>>>>> "Kurt" == Kurt Roeckx <k...@roeckx.be> writes:
Kurt> On Mon, Aug 31, 2015 at 04:49:08PM +0100, Dimitri John Ledkov wrote:
Kurt> One of the problems, and I consider that to be the most
Kurt> important one, is about the stratigic vote that you can do.
Kurt> For example, condiser that there are 2 options (A and B) plus
Kurt> the default option All options are acceptable for everybody,
Kurt> but 75% prefer A and 25% prefer B. You would except the
Kurt> following vote: 75%: 123 25%: 213
Kurt> Option A would win as expected.
Kurt> If there is a 3:1 majority requirement, you could instead
Kurt> vote: 75%: 123 25%: 312
Kurt> As in, the 2nd group says that option A is not acceptable
Kurt> while in fact it was.
Kurt> This results in the option A being dropped because it does not
Kurt> reach majority. 75% say A acceptable and 25% say it's not
Kurt> resulting in a 3:1 majority saying it's acceptable. The 75%
Kurt> just don't reach the "strictly greater" than the 3:1 majority
Kurt> requirement.
Kurt> In the end option B wins because of stratigic voting, while if
Kurt> they were honest option A would have won.
So, first, I think we're all agreed that we want to fix the strictly
greater issue.
That is, the example above fails both because you need to be strictly
greater than the majority requirement (equal doesn't count) and because
of the strategic voting.
I'd like to set aside the difference between strictly greater and
greater because that's unlikely to come up in a GR, and because I think
many of us agree we'd like to fix it.
However, I'm skeptical of the strategic voting problem and even more
skeptical of the fix.
What you're saying above is that 25% of the people find option 1
distasteful enough that they are willing to game the system but somehow
they still find option 1 "acceptable."
Assuming we believe that a super majority should apply to this
situation, we believe that enough folks who consider option 1 to be
unacceptable should cause option 1 to lose.
In my mind, distasteful enough to try and game the situation is fairly
good evidence that someone finds an option unacceptable, even if they
wouldn't phrase it that way themselves.
I have been watching how people rank things above and below FD both in
GR elections and in TC votes over the years. I have not done any
scientific analysis, but I've been amazed at the thought and care people
seem to put into that decision. And from what I can tell, people seem
to do a good job of thinking of the question as "would I rather get
stuck in another round of discussions than have this option win?"
As an example, the only reason we ever exited the process on init
systems in the TC is that enough TC members felt being done was more
valuable than having their preferred option win. If you look at the
discussion, you see that people on both sides put a lot of thought into
this.
There's a natural defense to this type of strategic voting that I've
seen us employ. Have an option similar to option 1 on the ballot that
is a statement of position rather than a constitutional change;
something without a super-majority requirement.
We're fairly good about employing such options on ballots where they
make sense. For example I'm thinking of the discussions surrounding
amending the DFSG to remove non-free.
Kurt> The solution to this problem is moving the majority check
Kurt> later in the process, so that option B would have been dropped
Kurt> first. If they did this stratigic voting in that case both
Kurt> options would have been dropped.
So, I think this opens up far worse problems than it solves.
let's take a specific example.
Let's assume that option 1 is amend the social contract to remove
non-free.
Option 2 is some statement that discourages non-free, but isn't strong
enough to be a supermajority change.
If option 1 wins but fails majority, w end up with FD winning.
So, then what. We'd like to have a ballot without option 1 so we can
make progress. Except now we've created a strategic opportunity for
five developers to put us back into the same position and put option 1
back onto the next ballot.
Forcing ourselves into endless rounds of more discussion on the most
controversial issues is not an improvement.
If we don't make a change, it's important that that we have a
non-supermajority option that is in the same direction as a
super-majority option if there are any options on the ballot that are in
significantly different directions.
That is, a ballot that was remove non-free and FD would be OK, but a
ballot that is remove non-free, reaffirm non-free and FD would be a bad
idea. Instead you'd want some version of discourage
non-free-but-not-requiring supermajority to be on the second ballot.
In conclusion, endless discussion is not a win. And I think this
strategic voting fix may bring us there. If I were to put together an
amendment that fixed the strictly greater issue but did not tackle the
strategic voting issue, would people second?
