Re: [Debconf-team] Early travel sponsorship
Moray Allan dijo [Thu, Mar 15, 2012 at 10:22:48PM +0000]:
> - In an ideal world, I don't think we would do travel money allocation
> completely independently each year. While a new method for choosing voters
> might increase the variation in results between years, making the problem
> less bad, if you suppose that we magically knew the correct fair ordering
> of everyone, I don't think we should simply allocate the money to the same
> top-ranking people each year, but spread it further down the list. (If we
> ignore this question for this year, we might at least start to keep more
> accessible records of how much money has been allocated to who, without
> other private information, that can be used in future years, if desired, to
> even things out a bit.)
>
> - In an ideal world, I don't think we should ignore the amounts of money
> being requested. While it's clearly hard to make judgements between
> specific individuals, if you imagine a case where person N on the list will
> cost the same amount of money to transport as people N+1, N+2, ... N+10, it
> might make sense to bring the ten following people before person N. In
> practice, we *do* see big variations between flight costs; this isn't
> merely a hypothetical problem. (I also realise that we might need to take
> into account the reasons for high flight costs, and e.g. whether they're
> likely to be lower the following year.)
>
> - To deal directly with the most common case of that: just as we had a
> separate pot of money for "DebConf newbies" in the last couple of years, it
> might make sense to have separate pots for regional vs. distant people.
> It's hard to trade off the two cases when looking at individual people, so
> it might be easier to make an overall split of the money. It's good to
> bring people to DebConf from far away, but it's also an efficient use of
> Debian money to bring people to years when they are relatively closer to
> the venue. (What is considered "regional" would vary between years
> depending on how travel prices look to the destination, rather than being
> defined as a specific level of distance.)
Hmm... I agree with you up to a certain bit. I realy approve and like
the "DebConf newbies" separate queue, and the idea of having several
queues has quite a bit of merit. However, I cannot be too enthusiastic
about it. It would not be optimal to have too many queues (as there
are too many vectors for wanting somebody to attend). A needless
balkanization of the travel queues (hey, there will be no problems!)
would be for all purposes identical to the system (or lack of it) we
currently have. Also, the more categories we have, the harder it will
be to categorize somebody — Just following your stated examples,
having newbies/regular queues and regional/distant queues means there
are at least four effective queues (regional newbie, distant newbie,
regional repeating, distant repeating). And many queues will lead to a
third state (of people we don't know to which sub-queue they would
belong).
> - It would be fair to require some more explicit information from people
> seeking travel money, to make the voters' job easier. We should also
> perhaps also require people who receive it (and if so probably anyone who
> received sponsored accommodation/food) to write something about what they
> did at DebConf, and keep that information such that it could potentially be
> used in future decision-making.
Right, this is very true. We have asked attendees to provide some
information to help the ranking, but it's often far from enough. And
this information should probably be quantifiable (i.e. not free-form
text).
Reply to: