[CREATE] Decision for LGM 2011 venue — Action required

[Yuval Levy]

Hi Louis,

>> *2* Evaluation of proposals
>>
>> the above said, looking at the three proposals and given the information
>> provided in the proposals, my pecking order would be:
>> 1. Canada
>> 2. Vietnam
>> 3. Brazil
>>
>> It was a difficult choice - I feel I am lacking some critical
>> information,
>>
>
> Like we all do. This is why we insist so much on having this info on the
> table.

Which info, and from which source?  I had a little bit of more time to
think about (and will pay for it tomorrow morning, sleep deprived).


>> and of course I am biased.
>
>
> Who isn’t?

we should put together a framework that set bias in perspective.


>> @Brazil: how much for a hotel room?  how much for a meal?  how much for
>> a
>> flight from Europe? from Asia? from North America ?  how would we be
>> clustered locally / distance between hotels and conference? local
>> traffic?
>>  I feel uncomfortable with the not yet secured venue.
>>
>> @Vietnam: realistically, do you think we'll sleep in the 10-15$/night
>> place, or in the 100$/night place in "district 1"?  I don't want to
>> guess,
>> I do want "developed country" level budget comfort (e.g. no
>> cockroaches).
>> Allergy from airco does not bode well for health... and having a long
>> commute from the hotels to the conference venue does not appeal to me.
>
>
>> @Canada: OK, you've hosted LGM twice, you have the advantage of "better
>> the devil you know" from my perspective.
>>
>
> Let’s keep all those comments for the next round, once we have all the
> informations on the table.

Rethinking this, my comments are a push-back on the proposals because I
feel they don't give enough information.

I would like to have a quantitative model for the "distance cost" as part
of the guidelines.  "Distance cost" is IMHO the most significant factor in
the decision.  A quant model can abstract and do away much of the
subjectivity.

For that model, I need global and local input.  Local input comes from the
proposals.

Global input: what is the potential pool of traveler-participants from
each region of the world?  My approximation for this would be the number
of active developers in each region because I assume that artists are less
inclined to travel to a conference.

Can we have a single list of regions? how detailed must it be?

can we live with:
- North America
- Central America
- South America
- Europe
- Africa
- Asia
- Australia/NZ

or do we need a more detailed one? regions should be built around "travel
clusters" - so maybe it makes sense to distinguish between East and West
Coast in North-America (big different for Pacific/Atlantic travel); and
between Sub-Saharan Africa and North-Africa); and between Northern and
Southern Europe; etc.

Can we have an idea of how many devs live in each of these regions?  Can
each project, or at least the largest projects, do a quick head count and
quantify their contributors and split them geographically?  Maybe with two
separate categories for "core contributors" and "other contributors"?  Of
course there will still be some subjectivity (e.g. what qualifies for an
active dev), but we need some estimates to start with, and we can keep the
data and refine it as we go on.  Note: developer in this context
encompasses translator, document-writer, etc... all roles that contribute
to the general development of the project, not just software developers.

Local input:  The minimum I would expect to see, from each bid is:
- flight cost per traveler from each of these regions
- lodging cost per traveler
- food cost per traveler
- local transport cost per traveler

all expressed in a single currency to be comparable. USD.

I want this summarized into a single number per region of origin.  My
critique of the Vietnam proposal is that I can not discern whether the
15$/hotel will be up to standard or not.

Of course we need to set a global standard, a minimum of what we would
expect.  Lodging cost should be for a double room, with bathroom and
decent hygienic standards.  Sorry for picking up on Vietnam again: airco
allergy is not a decent hygienic standard since it means that the airco's
filters are not serviced often enough.  If individual participants decide
to cut corners at their own risk and reduce expenses, that's a personal
choice.  For the budget, I want numbers that are realistic for the average
traveler, and I can't discern if the 15$/night hotel in Vietnam is such a
thing.

Once we have the numbers, we can do the math.

I would like to add a weighting factor to the math:
*distance-inconvenience*. Bear with me and the explanation will follow
below.

For each region of the world, its distance-inconvenience factor express
how inconvenienced people from that region were at attending LGM.

Let y be the number of years elapsed since a specific LGM (e.g. y=1 for
LGM2009 Montreal), and let Kry be the distance, in Km, from the center of
a region r to the LGM location y years ago.  The distance-inconvenience
factor Dr is the sum of all Kry/y since LGM's inception.

We want to bring LGM to a region with high Dr to share fairly the burden
of travel (should be stated in the guidelines, IMHO).

For every destination d, and for every origin of travel region r, I would
like to see a single number, in USD, Vrd summarizing total travel cost for
an average visitor, from region r to destination d.

Now we can calculate for every bid a potential maximum travel cost by
multiplying Vrd with the number of potential traveler-participant from
each region (Pr).  We want to keep the total travel burden low, i.e.
minimize Vrd*Pr

Next comes some (unavoidable) subjectivity.  What is the estimated
probability that a potential participant from region r will actually
participate at an event at destination d?  We can only guesstimate this
probability Xrd.  But it is important to do so because we want to maximize
participation, i.e. the sum of Xrd*Pr

To make it fair, we want to maximize weighted participation, i.e. the sum
of Xrd*Pr*Dr

If you managed to follow me until here, thank you for your patience and
congratulation.  We should now have a set of comparable numbers, for each
bid, to express their distance-incovenience adjusted travel cost.

Hard cost numbers can be sorted and ranked easily, but are not the only
factor to consider.  The other side of the numbers is revenue.  How much
sponsorship revenues stands against the likely potential travel cost of
Vrd*Xrd*Br ?

Having a quant model for this part of the decision takes some subjectivity
away and enables us to consciously minimize total travel impact while
dispersing cost as fairly as possible over the different regions of the
world.

Still, softer, subjective factors remain to consider, such as the
definition of standards for travel, lodging, food, local transportation.

Sorry Brazil, your proposal as I read it on the Wiki page when I was asked
to read was superficial and incomplete.  I don't want to collect
information from any other place than from a complete and submitted
proposal on the wiki.  Go back and do your homework.  As me to read and
judge a bid on a wiki page when it is complete.

Congratulations Vietnam, your proposal has a lot of information.  The
issue I have when reading your proposal is that I can not make the call as
to whether the "district 1" 100$/night hotel is necessary or whether the
15$/night hotel is up to standard.  I think we should give you the
standard and you should judge if the hotels, restaurant, and local
transport are all up to those standard and what the cost are.  The choice
between you and Canada was the difficult one for me at this stage, given
the currently available information.

And for Montreal, sorry Louis. You have the advantage of being a known
(and very solid) value; which is also a disadvantage because people who
have seen 2009 perceive it as "this is Montreal", while in reality it was
a last minute herculean effort to salvage LGM from a catastrophe, and did
not give you the time to unfold all of your potential.

In any case, besides whatever I ranked, I support an LGM in any of the
three locations.  It is not the *where* that counts.  It is the *why*, the
*how*, the *who*.

Yuv