<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
<title></title>
</head>
<body>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
<title></title>
Larry,<br>
<br>
Thanks for the additional input and let me try to see if I can both
summarize some of the positions that have been expressed and maybe
identify a possible consensus, mostly along similar lines as what Mark
outlined in his last email.<br>
<br>
I don't think that anyone is disagreeing with the potential benefits
associated with small, focused venues such as HotNets, and in
particular of the positive role that HotNets has played over the past
few years. In that respect, I believe that the concerns that Scott
expressed in his last email are not the issue.<br>
<br>
The main question is how to reconcile the need for keeping the audience
sufficiently small to be conducive to interactive and productive
exchanges, with the need to be "open" regarding who has (direct) access
to the benefits associated with participating in those exchanges. The
latter being also key in broadening our community and bringing in new
people.<br>
<br>
This, as you point out, is probably largely an issue of principle, and
although I wont really follow you in your argument that "private
meetings will happen anyhow because of their intrinsic value, so lets
make the best of it by ensuring that they are as fair as possible", I
also don't think that things are completely black and white and there
should be room for a consensus middle-ground. <br>
<br>
The main sticking point is again that of openness of attendance, but
I'll be
the first to agree that openness itself does not have an unambiguous
definition. Charging a $500 admission fee rules out participating in
some conferences for many. Similarly, as has been pointed out, the
selection of a given location will open attendance to some and close it
for others, etc. So the question is where to draw the line to on one
hand keep the overall audience small enough, and on the other hand open
attendance enough so as to give "outsiders" a reasonable chance of
getting in.<br>
<br>
I think there is some consensus that authors of accepted papers and PC
members should be given priority in terms of attendance. So the
question is then how many more slots can be accommodated after that,
how are they to be allocated, and what is the threshold/process that
would
meet the goal of sufficient openness?<br>
<br>
I don't think it would be productive to go into bean counting
specifications regarding the number of "open" slots, but it might be
possible to generate some rough guidelines on what would be a
reasonable percentage and how these could be allocated. There have
been a number of proposals put forward to accomplish this, but my own
preference is to go with something with the least number of control
knobs, e.g., 30% of open slots that are allocated on a fcfs basis or
any other scheme that is as transparent as possible. I actually don't
think that it is that meaningful or even practical to tightly
control/monitor a single instance of a venue, and instead I think that
Mark's "sliding window" approach that monitors attendance diversity
over a number of years is the better way to go about it (I must admit
having also been surprised by the number you gave regarding HotNets
attendance - 145 unique individuals out of 181 for the last 3 years -
which is really great). In that respect, I very much like the longer
term perspective of his proposal. I think that guidelines that on one
hand call for some partial opening of the attendance criteria of a
"closed" venue, e.g., having a certain percentage of open slots, and
more importantly enforce a monitoring of the actual attendance
diversity of a period of several years can represent a reasonable
approach and compromise. <br>
<br>
Obviously, there are some mechanical details that need to be addressed,
e.g., Mark's suggestion for a SC, and there has to be a commitment to
enforce the guidelines and be willing to pull the plug if necessary,
but this might be a solution that satisfies the principle of openness
and accessibility to the larger community, while preserving the ability
to keep audiences small enough to allow interactive discussions.<br>
<br>
Roch<br>
<blockquote
cite="mid6A973701-28AF-4FD9-B3D1-D3412DDD73D6@CS.Princeton.EDU"
type="cite">You've done a good job of articulating the issues. <br>
We can debate: <br>
<br>
o value of closed (in absolute terms) <br>
o value of closed (in terms of ROI to SIGCOMM) <br>
o if closed, how do we select attendees <br>
o if closed, how many people is too many <br>
<br>
I think you're pointing to the 2nd issue as the one on which <br>
we have a disagreement (with the caveat that the ROI is higher <br>
during the first couple of years, so maybe such events are worth <br>
sponsoring as they start). <br>
<br>
However, I believe SIGCOMM's investment in HotNets is pretty <br>
small. It was larger when SIGCOMM helped jump start the workshop <br>
by publishing the proceedings, but that's no longer the case. <br>
<br>
Does this then boil down to an issue of principle? Maybe... <br>
>From one perspective, framing the issue as one of "fairness <br>
and openness" certainly leads to the position you're taking. <br>
>From another perspective, "fostering scientific dialogue" <br>
seems to be the tag line I'm advocating. <br>
<br>
Let me try the following argument. SIGCOMM can't prohibit <br>
private meetings among researchers with common interests. <br>
They happen because they have value, and quite frankly, we <br>
all we to be invited to those meetings because we know their <br>
tremendous value. HotNets is an opportunity for SIGCOMM to <br>
sponsor such a meeting with a much greater degree of fairness <br>
and equal-opportunity than the best of these private meetings. <br>
A much broader and changing set of people get to participate, <br>
and budgets willing, the proceedings might even be published. <br>
My view is that there's room under SIGCOMM's tent to support <br>
a limited number of such events. <br>
<br>
Larry
</blockquote>
<br>
</body>
</html>