[Chicago] selecting talks Re: [PyCON-Organizers] Talk slot durations (was: FWD: Re: Pycon disappointment)

[Lukasz Szybalski]

On Sun, Mar 23, 2008 at 10:31 AM, sheila miguez <shekay at pobox.com> wrote:
> I'm moving this thread here because I think my reply is too
>  hare-brained and informal for the organizers list. I hate lengthy
>  blahblahblahs on how things should get done and would like some
>  concrete way to test out ideas so that we can cut the crap and get on
>  with it. where "it" is more fun.
>
>  2008/3/21 Atul Varma <varmaa at gmail.com>:
>
>  > I like this idea of people having to "prove themselves" to some extent to
>  > present at PyCon, and I wonder if there's a way to integrate this kind of
>  > concept with local Python interest groups [snip]
>
>  I think the pycon organizers should find a way to test ideas being
>  discussed in these threads.
>
>  Perhaps we could test ideas on previous conferences. e.g. apply
>  proposed selection procedures to historical data sets, and judge the
>  resulting conference models with the historical outcomes. Which
>  algorithm rejected/selected what were considered high quality talks,
>  which algorithms selected talks that were not selected in the
>  historical conference? &c.
>
>  We won't have the benefit of knowing the quality of talks that didn't
>  go in to previous conferences (unless rejected talks were given
>  somewhere else and we have that data), but perhaps we have enough data
>  to run through voting scenarios on past conference talks to compare
>  generated scores with actual talk quality.
>
>  Alternatively, If we can obtain access to data from upcoming
>  conferences we could run through scenarios to compare hypothetical
>  model against actual results.
>
>  If this is easy to test, we could at least get an idea of "holy crap
>  that method sucks!" though perhaps they will all be too close to call.
>
>
>  This is just something I was pondering whilst thinking of different
>  ideas on voting and also ranking talks to consider scheduling--Randall
>  Monroe once had a voting scheme on ideas for the perfect date, perhaps
>  similar to the voting schemes he uses on sites like
>  <http://bestthing.info/algorithms.html>. (loosely based, he says, on
>  Condorcet voting criteria)
>
>  Also, being a gamer, I thought of the voting system with different
>  criteria for giving voters more points to vote on talks. and talks
>  would have different costs for voting. by analogy, designing stats and
>  leveling up characters.
>


Are there any data available that tells you who picked what talk, how
many people showed up, who registered for what, some survey on a talk
(good +, bad -), any other indicators?

pycon 2008 and previous pycons?

Lucas