kwlug-disc] What is all this about systemd?

[Hubert Chathi]

On Wed, 19 Nov 2014 15:26:06 -0500, Ralph Janke <txwikinger at ubuntu.com> said:

> I think one big issue (like with lots of things in Democracy) is the
> way the vote has been set up. Besides, that it does not make sense
> that all 4 options pass majority, ... Option 1 and 2 (Mandatory /
> Recommended not to require one init system) should be seen in context,
> so if I read the numbers right, Option 1 & Option 2 together had more
> votes than option 4 (no GR required)

I think that you are misunderstanding the voting system that Debian
uses.  Debian uses a Condorcet method, which can be seen as a
simultaneous pairwise vote on all the options.  In essence, each vote
encodes the voter's preference between any to options.  The winner is
the option that wins all the pairwise votes.  So you cannot add the
numbers for two options together to compare to a third option, because
the voting method already compares each pair of options together.  i.e.,
the voting method already determined that option 4 is preferred over
option 1, and option 4 is preferred over option 2.  If you try to add
votes together, you would be counting some votes (i.e. any vote that
ranks both option 1 and option 2 above FD) twice.

It's a pretty awesome voting system, and once you get used to it,
everything else seems suboptimal.

The fact that all 4 options pass majority can be read as the Debian
Developers wanting *some* decision to be reached, no matter what it is.

> I also think the problem is democracy in governance, since it is about
> popularity and requires participation. Out of 1066 possible voters,
> only 622 ballots where received and only 458 (< 50%) where unique
> voters.

It's actually a pretty significant voter turnout.
https://lists.debian.org/debian-vote/2014/11/msg00192.html
compares the voter turnout of this vote against other Debian votes in
the past, and this vote is one of the highest turnouts, both by absolute
numbers, and by percentage, especially if you omit the DPL elections.
(I don't know why DPL elections generally have higher voter turnouts
than GRs.)

> And this are only Debian Developers, not even Debian Users.

> There are certainly arguments for and against if the vote should be
> broadened to the public at large or users, however, it would be nice
> to have at least a non-binding additional vote from that pool to see
> the differences if they exist.

As much as I would agree that "it would be nice to have", I think that
various practical considerations would make such a vote pretty much
impossible to run fairly, and to give real meaningful results.  In any
event, the Debian Social Contract[1] section 4 means that Debian
Developers should have, to the best of their abilities, taken into
account the opinion of Debian users.  So even though the vote was only
among Debian Developers, this does not mean that the opinions of Debian
users was not taken into account at all.

[1] https://www.debian.org/social_contract