Hi Marshall - A method that works for any ratio using running totals:
Let NAp, Ep, Ap be the value of each regions part of the ratio (e.g. NAp = 1.7 for a 1.7:1:1) Set NAt(0), Et(0), At(0) = 0 Set NAs= NAp/(NAp+Ep + Ap) (basically the decimal version of the ratio), repeat for Es and As For each meeting 'i' let NAt(i) = NAt(i-1) +NAs - (if meeting in NA ? 1 : 0) and repeat the same for each region. [Basically add the meeting credit XXs (NAs, Es, As) to the running credit total and subtract one from the credit if the meeting is held in the region] XXt(i) gives you the number of meetings "owed" to that region - or if negative, the surplus. This is easiest to see if you map it out on a spread sheet. A simple model is to attempt first to plan a meeting in the region with the greatest arrears falling back to the next greatest etc. Any luck in deriving the other data I asked about? E.g. looking only at the Nomcom qualified and the "contributors"? Thanks - Mike At 12:24 PM 9/1/2010, Marshall Eubanks wrote: >On Sep 1, 2010, at 11:24 AM, Ross Callon wrote: > >> Why does this have to be precisely on an integer-number year boundary? > >It doesn't have to be. It could, in principle, be anything that the community >wants. > >My real point here, which I may not have gotten across, is that meeting >scheduling is a fairly blunt tool. The community cannot expect too much >precision in this. At any time there are meetings at various degrees of being >scheduled some years out, meetings have had to be rescheduled (i.e., >prospective venues have fallen through) in the past and likely will again in >the future, and external events also sometimes constrain when we can meet >where. If the desired X:Y:Z goal is not being obtained all the IAOC can do is >to change or swap meeting locations, and there are generally strong >constraints on that (i.e., some meetings may be firmly scheduled some time >out, there may be cancelation penalties on some Hotel contracts, certain >sponsors may insist on "their" meeting being in a certain location at a >certain time, etc.). Then the IAOC is open to complaints such as "there are 2 >meetings in Region X back to back", or "there are no meetings in Region Y at >all this year." > >So, I would recommend simple goals with short repeat cycles, such as 3:2:1 or >1:1:1, and also repeat cycles that commensurate with a integer number of years >(where that integer is 1, 2 or 3). I don't think the system is likely to >deliver more fine grained performance than that. > >Regards >Marshall > > >> >> Ross >> >> -----Original Message----- >> From: [email protected] [mailto:[email protected]] On Behalf Of >> Marshall Eubanks >> Sent: Wednesday, September 01, 2010 10:56 AM >> To: Scott Brim >> Cc: Adrian Farrel; [email protected] >> Subject: Re: IETF Attendance by continent >> >> On Aug 28, 2010, at 1:25 PM, Scott Brim wrote: >> >>> On 08/28/2010 12:28 EDT, Adrian Farrel wrote: >>>> And even closer to 3:2:2 ? >> >> I think that people have unreasonable expectations about what we can do here. >> >> There are 3 meetings per year, and 3 meeting regions being considered, and >> we are generally considering something between 1 and 3 years out at any >> time. >> >> Suppose that the time horizon is 2 years. Then, an equal meeting schedule is >> >> 2:2:2 (which is equivalent to 1:1:1, of course). >> >> If we shift one meeting, we have >> >> 3:2:1 (the current proposal) - or 1:0.66:0.33 >> >> If we shift 2 meetings, we have >> >> 4:1:1 - or 1:0.25:0.25 >> >> and that's it. Without having no meetings in some region, 1:1:1, 3:2:1, or >> 4:1:1 is all we can chose between with a 2 year horizon. >> >> (You have to chunk the meetings somehow to get these ratios; doing by >> calendar years is a very reasonable chunk that fits well with the way that >> meetings are scheduled.) >> >> Suppose that our time horizon is 3 years - then an equal meeting schedule is >> >> 3:3:3 and we can shift meetings to produce >> >> 4:3:2 - or 1:0.75:0.5 >> 4:4:1 - or 1:1:0.25 >> 5:2:2 - or 1:0.4:0.4 >> 5:3:1 - or 1:0.6:0.2 >> 6:2:1 - or 1:0.33:0.16 >> 7:1:1 - or 1:0.14:0.14 >> >> and that's it (without dropping some region entirely). >> >> So, for example, instead of 3:2:2 (or 1:0.66:0.66) I would recommend 4:3:2 >> for the next 3 years >> (the closest triplet using an absolute value sum metric on the differences). >> 4:3:2 would be easier to do than 3:2:2 based on the way we schedule and >> review meeting locations. >> >> Now, of course, meeting locations do get moved, and 4:3:2 might easily turn >> into 4:4:1 or 3:3:3 based on contingencies. >> >> I do not think it is reasonable to apply a time horizon of > 3 years to IETF >> meeting locations. Attendance is changing too rapidly for that. >> >> Regards >> Marshall >> >>> >>> +0.2 >>> _______________________________________________ >>> Ietf mailing list >>> [email protected] >>> https://www.ietf.org/mailman/listinfo/ietf >>> >> >> _______________________________________________ >> Ietf mailing list >> [email protected] >> https://www.ietf.org/mailman/listinfo/ietf >> > >_______________________________________________ >Ietf mailing list >[email protected] >https://www.ietf.org/mailman/listinfo/ietf _______________________________________________ Ietf mailing list [email protected] https://www.ietf.org/mailman/listinfo/ietf
