On Sun, Mar 17, 2013 at 10:36:52PM +0100, Gaudenz Steinlin wrote: > It was pointed out several times, that filling the largest rooms at Le > Camp to their maximum capacity might be unacceptable. I agree with this > and propose to fill the 3 largest rooms (all rooms in the "Large > sleeping-bag" category) only to 2/3 of their capacity, even if we have > more people applying for accomodation in these rooms. I find this proposal to be fatally flawed, driven by a misguided standard of "fairness" to attendees. A person's tolerance for additional roommates follows a logarithmic curve, not a linear one, and each person has their own multipliers. One person will be uncomfortable with but accept three roommates, but will consider five roommates unacceptable; another person will be ok sharing a sleeping space with 12 people, in which case it's very unlikely they will object to sharing a space with 13 (or 14, or 16) people. So the goal here should not be to try to even out the number of people per room, the goal should be to try to even out each attendee's comfort level, with the understanding that there will be *quantum* differences in what individual people will be comfortable with. Some people will accept communal accomodation, and some will not. You can declare that you're going to make it "fair" to all attendees by only offering communal sponsored accomodations; but the effect of this "fairness" is that some people will choose not to attend DebConf at all as a result; and when that happens with people who are at the top of the to-be-sponsored list, it diminishes the conference as a whole. So I posit that: - The attendees who are willing to accept communal accomodations at all are not going to be bothered by having only a small amount of personal space. Underselling the communal rooms reduces the capacity for the conference but brings only a marginal comfort benefit. - Attendees who are *not* willing to accept communal accomodations will represent a significant portion of the attendees, and their needs should be factored into the planning. - The distribution of attendee preferences *may* mean that it's better to undersell the medium-sized rooms (to make them small-occupancy rooms, with a max of 4 people each). But we can't know *for sure* if this is true until after we've collected the registration data, and then only if we make sure we ask this question. As requested at the sponsorship IRC meeting, I've prepared a wiki page that lays out an alternate accomodation model: https://wiki.debconf.org/wiki/DebConf13/Pricing/Strawman This proposal allows for a sliding scale of small-occupancy vs. communal rooms, in response to the actual data that comes in from registrants. At the high end, it accomodates 294 people ("Proposed occupancy (high)"), which is only 5 less than Gaudenz's proposal. At the low end ("Proposed occupancy (low)"), it accomodates only 237 people - but if that's the configuration that actually matches the requirements of the people we want to sponsor, I insist that this is the correct usage of the space, *not* a usage that packs more beds with people we wouldn't actively seek to sponsor. Now, in terms of opening registration, I don't think we need to actually figure out which of the two of us has more accurately predicted the distribution of preferences among registrants. The only thing we need to do is make sure we *capture* those preferences accurately, so we can make an informed decision about room allocation once the herb team has said who should be sponsored. This implies adding a single yes/no question to the registration form for sponsored attendees: [ ] I will accept sponsored communal accomodation (possibly with a link to a more detailed description of the room configurations) If the distribution of responses is really biased as far to "yes" as Gaudenz's model implies, then we can even go ahead with that room allocation model. But if, as I believe, there will be a lot of people we want to sponsor who will not attend if communal accomodation is their only option, then we should follow the data and allocate rooms accordingly, instead of trying to fit the attendees to a preconceived model. I would also note that this is of course a stepwise function. If person n in the sponsorship rankings requires small-occupancy accomodation, and sponsoring them has the consequence that the total number of beds is reduced by 4 and persons n+1..n+5, all of whom would accept communal accomodation, are not able to attend, we should be smart about weighing the value of having 5 attendees vs. 1 attendee who are all ranked relatively close together. But by the same token, if persons n..n+3 all will only attend if they can get small-occupancy accomodation, I don't think it makes sense to skip over them and sponsor persons n+4..n+9 in the same space just because it "fills" the space better. Thanks, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. Ubuntu Developer http://www.debian.org/ slangasek@ubuntu.com vorlon@debian.org
Attachment:
signature.asc
Description: Digital signature