Thanks Mark,
This is very interesting results, it is ok if not 100% correct which I
think the error can be less than 10%, but I may have different
analysis of results. You concluded that homes in Europe had better
shortest distances to IETF meetings (assuming that thoes homes have
full participation of all meeting from 2009). Always Europe gets
better results because it is the favoriate meeting-location for ALL
businesses, but hope that IETF changes most of its meetings in Europe
which will serve all. I will post in future another message of my
opinion of a best practice of meeting locations which may become a
good I-D start (but need your assistance and your data results)
My analysis will discover that the past locations of IETF meetings
(from 2009 until now) does not serve-best the majority of full
attendance-participants of IETF from all regions. Thoes past locations
serve only regional-meeting-participants.
My analysis discovers that there is not close air-time-results when
comparing between homes/regions, the results variances between home
cities are large.
Overall, I suggest to consider the number of participants that are
full time attendance in all most meetings 90% attendance (not 75%).
Also to consider the number of past attendance of meetings in Europe
and Asia (there are large differents I think). These two
considerations will show that past locations of meetings did not serve
better the IETF but served some regional-interests. IMHO, two meetings
in Europe per year is a better practice for IETF activities.
AB
On 5/31/13, Mark Nottingham <mnot(_at_)mnot(_dot_)net> wrote:
In an attempt to inject some data into the discussion, I wrote a bit of code
that figures out how much time, given your home city, you would have spent
in the air if you'd attended all IETF meetings since IETF74 (i.e., from 2009
onwards).
The first column is the "home" airport.
The second column is the great circle time between the home airport and the
nearest large airport to the IETF meeting, hhh:mm. This doesn't count things
like transit time, taxiing, takeoff and landing overhead, indirect routing,
etc. As such, this is an ideal number; the only way to achieve anything
close to it is to have a private jet (with exceptional range).
The third column is the time (hhh:mm) using the shortest-time routing on a
travel booking engine. This is first-takeoff-to-last-landing time.
Both numbers assume round trip between "home" and the IETF airports.
SFO 204:10 282:04 // San Francisco
BOS 197:42 297:38 // Boston
ATL 205:44 297:28 // Atlanta
ANC 197:12 345:54 // Anchorage
LHR 198:02 249:44 // London
FRA 202:10 255:22 // Frankfurt
FCO 223:52 283:04 // Rome
SVO 211:28 287:14 // Moscow
TLV 264:12 334:22 // Israel
DXB 293:26 344:34 // Dubai
NRT 259:00 314:38 // Tokyo
HKG 296:38 359:22 // Hong Kong
BLR 332:28 448:24 // Bangalore
MEL 450:28 556:04 // Melbourne
AKL 442:24 569:04 // Auckland
JNB 414:30 498:22 // Johannesburg
EZE 411:10 522:56 // Buenos Aires
GIG 381:56 488:32 // Rio de Janeiro
Draw your own conclusions, of course.
One observation is that there's a 3+ days-in-the-air per year variance if
you're a full-time participant, depending on where you live. I.e., more than
one day-per-meeting difference, on average. In the air alone.
Another is that, perhaps surprisingly, the "closest" homes to all meetings
are in Europe, not the US (at least by shortest-time routing).
I can run other airports upon request, as well as make source available, but
will do so conservatively, so as not to incur the ire of the services I'm
(ab)using.
Regards,
P.S. The IETF airports chosen were:
IETF_airports: [
"ORL",
"ATL",
"YVR",
"CDG",
"TPE",
"YQB",
"PRG",
"PEK",
"AMS",
"LAX",
"HIJ",
"ARN",
"SFO"
],
--
Mark Nottingham http://www.mnot.net/