On Thu, May 19, 2016 at 4:37 PM, Sylvain Bauza <sbauza@redhat.com> wrote:
Le 19/05/2016 16:04, Florian Haas a écrit :
On Thu, May 19, 2016 at 3:40 PM, Sylvain Bauza <sbauza@redhat.com> wrote:
Le 19/05/2016 15:19, Tristan Goode a écrit :
So let me get this straight... You're proposing that only speakers vote for talks yeah?
Because hey, fuck the audience right?
I'd be totally opposed to that idea that would induce a clear bias. Let me explain : while the Foundation is trusting a different set of people at every Summit for each track, the above would create a define list of people that would be quite the same for each Summit - because we know that people naturally tend to prefer their own close relations.
How is that a factor if out of hundreds of talks you only ever review 8-10, randomly selected? What are the odds that your own close relations will even be in the subset you review?
Because you're not randomly picking a list of talks when reviewing. You naturally tend to review the ones you feel good at, where you built your network and where you know people. Or, on the other hand, you naturally tend to select talks made either by your company peers, or the social network relatives you're close to.
As I have tried to point out repeatedly, I am proposing a system where the randomized talk selection *is done for you*, so you never see all talks up for review, you only see those allotted to you. You review those, and your review is done. You don't get to see any others. Someone else gets a different set, which may or may not overlap with yours. If your group of reviewers is big enough (a given at the Summit), that means you get multiple eyeballs on every single proposal, no proposal goes unreviewed, and no-one has to deal with a daunting pile of submissions to review.
Anyway, my point is that you're already biased when picking a subset of talks to vote to (because you can't, as an human, vote for 800+ talks). The vote itself is another bias, where you tend to +3 the ones you trust, because you know the proposer, and not because the proposed abstract is great.
Don't blame me, but I'm seeing this as an argument about how much we currently trust the track chairs as non-biased people. FWIW, if we agree with the fact that track chairs are good for their duty, why should we change how we select them ? Maybe the proposal is to leave the track chairs, and only allow votes from previous speakers ? If so, that's even more terrible : we're moving from a representative democracy (the track chairs) to an oligarchy.
Nope, that's not my proposal. Mine isn't about previous speakers, it's about current talk submitters.
Fair point, that doesn't really change my opinion. My paranoid mind also leads me to consider a Foo company wanting to promote a certain talk by adding more talks hence more voters for that specific talk.
Yes, this is Duncan Thomas' point from several months ago that I've already addressed upthread. Cheers, Florian