From vern at icir.org Mon Oct 24 23:22:12 2005 From: vern at icir.org (Vern Paxson) Date: Mon, 24 Oct 2005 23:22:12 -0700 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events Message-ID: <200510250622.j9P6MCI9056174@jaguar.icir.org> At the last SIGCOMM business meeting, one of the issues we framed for discussion by the SIGCOMM community concerns attendance policies at SIGCOMM-affiliated events such as HotNets. The main question is to what degree should such events be given latitude to limit their attendance along dimensions such as the following: - limiting size in order to facilitate discussion - given limited attendance, imposing criteria on who can attend, such as paper presenters / PC members / paper authors / paper submitters - filling some limited-attendance slots by invitation There was considerable discussion at the meeting of this topic, which I imperfectly summarize as: (1) A view by quite a few who commented at the microphone that limited attendance has utility for some events and that SIGCOMM should find ways to facilitate this. (2) A view by others (not as many at the microphone) that any policy other than first-come-first-serve is counter to the principles of fairness (both to individuals and in terms of how the SIG uses its resources) by which the SIG should abide. (3) A question as to whether smaller venues that benefit from closed attendance need SIG sponsorship anyway. Benefits of sponsorship include raising awareness of the event and providing a means/imprimatur for publishing proceedings for the event. Some questioned whether small events need proceedings; others view this as desirable as it makes publicly available the research ideas that went into the event. (4) The view that if attendance is closed, the invitation policy needs to be made clear. (5) Notions of pursuing hybrids in which some slots are left open to first-come-first-serve, and more generally with experimenting with different forms to see what works best. (6) Thoughts on how to "mitigate" the impact of an event being closed, such as by recording some of the discussion to make it available to those who were not able to attend (which on the other hand some viewed as likely to dampen the nature of the exchanges). We'd like to solicit further views from the community to get a sense of whether there's rough agreement on the best policy for the SIG to adopt. Please let us know your thoughts. - Vern, speaking as SIGCOMM vi-chair From jrex at CS.Princeton.EDU Tue Oct 25 06:36:08 2005 From: jrex at CS.Princeton.EDU (Jennifer Rexford) Date: Tue, 25 Oct 2005 09:36:08 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <200510250622.j9P6MCI9056174@jaguar.icir.org> Message-ID: <200510251336.j9PDaD9T015064@bluebox.CS.Princeton.EDU> Vern, Thanks for launching the discussion. I have a few thoughts to add, with the goal of encouraging more discussion. First, I think that, realistically, some events need limited attendance in their first couple of years, if for no other reason than the fact that it is hard to organize a venue when you don't know how big the audience will be. In addition, limited-attendance events help build the energy necessary for a sub-community to come together by facilitating more discussion, though frankly the size limits for really achieving this goal are pretty tight, like 30-40 people I would guess. I think many small "start up" events could easily forego SIG sponsorship if they have a natural way to contain cost -- having basically no fixed costs (e.g., a university meeting room, if a free one is available) and having all variable costs covered by registration fees -- and aren't worried that someone will sue them after a slip-and-fall accident... ;) That said, having the SIG be able to sponsor such start-up events might lower the barrier to starting new and exciting workshops and conferences by carrying the financial risk, arranging publication of proceedings (if needed), advertising the call for papers and the program, branding the event, etc. Creating new events is yeoman service to the community, and we should do what we can to make it easier for dedicated volunteers to do their jobs, which includes sponsoring new events that are trying to gauge community interest or fill a previously-overlooked need. So, personally, I think it would be nice for the SIG to support limited attendance events, at least for the first year or two. That said, I think a transparent attendance policy would be crucial, given the goals the SIG has for fairness and openness. (As someone mentioned at the business meeting, event organizers always have an option of foregoing SIG sponsorship if they want complete autonomy in determining attendance policies.) The word "transparent" might not be strong enough, but I hesitate to suggest exactly what the policy should be, as it may depend on the event. Clearly, you want at least one author for each paper, and the program and steering committees, to be able to attend. Beyond that, do you favor other authors of accepted papers? FIFO? Random? I don't know. Given that restricted attendance is motivated by the desire to encourage discussion, having many more people than authors, PC, and SC might too many anyway. Beyond a certain size, why not have open attendance anyway? However, once an event is successful enough to have a large community interest, and to sustain itself, I personally think it makes sense to have open attendance go hand in hand with SIG sponsorship. In some cases, this might argue for some changes in the nature of the event. For example, the Internet Meaurement Workshop (which had closed attendance the first couple of years, due to venue size constraints the first year and a desire to keep the event a "workshop" with lots of discussion) ultimately became an open-attenance Conference once it became clear that we had a large subcommunity with lots of mature work on our hands. Once an event is off the ground, I think it should take on a life of its own, and adapt to the community needs as best it can. After the initial "burn in" period, I believe that the goal of the SIG to serve its larger community trumps the goals achieved by limited attendance. Anyway, that's my 0.02 Euro... -- Jen -----Original Message----- From: sigcomm-bounces at postel.org [mailto:sigcomm-bounces at postel.org] On Behalf Of Vern Paxson Sent: Tuesday, October 25, 2005 1:22 AM To: sigcomm at postel.org Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events At the last SIGCOMM business meeting, one of the issues we framed for discussion by the SIGCOMM community concerns attendance policies at SIGCOMM-affiliated events such as HotNets. The main question is to what degree should such events be given latitude to limit their attendance along dimensions such as the following: - limiting size in order to facilitate discussion - given limited attendance, imposing criteria on who can attend, such as paper presenters / PC members / paper authors / paper submitters - filling some limited-attendance slots by invitation There was considerable discussion at the meeting of this topic, which I imperfectly summarize as: (1) A view by quite a few who commented at the microphone that limited attendance has utility for some events and that SIGCOMM should find ways to facilitate this. (2) A view by others (not as many at the microphone) that any policy other than first-come-first-serve is counter to the principles of fairness (both to individuals and in terms of how the SIG uses its resources) by which the SIG should abide. (3) A question as to whether smaller venues that benefit from closed attendance need SIG sponsorship anyway. Benefits of sponsorship include raising awareness of the event and providing a means/imprimatur for publishing proceedings for the event. Some questioned whether small events need proceedings; others view this as desirable as it makes publicly available the research ideas that went into the event. (4) The view that if attendance is closed, the invitation policy needs to be made clear. (5) Notions of pursuing hybrids in which some slots are left open to first-come-first-serve, and more generally with experimenting with different forms to see what works best. (6) Thoughts on how to "mitigate" the impact of an event being closed, such as by recording some of the discussion to make it available to those who were not able to attend (which on the other hand some viewed as likely to dampen the nature of the exchanges). We'd like to solicit further views from the community to get a sense of whether there's rough agreement on the best policy for the SIG to adopt. Please let us know your thoughts. - Vern, speaking as SIGCOMM vi-chair _______________________________________________ sigcomm mailing list sigcomm at postel.org http://www.postel.org/mailman/listinfo/sigcomm From llp at CS.Princeton.EDU Tue Oct 25 07:29:52 2005 From: llp at CS.Princeton.EDU (Larry Peterson) Date: Tue, 25 Oct 2005 10:29:52 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <200510251336.j9PDaD9T015064@bluebox.CS.Princeton.EDU> References: <200510251336.j9PDaD9T015064@bluebox.CS.Princeton.EDU> Message-ID: Let me try to say a couple things about HotNets, which I understand is at the heart of this discussion. The history is that HotNets was created through a grass-roots effort, outside the the SIGCOMM umbrella. To help lower the barrier-to-entry, we asked SIGCOMM to sponsor, so we could leverage existing know-how and infrastructure. Thus, I strongly endorse Jen's observation that SIGCOMM can play a role in helping get a new workshop off the ground. I also strongly believe that there is a valuable place for workshops with limited attendance. I don't think anyone is disputing that point. As a practical matter, once you invite one author per paper, the PC and SC, and as many students as you find room for (typically co-authors on accepted papers and co-authors on rejected papers), you've pretty much filled out the workshop. My recollection is that the number of invitations to play with is quite small, and usually go to a non-student co-author. So, in my view, it seems to come down to two alternatives: 1) HotNets continues with SIGCOMM sponsorship, remains closed, but has a more transparent admissions policy; or 2) HotNets leaves the SIGCOMM nest. Jen points to IMW as an example of a workshop "growing up" but I don't think that applies to HotNets, which has to remain a workshop to maintain its value... as a place to publish early ideas and vision papers. HotNets can survive without SIGCOMM sponsorship, but I worry that the extra burden of running a workshop will make it harder to get people to volunteer. Regarding a transparent policy, I think we're pretty close to having one (see above). Said another way, the PC decides who attends based on the papers that are submitted. Larry On Oct 25, 2005, at 9:36 AM, Jennifer Rexford wrote: > Vern, > > Thanks for launching the discussion. I have a few thoughts to add, > with the > goal of encouraging more discussion. First, I think that, > realistically, > some events need limited attendance in their first couple of years, > if for > no other reason than the fact that it is hard to organize a venue > when you > don't know how big the audience will be. In addition, limited- > attendance > events help build the energy necessary for a sub-community to come > together > by facilitating more discussion, though frankly the size limits for > really > achieving this goal are pretty tight, like 30-40 people I would guess. > > I think many small "start up" events could easily forego SIG > sponsorship if > they have a natural way to contain cost -- having basically no > fixed costs > (e.g., a university meeting room, if a free one is available) and > having all > variable costs covered by registration fees -- and aren't worried that > someone will sue them after a slip-and-fall accident... ;) That said, > having the SIG be able to sponsor such start-up events might lower the > barrier to starting new and exciting workshops and conferences by > carrying > the financial risk, arranging publication of proceedings (if needed), > advertising the call for papers and the program, branding the > event, etc. > Creating new events is yeoman service to the community, and we > should do > what we can to make it easier for dedicated volunteers to do their > jobs, > which includes sponsoring new events that are trying to gauge > community > interest or fill a previously-overlooked need. > > So, personally, I think it would be nice for the SIG to support > limited > attendance events, at least for the first year or two. That said, > I think a > transparent attendance policy would be crucial, given the goals the > SIG has > for fairness and openness. (As someone mentioned at the business > meeting, > event organizers always have an option of foregoing SIG sponsorship > if they > want complete autonomy in determining attendance policies.) The word > "transparent" might not be strong enough, but I hesitate to suggest > exactly > what the policy should be, as it may depend on the event. Clearly, > you want > at least one author for each paper, and the program and steering > committees, > to be able to attend. Beyond that, do you favor other authors of > accepted > papers? FIFO? Random? I don't know. Given that restricted > attendance is > motivated by the desire to encourage discussion, having many more > people > than authors, PC, and SC might too many anyway. Beyond a certain > size, why > not have open attendance anyway? > > However, once an event is successful enough to have a large community > interest, and to sustain itself, I personally think it makes sense > to have > open attendance go hand in hand with SIG sponsorship. In some > cases, this > might argue for some changes in the nature of the event. For > example, the > Internet Meaurement Workshop (which had closed attendance the first > couple > of years, due to venue size constraints the first year and a desire > to keep > the event a "workshop" with lots of discussion) ultimately became an > open-attenance Conference once it became clear that we had a large > subcommunity with lots of mature work on our hands. Once an event > is off > the ground, I think it should take on a life of its own, and adapt > to the > community needs as best it can. After the initial "burn in" period, I > believe that the goal of the SIG to serve its larger community > trumps the > goals achieved by limited attendance. > > Anyway, that's my 0.02 Euro... > > -- Jen > > -----Original Message----- > From: sigcomm-bounces at postel.org [mailto:sigcomm- > bounces at postel.org] On > Behalf Of Vern Paxson > Sent: Tuesday, October 25, 2005 1:22 AM > To: sigcomm at postel.org > Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events > > At the last SIGCOMM business meeting, one of the issues we framed for > discussion by the SIGCOMM community concerns attendance policies at > SIGCOMM-affiliated events such as HotNets. The main question is to > what > degree should such events be given latitude to limit their attendance > along dimensions such as the following: > > - limiting size in order to facilitate discussion > - given limited attendance, imposing criteria on who can attend, > such as paper presenters / PC members / paper authors / paper > submitters > - filling some limited-attendance slots by invitation > > There was considerable discussion at the meeting of this topic, which > I imperfectly summarize as: > > (1) A view by quite a few who commented at the microphone that > limited attendance has utility for some events and that > SIGCOMM should find ways to facilitate this. > > (2) A view by others (not as many at the microphone) that any > policy other than first-come-first-serve is counter to the > principles of fairness (both to individuals and in terms of > how the SIG uses its resources) by which the SIG should abide. > > (3) A question as to whether smaller venues that benefit from > closed attendance need SIG sponsorship anyway. Benefits of > sponsorship include raising awareness of the event and > providing > a means/imprimatur for publishing proceedings for the event. > Some questioned whether small events need proceedings; others > view this as desirable as it makes publicly available the > research ideas that went into the event. > > (4) The view that if attendance is closed, the invitation policy > needs to be made clear. > > (5) Notions of pursuing hybrids in which some slots are left open > to first-come-first-serve, and more generally with > experimenting > with different forms to see what works best. > > (6) Thoughts on how to "mitigate" the impact of an event being > closed, such as by recording some of the discussion to make > it available to those who were not able to attend (which on > the > other hand some viewed as likely to dampen the nature of the > exchanges). > > We'd like to solicit further views from the community to get a > sense of > whether there's rough agreement on the best policy for the SIG to > adopt. > Please let us know your thoughts. > > - Vern, speaking as SIGCOMM vi-chair > _______________________________________________ > sigcomm mailing list > sigcomm at postel.org > http://www.postel.org/mailman/listinfo/sigcomm > > _______________________________________________ > sigcomm mailing list > sigcomm at postel.org > http://www.postel.org/mailman/listinfo/sigcomm > > From acaro at bbn.com Tue Oct 25 10:26:07 2005 From: acaro at bbn.com (Armando L. Caro, Jr.) Date: Tue, 25 Oct 2005 13:26:07 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: References: <200510251336.j9PDaD9T015064@bluebox.CS.Princeton.EDU> Message-ID: <435E6AAF.1020306@bbn.com> Larry Peterson wrote: > As a practical matter, once you invite one author per paper, the PC > and SC, and as many students as you find room for (typically co-authors > on accepted papers and co-authors on rejected papers), you've pretty > much filled out the workshop. My recollection is that the number of > invitations to play with is quite small, and usually go to a > non-student co-author. > Regarding a transparent policy, I think we're pretty close to > having one (see above). Said another way, the PC decides who > attends based on the papers that are submitted. I think inviting co-authors on accepted papers makes sense, but inviting co-authors on rejected papers may shift the problem. Anyone can submit a paper and have it rejected. How does the PC determine _which_ authors of rejected papers will be invited? Maybe this hasn't been a problem in the past, but it could come up. -- Armando www.armandocaro.net From douglis at acm.org Tue Oct 25 10:39:44 2005 From: douglis at acm.org (Fred Douglis) Date: Tue, 25 Oct 2005 13:39:44 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events Message-ID: <435E6DE0.1090601@acm.org> As a past organizer of both an IEEE-CS event with similar issues, and an independent workshop, I thought it might be worth speaking up. Apologies for the length. First, the IEEE-CS event. I was the general chair of WWOS-IV, the precursor to HOTOS. When I became chair of TCOS and IEEE-CS wanted a formal steering committee for all conferences, I was the SC chair for several years. As Larry has now pointed out, the example of IMW becoming IMC is very different, I think, from the HOTOS/HOTNETS/HOT* mold. HOTOS and several conferences like it are targeted for small audiences by design. IMW being limited because of venue size suggests a FCFS registration policy, with spaces set aside for authors and organizers. HOTOS would set aside spaces for authors and organizers too, but the difference was that after those spaces were set aside, there was just about nothing left. Back in 1993, when we did WWOS, someone pointed out that IEEE-CS laws precluded these sorts of biased limited attendance. So we could say there are only 50 attendees, but officially, we had to fill them FCFS and ignore who actually was presenting papers. A rather ludicrous way to run a small workshop, and after some debate, we basically ignored the official requirement and did the right thing. I have no idea whether IEEE-CS still officially precludes the authors-only style of workshop. It seems like Vern's note was suggesting that ACM SIGCOMM might decide as an organization to frown on it or support it, but without considering whether ACM itself has a policy on the matter. From the standpoint of SIGCOMM policy, I have two concrete recommendations: 1) Yes, you should allow HOTNETS to run indefinitely as a limited, by-invitation-only workshop, where invitations are mostly the result of submissions. But these should be the exception, not the rule. One workshop like this per SIG or TC seems quite reasonable. 2) Any policy on this should start at the grass-roots level, but then rather than having different policies for each SIG, I believe you should push for an ACM-wide agreement. Now for the flip side. I was the program chair of the Web Caching Workshop a couple of years ago, and wound up being its SC chair as well as it moved under the IEEE-CS Technical Committee on the Internet. I felt *very* strongly that the workshop should get the backing of a professional organization, after my role as PC chair turned into also hosting the event at my company (when SARS forced the relocation of it from China). The reason is simple: ACM/IEEE/USENIX/etc provide financial support, both in the event the conference loses money and in the event of some sort of liability. When WCW moved to IBM, only an additional donation from IBM in lieu of particpant registration fees kept it from not only losing money but burning up the leftover money from the previous conferences. (Note that we had not obtained in advance any official financial commitment to absorb these losses -- we weren't even planning to host the workshop. That was my mistake, due to so many years of conferences with IEEE backing.) But what if you run a bus tour (as we did) and something Very Bad happens? Obviously the charter company will be at risk, but in this day I can't imagine people not coming after anyone else who had a hand in it. If sponsored by ACM, then ACM gets to handle this with their insurance. If run by a haphazard consortium of universities or companies, this is less clear to me. I think the only alternative to a professional organization is a "company" chartered specifically to run the conference, a la IW3C2 for the WWW conference. I really think you want HOTNETS to stick to ACM. Or at least make sure that every volunteer knows that the "extra burden of running a workshop" (in Larry's words) includes very serious financial questions. Note, I'm not a lawyer, and I may be completely deluded about the risks involved, or lack thereof. If others have more confidence in these independent workshops, I'd love to hear from them. But I was very sorry to have gotten into that situation, and I advise against it. -- Fred Douglis IBM T.J. Watson Research Center fdouglis at us.ibm.com http://www.research.ibm.com/people/f/fdouglis/ From touch at ISI.EDU Tue Oct 25 11:35:20 2005 From: touch at ISI.EDU (Joe Touch) Date: Tue, 25 Oct 2005 11:35:20 -0700 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <435E6DE0.1090601@acm.org> References: <435E6DE0.1090601@acm.org> Message-ID: <435E7AE8.5040701@isi.edu> Fred Douglis wrote: ... > Now for the flip side. I was the program chair of the Web Caching > Workshop a couple of years ago, and wound up being its SC chair as > well as it moved under the IEEE-CS Technical Committee on the > Internet. I felt *very* strongly that the workshop should get the > backing of a professional organization, after my role as PC chair > turned into also hosting the event at my company (when SARS forced the > relocation of it from China). The reason is simple: > ACM/IEEE/USENIX/etc provide financial support, both in the event the > conference loses money and in the event of some sort of liability. This is a key issue. These meetings, at some level, use or occupy shared community resources. That's exactly why the IEEE requires open attendance, and the ACM should as well. Private meetings should be underwritten by private resources. As to risks, they can be mitigated, and this is often lost in the shuffle. I.e., cancel the bus tour if you don't get enough sign-ups. Or keep in mind that workshops don't always have bus tours, banquets, or the like. They are necessarily as 'risky' as the content they present, and should be correspondingly more modest. Joe From touch at ISI.EDU Tue Oct 25 11:48:37 2005 From: touch at ISI.EDU (Joe Touch) Date: Tue, 25 Oct 2005 11:48:37 -0700 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <200510250622.j9P6MCI9056174@jaguar.icir.org> References: <200510250622.j9P6MCI9056174@jaguar.icir.org> Message-ID: <435E7E05.7020005@isi.edu> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Some other comments: Vern Paxson wrote: ... > (1) A view by quite a few who commented at the microphone that > limited attendance has utility for some events and that > SIGCOMM should find ways to facilitate this. Limited attendance has two useful purposes: a- encouraging lively debate b- fitting a limited-space venue (b) can and should be handled by FCFS beyond the author list. Conferences aren't journals; making them more exclusive is counter to the ACM notion of a public venue for discourse (which is core to the ACM's charter), and further clouds the 'is a workshop a conference' or 'is a workshop a journal' debate (which is a different debate; we can have that one later). Larry said: > I also strongly believe that there is a valuable place for workshops > with limited attendance. I don't think anyone is disputing that > point. I will: (a) seems the key point in that regard, but it is well known that the breakpoint for the size of such groups is around 20-40, and most of the meetings we're talking about are around 80, so that point is moot. Larry also noted that there are two options: > 1) HotNets continues with SIGCOMM sponsorship, remains closed, > but has a more transparent admissions policy; or > > 2) HotNets leaves the SIGCOMM nest. There are others, i.e., that SIGCOMM requires open attendance and Hotnets complies. We shouldn't limit this debate based on whether HotNets will stay with the SIG. That seems too much like closed the attendance policy we're debating (i.e., "only people who agree we want to keep HotNets should participate in this debate"), and I'm in favor of _open discourse_. Joe -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (MingW32) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFDXn4FE5f5cImnZrsRAggcAJ9MzAY0oXVVNgWVip7yrjGgmuiX2ACffml0 6G/fqcdX4hQU89wppMwDA/U= =+p91 -----END PGP SIGNATURE----- From mallman at icir.org Tue Oct 25 11:55:29 2005 From: mallman at icir.org (Mark Allman) Date: Tue, 25 Oct 2005 14:55:29 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: Message-ID: <20051025185529.EE949371246@lawyers.icir.org> An embedded and charset-unspecified text was scrubbed... Name: not available Url: http://www.postel.org/pipermail/sigcomm/attachments/20051025/577eb548/attachment.ksh -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 185 bytes Desc: not available Url : http://www.postel.org/pipermail/sigcomm/attachments/20051025/577eb548/attachment.bin From craig at aland.bbn.com Tue Oct 25 12:17:47 2005 From: craig at aland.bbn.com (Craig Partridge) Date: Tue, 25 Oct 2005 15:17:47 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: Your message of "Tue, 25 Oct 2005 12:00:01 PDT." Message-ID: <20051025191747.B1C6E1FF@aland.bbn.com> Joe Touch notes: >(a) seems the key point in that regard, but it is well known that the >breakpoint for the size of such groups is around 20-40, and most of the >meetings we're talking about are around 80, so that point is moot. Since we're in the exploring possibilities phase, a small point. You can keep a group lively as it grows -- but it requires inviting people known to be willing to provoke discussion. Eventually the process breaks and I don't know if it breaks at 60 or 80 or 100 people. However, this invitation of select gadflies/commentators/etc creates another fairness issue -- how do you identify these folks properly and justify giving them a seat to the exclusion of someone else (who perhaps, might also prove a gadfly but lacks the reputation)? Finally, a comment on the larger issue. It seems to me that the issue is whether the benefit *to the SIGCOMM community* of having a closed workshop is large enough to exclude some parties who wish to attend the workshop. My view is that, provided the workshop produces valuable results which are disseminated (via proceedings and other means), it is worth it. Everyone sees the proceedings (and only a small group was going to be able to afford to attend the workshop). But we need to hold the workshop to a high standard of results. Craig From douglis at acm.org Tue Oct 25 13:07:28 2005 From: douglis at acm.org (Fred Douglis) Date: Tue, 25 Oct 2005 16:07:28 -0400 Subject: [sigcomm] comments from the WWOS (HOTOS) founder Message-ID: <435E9080.8020501@acm.org> I pointed Joe Boykin, who started HOTOS when it was called WWOS, at the SIGCOMM discussion. He replied as follows, and granted permission for me to forward to the list, as he's not a subscriber. I'm ccing him here. I thought his comments are very helpful: --- I started WWOS and ran the first bunch of them and most definitely helped push others through, so I guess I'm not unbiased. Personally, I think there are 3 different issues here: 1) What size? A 'Workshop' is a small (50-75) person event designed with the goal of interaction. Hence, attendees should not be looking for a lecture, but a discussion among colleagues. Workshops are single-topic, indeed, a single topic pointed at a particular part of a larger topic. A 'Conference' is a presentation/lecture type of setting, typically 100+. People attend from "all over". They often are not experts in the particular material being discussed. Topics are varied. A 'Symposium' is a cross between a Workshop and a Conference. Mostly, that means a semi-narrowly focused conference, but not a lot of interaction. 2) Who gets to attend, and by what rules? WWOS and HotOS ran by the following set of rules: 60 People max (well, we sometimes went over by a bit); Presenters of papers, Conference Organizers, PC members were all automatically in. After that, people who wished to attend needed to submit a "position paper". If we had more people who wanted to attend than slots, we chose "intelligently" to get a mixture of people from across the board. Note that you said the CS required FCFS; memory says that was not true for Workshops (although it is true for Symposia and Conferences). We were not allowed to discriminate for any reason, but we *could* 'balance' attendance. 3) Who sponsors it? First, small events often lose money, so getting sponsorship is a 'good thing'. Second, having a 'professional' organization behind it means you do get conference insurance -- a good thing these days. You also get things like credit card processing and the like. You also get help (if needed) negotiating contracts -- let me tell you, hotels *will* try to rip off the uninitiated!. And last, you (usually) get a technical conference, not a marketing event. If you go the corporate only route, things often get muddy. Of course, it all depends on the conference organizers, but as it *always* happens, different people come along wtih different agendas. It sounds like the discussion is mixing the 3. From touch at ISI.EDU Tue Oct 25 13:08:04 2005 From: touch at ISI.EDU (Joe Touch) Date: Tue, 25 Oct 2005 13:08:04 -0700 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <20051025191747.B1C6E1FF@aland.bbn.com> References: <20051025191747.B1C6E1FF@aland.bbn.com> Message-ID: <435E90A4.2030601@isi.edu> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Craig Partridge wrote: > Joe Touch notes: > > >>(a) seems the key point in that regard, but it is well known that the >>breakpoint for the size of such groups is around 20-40, and most of the >>meetings we're talking about are around 80, so that point is moot. > > Since we're in the exploring possibilities phase, a small point. > > You can keep a group lively as it grows -- but it requires inviting > people known to be willing to provoke discussion. Eventually the process > breaks and I don't know if it breaks at 60 or 80 or 100 people. That might also be facilitated by discounted registration. If you really want gadflies there, attract them (this can be incorporated into the budget fairly easily if it's limited). Joe -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (MingW32) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFDXpCkE5f5cImnZrsRAnNBAJsGjYfN99z5NeHDUoad2IvL4JFABwCgq3yH KaPLt7KAONC3iyW3TZ/rcTM= =uT1i -----END PGP SIGNATURE----- From craig at aland.bbn.com Tue Oct 25 13:09:48 2005 From: craig at aland.bbn.com (Craig Partridge) Date: Tue, 25 Oct 2005 16:09:48 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: Your message of "Tue, 25 Oct 2005 13:08:04 PDT." <435E90A4.2030601@isi.edu> Message-ID: <20051025200948.ADA361FF@aland.bbn.com> In message <435E90A4.2030601 at isi.edu>, Joe Touch writes: >> You can keep a group lively as it grows -- but it requires inviting >> people known to be willing to provoke discussion. Eventually the process >> breaks and I don't know if it breaks at 60 or 80 or 100 people. > >That might also be facilitated by discounted registration. If you really >want gadflies there, attract them (this can be incorporated into the >budget fairly easily if it's limited). Hi Joe: I'm afraid I don't understand the economic model here. Is the idea that gadflies are cheap -- and only come if the registration cost is low? Craig From touch at ISI.EDU Tue Oct 25 13:17:26 2005 From: touch at ISI.EDU (Joe Touch) Date: Tue, 25 Oct 2005 13:17:26 -0700 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <20051025200948.ADA361FF@aland.bbn.com> References: <20051025200948.ADA361FF@aland.bbn.com> Message-ID: <435E92D6.5070301@isi.edu> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Craig Partridge wrote: > In message <435E90A4.2030601 at isi.edu>, Joe Touch writes: > > >>>You can keep a group lively as it grows -- but it requires inviting >>>people known to be willing to provoke discussion. Eventually the process >>>breaks and I don't know if it breaks at 60 or 80 or 100 people. >> >>That might also be facilitated by discounted registration. If you really >>want gadflies there, attract them (this can be incorporated into the >>budget fairly easily if it's limited). > > Hi Joe: > > I'm afraid I don't understand the economic model here. Is the idea that > gadflies are cheap -- and only come if the registration cost is low? > > Craig Just exploring ways to get them to come - discounted registration, subsidized travel, etc. I'm not sure about invited papers - those tend to work well for some participants, but not others. The point is that getting gadflies to attend depends on attracting gadflies, not as much on keeping a meeting 'closed'. Joe -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (MingW32) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFDXpLWE5f5cImnZrsRAjqDAKC1LR5bxYMImH7IdtQYoPqJatTHhQCfXOB8 Yyf5xy7kwUscoe3SpIkkf3k= =TgRD -----END PGP SIGNATURE----- From vern at icir.org Tue Oct 25 22:26:29 2005 From: vern at icir.org (Vern Paxson) Date: Tue, 25 Oct 2005 22:26:29 -0700 Subject: [sigcomm] comments from the WWOS (HOTOS) founder In-Reply-To: Your message of Tue, 25 Oct 2005 16:07:28 EDT. Message-ID: <200510260526.j9Q5QTlH053434@jaguar.icir.org> Thanks, a good framing of issues. FWIW: > ... It sounds like the discussion is mixing the 3. I hope it isn't; at least, that wasn't my intent. The intent was to assume (1) what size = small enough that having closed attendance is relevant (so likely only Workshops, as Joe scopes things); (3) who sponsors = either SIGCOMM or not SIGCOMM; with the goal then being to resolve a SIG policy for: (2) Who gets to attend, and by what rules? Vern From llp at CS.Princeton.EDU Wed Oct 26 02:16:33 2005 From: llp at CS.Princeton.EDU (Larry Peterson) Date: Wed, 26 Oct 2005 05:16:33 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <20051025185529.EE949371246@lawyers.icir.org> References: <20051025185529.EE949371246@lawyers.icir.org> Message-ID: I didn't mean to cut off debate, but just to focus it on what I saw to be the two alternatives. You're right that people could have differences of opinion in reaching the two alternatives I suggest, but the main issue in that chain that I see is whether there is value in a limited attendance workshop. I don't know that I've seen anyone argue against that position. (Arguing that all SIG events should be open is a different position, and one that I certainly recognize). Larry On Oct 25, 2005, at 2:55 PM, Mark Allman wrote: > > Larry- > > Vern asked a set of good questions. We should discuss them. This > note > is not part of such a discussion. You started it. > > This: > > > > >> So, in my view, it seems to come down to two alternatives: >> >> 1) HotNets continues with SIGCOMM sponsorship, remains closed, >> but has a more transparent admissions policy; or >> >> 2) HotNets leaves the SIGCOMM nest. >> >> >> > > is not part of a useful discussion on the questions Vern asked. > Conferences / workshops / etc. that want to use the SIG umbrella > for the > benefits (which you clearly believe are present - from reading your > note) at least have the responsibility to engage in a discussion with > the community on issues like this, IMO. If the grass roots that pitch > ideas to the SIG cannot agree to that then I don't feel > particularly bad > about seeing the door hit 'em on the way out. > > Reasonable people can, of course, have philosophical differences > and one > could envision things coming to your (1) or (2). I would hope that > maybe we could all reason about these things and it would not come to > that. At least I would hope that we could *have* a discussion before > people start throwing down "my way or else" sorts of statements. > (Statements which are clearly ludicrous anyway ... as if there aren't > many very able people in this community who could drive a new SIGCOMM > workshop on "Hot Topics In Networking".) > > For instance, I'd be interested in hearing your reasoning behind this > statement ... > > > > >> I also strongly believe that there is a valuable place for workshops >> with limited attendance. >> >> >> > > That would be useful to me. (Not saying I agree or disagree ... just > that I'd like to have *that* conversation and I'd be quite > interested in > your thinking on the matter since you have been through all stages of > a number of such events.) > > allman > > > > > > From llp at CS.Princeton.EDU Wed Oct 26 02:23:01 2005 From: llp at CS.Princeton.EDU (Larry Peterson) Date: Wed, 26 Oct 2005 05:23:01 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <435E7AE8.5040701@isi.edu> References: <435E6DE0.1090601@acm.org> <435E7AE8.5040701@isi.edu> Message-ID: <2D3EDE30-2F36-416D-B012-5AE456456DB0@cs.princeton.edu> Joe, I disagree that it's a private meeting. The CFP is published widely, anyone can submit, and the these papers are the primary factor in deciding who can attend. Just to clarify another point (and I'm going from memory here), the only case I remember a decision being made to invite an author of a paper not accepted was to invite a student that had applied for a travel grant. BTW, I'm not claiming that the chairs have never invited anyone that was not a student or a co-author, just that the numbers are very small. I could make an argument that one of the PC and chair's obligations is to put together as interesting of program as possible, and one that has the most value to the authors (e.g., in terms of feedback). Having 4-5 invitations to use at their discretion is not an unreasonable tool to achieve that goal. Larry On Oct 25, 2005, at 2:35 PM, Joe Touch wrote: > > > Fred Douglis wrote: > ... > > > >> Now for the flip side. I was the program chair of the Web Caching >> Workshop a couple of years ago, and wound up being its SC chair as >> well as it moved under the IEEE-CS Technical Committee on the >> Internet. I felt *very* strongly that the workshop should get the >> backing of a professional organization, after my role as PC chair >> turned into also hosting the event at my company (when SARS forced >> the >> relocation of it from China). The reason is simple: >> ACM/IEEE/USENIX/etc provide financial support, both in the event the >> conference loses money and in the event of some sort of liability. >> >> >> > > This is a key issue. These meetings, at some level, use or occupy > shared > community resources. That's exactly why the IEEE requires open > attendance, and the ACM should as well. > > Private meetings should be underwritten by private resources. > > As to risks, they can be mitigated, and this is often lost in the > shuffle. I.e., cancel the bus tour if you don't get enough sign- > ups. Or > keep in mind that workshops don't always have bus tours, banquets, or > the like. They are necessarily as 'risky' as the content they present, > and should be correspondingly more modest. > > Joe > _______________________________________________ > sigcomm mailing list > sigcomm at postel.org > http://www.postel.org/mailman/listinfo/sigcomm > > > > From calvert at netlab.uky.edu Wed Oct 26 06:25:06 2005 From: calvert at netlab.uky.edu (Ken Calvert) Date: Wed, 26 Oct 2005 09:25:06 -0400 (EDT) Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: References: <20051025185529.EE949371246@lawyers.icir.org> Message-ID: I second Craig's point: > It seems to me that the issue is whether the benefit *to > the SIGCOMM community* of having a closed workshop is > large enough to exclude some parties who wish to attend > the workshop. Most participants in the discussion so far have been organizers/attendees of such meetings; it's not surprising that they would support SIG sponsorship. If any conclusions are going to be reached as a result of the current discussion, participation from a much wider cross-section of the SIGCOMM community is needed. (In fact the question could be answered pretty straightforwardly by a poll, at least for meetings that have a history: "Do you, as a SIGCOMM member, see enough value in [insert name of meeting] to justify SIG sponsorship, even though some members who want to attend will not be able to?") Anyway, here's another data point. I'm skeptical about limiting attendance for the sole purpose of "encouraging lively discussion". As others have suggested, once you have too many participants to sit around one table, a particular size is neither necessary nor sufficient for good discussion, and having an arbitrary limit just risks the perception of exclusivity (which I believe has been acknowledged as an issue for SIGCOMM in the past). I am somewhat more sympathetic with limiting numbers to reduce uncertainty in planning, or to meet venue constraints in some cases. Finally, I think the more important transparency question is one level up: What is SIGCOMM's policy on sponsorship of meetings in general? What criteria are used to decide whether SIGCOMM will sponsor a meeting? Who makes the decision? (A quick perusal of the web site found lots of advice for those wanting to organize meetings, but no policies, though I may have missed something.) This part of the process should be as open and transparent as possible, to avoid any appearance of exclusivity. [Full disclosure: I have (so far) not attended or submitted to any recent closed-attendance event. Once, a looong time ago, my registration for SOSP was rejected because attendance was limited and the big company I worked for at that time already had its share of attendees.] KC -- Ken Calvert, Associate Professor Lab for Advanced Networking calvert at netlab.uky.edu University of Kentucky Tel: +1.859.257.6745 Hardymon Building, 2nd Floor Fax: +1.859.323.1971 301 Rose Street http://www.cs.uky.edu/~calvert/ Lexington, KY 40506-0495 From minshall at acm.org Wed Oct 26 06:14:06 2005 From: minshall at acm.org (Greg Minshall) Date: Wed, 26 Oct 2005 15:14:06 +0200 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <435E7AE8.5040701@isi.edu> References: <435E6DE0.1090601@acm.org> <435E7AE8.5040701@isi.edu> Message-ID: <7812-SnapperMsg8AEA1AA7BF85329C@[10.197.70.197]> my thoughts: a) one of SIGCOMM's roles is to support the networking research community. (we do that, for example, by providing access to SIGCOMM publications in the ACM Digital Library free of charge). the "networking research community" overlaps with SIGCOMM membership. b) closed, small, workshops are very important for the research community. (i've seen "small" up to 60 or so people.) (therefore, I support SIGCOMM supporting some by-invitation-only workshops.) c) in terms of who gets invited, i'd leave it up to the organizing committee, but *my* default would be: authors of accepted papers (ambiguity of primary or all secondary); PC; primary authors of rejected papers, assuming the rejected paper had a clue; other invitees (this last class to make up < 10 % of the total attendees). (somebody mentioned "all the students you can squeeze in", which I reject: hopefully, students will be authors of papers.) cheers, Greg p.s., years ago Scott Brim of Cisco was ruminating about the IETF, now versus then, and said something like, "before, we tried to do what was good, and now, we try to do what is 'right'; the latter is much harder." From guerin at ee.upenn.edu Wed Oct 26 07:46:29 2005 From: guerin at ee.upenn.edu (Roch Guerin) Date: Wed, 26 Oct 2005 10:46:29 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <7812-SnapperMsg8AEA1AA7BF85329C@[10.197.70.197]> References: <435E6DE0.1090601@acm.org> <435E7AE8.5040701@isi.edu> <7812-SnapperMsg8AEA1AA7BF85329C@[10.197.70.197]> Message-ID: <435F96C5.2000204@ee.upenn.edu> My 2c in part in reaction to Greg's email on the topic, but also based on all the input provided so far. >a) one of SIGCOMM's roles is to support the networking research community. > (we do that, for example, by providing access to SIGCOMM publications in >the ACM Digital Library free of charge). the "networking research >community" overlaps with SIGCOMM membership. > > Absolutely. >b) closed, small, workshops are very important for the research community. > (i've seen "small" up to 60 or so people.) > > Yes again, BUT >(therefore, I support SIGCOMM supporting some by-invitation-only workshops.) > > I view this role as taking the form of "nurturing" seed projects so that we make it easier for people to explore new directions that ultimately will be benefit the SIGCOMM community at large. And this does typically require small, focused group and is, therefore, in my mind a good justification for closed workshops. The point though is that this should NOT be a steady-state situation, and only meant to lower the barrier to entry in getting things started, i.e., provide focused forums to explore new directions and avoid doing so while overly burdening volunteers trying to get these started. This should NOT extent to long-running event. In my opinion, those should either grow-up and learn to fly on their own or go-away. I just don't buy into the premises that publishing the proceedings of a closed event represents sufficient value to the community. So more specifically, in the case of HOTNETS, I believe it needs to either open-up or gather enough direct support from other sources to continue operating under the current model. >c) in terms of who gets invited, i'd leave it up to the organizing >committee, but *my* default would be: authors of accepted papers (ambiguity >of primary or all secondary); PC; primary authors of rejected papers, >assuming the rejected paper had a clue; other invitees (this last class to >make up < 10 % of the total attendees). (somebody mentioned "all the >students you can squeeze in", which I reject: hopefully, students will be >authors of papers.) > > That's again perfectly OK as part of a startup phase. I don't consider this justified in the context of events that repeat regularly. My 2c. Roch From mallman at icir.org Wed Oct 26 08:16:25 2005 From: mallman at icir.org (Mark Allman) Date: Wed, 26 Oct 2005 11:16:25 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <20051025191747.B1C6E1FF@aland.bbn.com> Message-ID: <20051026151625.2D583371AE4@lawyers.icir.org> An embedded and charset-unspecified text was scrubbed... Name: not available Url: http://www.postel.org/pipermail/sigcomm/attachments/20051026/58f2592b/attachment.ksh -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 185 bytes Desc: not available Url : http://www.postel.org/pipermail/sigcomm/attachments/20051026/58f2592b/attachment.bin From jrex at CS.Princeton.EDU Wed Oct 26 08:17:39 2005 From: jrex at CS.Princeton.EDU (Jennifer Rexford) Date: Wed, 26 Oct 2005 11:17:39 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: Message-ID: <200510261517.j9QFHeMT012678@bluebox.CS.Princeton.EDU> Ken, > Finally, I think the more important transparency question is > one level up: What is SIGCOMM's policy on sponsorship of > meetings in general? What criteria are used to decide > whether SIGCOMM will sponsor a meeting? Who makes the > decision? That's a great question, and we should make the policies more readily available online from the main SIGCOMM Web site. Let me give some context, and then explain how things work. My warning in advance for a long note, in the interest of completeness. First of all, there are two categories of SIGCOMM involvement, as defined by the ACM: "sponsorship" (which implies financial involvement) and "in cooperation with" (which implies endorsement and technical involvement). For example, SIGCOMM is the sole sponsor of the main SIGCOMM conference, HotNets, and IMC, and a co-sponsor of SenSys and ANCS along with other SIGs (and some IEEE groups, in the case of ANCS). IMC is "in cooperation with USENIX". CoNEXT is "in cooperation with SIGCOMM." Second, new events arise through grass-roots efforts. A small group of folks conceive of an event that fills a need and comes to SIGCOMM asking for sponsorship, or co-sponsorship, or "in cooperation with" status. Then, the SIG executive committee reviews the proposal. We discuss, and weigh the need for such an event, the likelihood of success, etc. We also have a bunch of other criteria, like: - conference shouldn't overlap significantly in content with an existing SIGCOMM-sponsored conference - conference shouldn't overlap significantly in the dates with a related SIGCOMM conference (especially the annual SIGCOMM conference) - topic should be of interest to SIGCOMM members, and some SIGCOMM kinds of people would be on the program committee - quality is high enough that we'd want to be associated with it - conference makes the accepted papers freely available online - conference ideally would give a registration discount to ACM/SIGCOMM members, and have reasonable registration fees in general - we learn about the event early enough to advertise the CFP to our members We've been trying to limit ourselves in the number of new sponsored events so things don't become unmanageable, either logistically or financially. (For example, many SIGs, including SIGCOMM, lost a fair amount of money in the year or two after the dot-com bust and 9/11, making all SIGs concerned about controlling financial risk. Plus, ACM has explicit requirements on how much money we need to maintain in reserve as a fraction of our total conference and member-related expenses, which also limits the number of new events we can realistically take on.) During the past five or six years, we've been taking on one (or at most two) per year, in the interest of serving a growing community and embracing new sub-specialties (e.g., measurement, sensornets, network hardware, etc.). So, we've typically focused on "fat topics" rather than narrow topics -- otherwise, we just couldn't reasonably contain the number of events we're sponsoring. For each sponsored event, there is a steering committee that takes ownership for running the event, putting together the budget, picking locations and venues, picking the PC chairs, etc. Our goal has been to give these steering committees considerable autonomy to give each event its own local flavor, with the SIG and ACM providing guidance and assistance around budget, Web site, formatting and printing the proceedings, registration, insurance, and stepping in when extra help is needed. With so many new events created in the past few years, we're starting to realize the need to consider having some SIG-level policies (e.g., around attendance policies, reviewing guidelines, etc.). During the public SIG business meeting at SIGCOMM'05 in August, we had a bunch of discussion about what kinds of SIG-level policies make sense for our SIG-sponsored events, to have some "SIG values" preserved while still erring on the side of steering committee autonomy whenever possible. Hence, the current discussion. As an aside, the barrier for "in cooperation with" events is much lower than for sponsorship, since the SIG takes no financial risk and generally has much less work to do. Normally, SIGCOMM would do "in cooperation" where there would be mutual benefit. Meetings where we expect substantial SIG participation - or want to promote that. The meetings shouldn't overlap topically with an existing SIGCOMM-sponsored event, and the dates should not conflict either. The event should usually be sponsored by a fairly large organization as well -- ACM (another SIG), IEEE, or IFIP to name a few -- and certainly by a non-profit institution promoting research, rather than a commercial entity or trade show. Typically, these would be international events (rather than small regional workshops). We typically want to see the current or proposed program committee members to get a sense of the event, and for papers to be freely available online, and for the SIG/ACM members to be eligible for whatever member discounts might be made available for conference registration. In general, we've been a somewhat conservative in doing "in cooperation with" to avoid diluting or compromising the meaning of "in cooperation with SIGCOMM," or creating conflicts (in dates or topics) with existing SIGCOMM-sponsored events. -- Jen, with SIG Chair hat on From touch at ISI.EDU Wed Oct 26 10:04:41 2005 From: touch at ISI.EDU (Joe Touch) Date: Wed, 26 Oct 2005 10:04:41 -0700 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <2D3EDE30-2F36-416D-B012-5AE456456DB0@cs.princeton.edu> References: <435E6DE0.1090601@acm.org> <435E7AE8.5040701@isi.edu> <2D3EDE30-2F36-416D-B012-5AE456456DB0@cs.princeton.edu> Message-ID: <435FB729.5030805@isi.edu> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Larry Peterson wrote: > Joe, I disagree that it's a private meeting. The CFP is published > widely, anyone can submit, and the these papers are the primary > factor in deciding who can attend. Private meeting. Public proceedings, public call. We may be splitting hairs here, but private parties are often publicly known of a-priori and publicly reported post-facto. > Just to clarify another point (and I'm going from memory here), > the only case I remember a decision being made to invite an > author of a paper not accepted was to invite a student that had > applied for a travel grant. For HotNets-I, I was.* My notes from the meeting indicate a few others - my notes indicate 6 other names not on the PC or author list, i.e., 7 in total, and that's just the ones I could find. > BTW, I'm not claiming that the chairs have never invited anyone > that was not a student or a co-author, just that the numbers are > very small. I could make an argument that one of the PC and chair's > obligations is to put together as interesting of program as > possible, and one that has the most value to the authors (e.g., > in terms of feedback). Having 4-5 invitations to use at their > discretion is not an unreasonable tool to achieve that goal. As above, I have evidence of 7. There were another 7 on the PC, which are also chair-based invitations. That's 14 out of 65 (as reported in Jan 2003 CCR). That's 20%. Consider invited papers - 20% would be huge (5 papers at Sigcomm, or 50 at Infocom). Is this a tool to achieve a goal? Perhaps. The key question is whether it's a tool that a public group should allow at meetings they sponsor. Joe - -------------------- * FWIW, HotNets-I had other problems, notably that the reviews received were horribly incomplete (of 3 reviews, two gave only a summary rank with no ranks for the 6 component questions, and one had only the scores, with no substantiating text). We can debate that as another thread. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (MingW32) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFDX7cpE5f5cImnZrsRAuMZAJ975PXDN685U8Xce2z+XNmlzb/SnwCbBJO/ uDa7pI2qRN+ouFZ4jArbzcI= =EHmq -----END PGP SIGNATURE----- From llp at CS.Princeton.EDU Thu Oct 27 02:30:15 2005 From: llp at CS.Princeton.EDU (Larry Peterson) Date: Thu, 27 Oct 2005 05:30:15 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <435FB729.5030805@isi.edu> References: <435E6DE0.1090601@acm.org> <435E7AE8.5040701@isi.edu> <2D3EDE30-2F36-416D-B012-5AE456456DB0@cs.princeton.edu> <435FB729.5030805@isi.edu> Message-ID: This seems pretty important. I would be thrilled to see the proceedings published, augmented with notes about the discussion. This happened for HotNets-I. I'm pretty sure Hotnets-II papers were published in CCR, but I don't recall if there were any meeting notes. I'm pretty sure the HotNets-III papers were not published. My understanding is that this is an issue of budgets, and the extent to which SIGCOMM can help offset CCR costs. Perhaps others that know more about budgets/costs can comment. Larry On Oct 26, 2005, at 1:04 PM, Joe Touch wrote: > Larry Peterson wrote: > >> Joe, I disagree that it's a private meeting. The CFP is published >> widely, anyone can submit, and the these papers are the primary >> factor in deciding who can attend. >> > > Private meeting. Public proceedings, public call. We may be splitting > hairs here, but private parties are often publicly known of a- > priori and > publicly reported post-facto. From jrex at CS.Princeton.EDU Thu Oct 27 06:11:00 2005 From: jrex at CS.Princeton.EDU (Jennifer Rexford) Date: Thu, 27 Oct 2005 09:11:00 -0400 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: Message-ID: <200510271311.j9RDB1fr027617@bluebox.CS.Princeton.EDU> Larry, Hi, good question. The way this works now is that each conference/workshop decides whether to print a proceedings, and budgets the costs; they can decide whether to print just enough for the conference attendees (and some extra, e.g., for libraries, sale, etc.) or to mail to the SIG membership. The key issue is that each event breaks even (and budgets a small contingency to protect against uncertainty), and that we use SIG membership dues to pay for printing and mailing CCR; all members also get a copy of the main SIGCOMM proceedings. A number of years ago, we used to send *all* proceedings to *all* members, at a cost that made us lose money on each member -- forcing us to use conference revenues to subsidize membership benefits. That was fine, for a while... Then, as the number of conferences and workshops started to grow in the past five or six years, this was getting increasingly untenable, from a cost perspective. Plus, with papers increasingly available online (for free, because of a deal the SIG negotiated with ACM), mailing all proceedings to all members didn't seem like the best use of funds. We looked at increasing the membership dues to avoid losing money on a per member basis (and making it up in volume!), but the increase struck us as too large, and some conference proceedings just wouldn't be of interest to some members. And, we looked at having two classes of memberships -- with and without conference/workshop proceedings -- but the cost of printing a smaller run of each proceeding was higher on a per-proceeding basis, which almost completely offset the benefit. (Joe, feel free to chime in -- I recall your doing a bunch of this analysis a number of years ago.) Anyway, this is all a long-winded way of explaining why things work the way they do now, where each conference/workshop budgets the cost for printing its own proceedings, and mailing it (if they choose to mail it). The bottom line is that SIG doesn't really have much of a revenue stream: membership dues cover member benefits, and conferences/workshops generally break even, with the SIG covering loses when they happen and keeping profits (when they occur) in the bank to help when an event loses money, and occassionally to fund other things (like funding student travel grants, or printing and mailing the first couple of HotNets proceedings). That is, the decision to print the first couple of HotNets proceedings in CCR was done as a way to help bootstrap the new event, on the SIG's nickel rather than the workshop budget. We use surpluses to support new events (when they are at their most critical time), support students, and protect against future losses (to allow events we sponsor to take some risk). -- Jen -----Original Message----- From: sigcomm-bounces at postel.org [mailto:sigcomm-bounces at postel.org] On Behalf Of Larry Peterson Sent: Thursday, October 27, 2005 4:30 AM To: Joe Touch Cc: sigcomm at postel.org Subject: Re: [sigcomm] attendance policies for SIGCOMM-affiliated events This seems pretty important. I would be thrilled to see the proceedings published, augmented with notes about the discussion. This happened for HotNets-I. I'm pretty sure Hotnets-II papers were published in CCR, but I don't recall if there were any meeting notes. I'm pretty sure the HotNets-III papers were not published. My understanding is that this is an issue of budgets, and the extent to which SIGCOMM can help offset CCR costs. Perhaps others that know more about budgets/costs can comment. Larry On Oct 26, 2005, at 1:04 PM, Joe Touch wrote: > Larry Peterson wrote: > >> Joe, I disagree that it's a private meeting. The CFP is published >> widely, anyone can submit, and the these papers are the primary >> factor in deciding who can attend. >> > > Private meeting. Public proceedings, public call. We may be splitting > hairs here, but private parties are often publicly known of a- > priori and > publicly reported post-facto. _______________________________________________ sigcomm mailing list sigcomm at postel.org http://www.postel.org/mailman/listinfo/sigcomm From touch at ISI.EDU Thu Oct 27 07:42:48 2005 From: touch at ISI.EDU (Joe Touch) Date: Thu, 27 Oct 2005 07:42:48 -0700 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <200510271311.j9RDB1fr027617@bluebox.CS.Princeton.EDU> References: <200510271311.j9RDB1fr027617@bluebox.CS.Princeton.EDU> Message-ID: <4360E768.40209@isi.edu> Chiming in as requested below ;-) Jennifer Rexford wrote: > Larry, > > Hi, good question. The way this works now is that each conference/workshop > decides whether to print a proceedings, and budgets the costs; they can > decide whether to print just enough for the conference attendees (and some > extra, e.g., for libraries, sale, etc.) or to mail to the SIG membership. > The key issue is that each event breaks even (and budgets a small > contingency to protect against uncertainty), and that we use SIG membership > dues to pay for printing and mailing CCR; all members also get a copy of the > main SIGCOMM proceedings. A number of years ago, we used to send *all* > proceedings to *all* members, at a cost that made us lose money on each > member -- forcing us to use conference revenues to subsidize membership > benefits. That was fine, for a while... FYI, the cost per member is around $5 each for such proceedings, and the cost to the conference/workshop is around $15,000 to do this. > Then, as the number of conferences and workshops started to grow in the past > five or six years, this was getting increasingly untenable, from a cost > perspective. Since the SIG membership was in the $25 range, getting CCR, IMW, HotNets, Sigcomm, LANC, etc. ate up the entire membership revenue - and then some. Keep in mind that the SIG expenses roughly break even with CCR and Sigcomm being the only 'all member' publications, so these others threw us into debt to the tune of $50K a year or more. > Plus, with papers increasingly available online (for free, > because of a deal the SIG negotiated with ACM), mailing all proceedings to > all members didn't seem like the best use of funds. There are past studies that show that a paper copy increases citation rates substantially; I haven't heard any new info that refutes that. We would prefer to be able to distribute the copies that way, but it's not financially feasible. Sending everyone a CDROM doesn't have the same effect - it would come out once a year, and doesn't encourage casual browsing the way a paper copy still does. > We looked at increasing > the membership dues to avoid losing money on a per member basis (and making > it up in volume!), but the increase struck us as too large, and some > conference proceedings just wouldn't be of interest to some members. And, > we looked at having two classes of memberships -- with and without > conference/workshop proceedings -- but the cost of printing a smaller run of > each proceeding was higher on a per-proceeding basis, which almost > completely offset the benefit. (Joe, feel free to chime in -- I recall your > doing a bunch of this analysis a number of years ago.) There are costs: the cost to manage the list of who-gets-what, the additional cost per issue (over 2x for small runs), etc. Overall, per-meeting basis isn't feasible managerially, but 'all or ccr/sigcomm' might be. However, we'd be talking about an additional $10 per issue or more, on top of the current $35 SIG membership. That might mean an extra cost of around $50 or more. > Anyway, this is all a long-winded way of explaining why things work the way > they do now, where each conference/workshop budgets the cost for printing > its own proceedings, and mailing it (if they choose to mail it). The bottom > line is that SIG doesn't really have much of a revenue stream: membership > dues cover member benefits, and conferences/workshops generally break even, As a group - individually, some make, some lose. > with the SIG covering loses when they happen and keeping profits (when they > occur) in the bank to help when an event loses money, and occassionally to > fund other things (like funding student travel grants, or printing and > mailing the first couple of HotNets proceedings). That is, the decision to > print the first couple of HotNets proceedings in CCR was done as a way to > help bootstrap the new event, on the SIG's nickel rather than the workshop > budget. We use surpluses to support new events (when they are at their most > critical time), support students, and protect against future losses (to > allow events we sponsor to take some risk). > > -- Jen > > -----Original Message----- > From: sigcomm-bounces at postel.org [mailto:sigcomm-bounces at postel.org] On > Behalf Of Larry Peterson > Sent: Thursday, October 27, 2005 4:30 AM > To: Joe Touch > Cc: sigcomm at postel.org > Subject: Re: [sigcomm] attendance policies for SIGCOMM-affiliated events > > This seems pretty important. I would be thrilled to see the > proceedings published, augmented with notes about the discussion. > This happened for HotNets-I. I'm pretty sure Hotnets-II papers > were published in CCR, but I don't recall if there were any > meeting notes. I'm pretty sure the HotNets-III papers were not > published. > > My understanding is that this is an issue of budgets, and the > extent to which SIGCOMM can help offset CCR costs. Perhaps > others that know more about budgets/costs can comment. > > Larry > > On Oct 26, 2005, at 1:04 PM, Joe Touch wrote: > >> Larry Peterson wrote: >> >>> Joe, I disagree that it's a private meeting. The CFP is published >>> widely, anyone can submit, and the these papers are the primary >>> factor in deciding who can attend. >>> >> Private meeting. Public proceedings, public call. We may be splitting >> hairs here, but private parties are often publicly known of a- >> priori and >> publicly reported post-facto. > > _______________________________________________ > sigcomm mailing list > sigcomm at postel.org > http://www.postel.org/mailman/listinfo/sigcomm -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 250 bytes Desc: OpenPGP digital signature Url : http://www.postel.org/pipermail/sigcomm/attachments/20051027/05477ebd/signature.bin From christophe.diot at thomson.net Mon Oct 31 09:19:29 2005 From: christophe.diot at thomson.net (Christophe Diot) Date: Mon, 31 Oct 2005 18:19:29 +0100 Subject: [sigcomm] SIGCOMM 2006 -- Call for Workshops Message-ID: <43665221.6050502@thomson.net> ANNOUNCEMENT and CALL FOR PROPOSAL SIGCOMM 2006 will propose up to four one day workshops that will be scheduled Monday September 11 and Friday September 15, 2006, In Pisa (Italy). The SIGCOMM conference is co-located in Pisa September 12-14, 2006. We invite you to submit workshop proposals on any topic related to computer communication and packet networking before November 30th, 2005 to Christophe Diot and/or Serge Fdida. A workshop proposal should contain: - a draft call for paper (as complete as possible) - the workshop deadlines (internal and external) - tentative composition of the committees - motivation and rationale for the workshop, expected number of submissions and participants. - potential sponsors if any. - more if you think of something else we might like to know :-) we reserve the right to ask proposers more information later :-) Christophe Diot Serge Fdida From mallman at icir.org Mon Oct 31 09:21:29 2005 From: mallman at icir.org (Mark Allman) Date: Mon, 31 Oct 2005 12:21:29 -0500 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <4360E768.40209@isi.edu> Message-ID: <20051031172130.4A267375FA2@lawyers.icir.org> An embedded and charset-unspecified text was scrubbed... Name: not available Url: http://www.postel.org/pipermail/sigcomm/attachments/20051031/658e7e66/attachment.ksh -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 185 bytes Desc: not available Url : http://www.postel.org/pipermail/sigcomm/attachments/20051031/658e7e66/attachment.bin From touch at ISI.EDU Mon Oct 31 09:31:31 2005 From: touch at ISI.EDU (Joe Touch) Date: Mon, 31 Oct 2005 09:31:31 -0800 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <20051031172130.4A267375FA2@lawyers.icir.org> References: <20051031172130.4A267375FA2@lawyers.icir.org> Message-ID: <436654F3.4010004@isi.edu> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Mark Allman wrote: >>Sending everyone a CDROM doesn't have the same effect - it would come >>out once a year, and doesn't encourage casual browsing the way a paper >>copy still does. > > Would it make sense to send members (at their choosing) one email / > venue with a list of the papers and pointers to the electronic versions > in the digital library? Might not be like browsing paper, but it'd get > things in front of people without much effort on their part and seems > like it ought to be easy and cheap. > > allman We already do in most cases - it's the call for participation (at least some versions have the full linked paper list) Joe -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (MingW32) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFDZlTzE5f5cImnZrsRAiZTAJ9vieKAhQx1vbpf+ln4fKJhKk/qhACfcDZ5 vVUcHMF26G4a/n11eGbI2EQ= =EhdX -----END PGP SIGNATURE----- From jrex at CS.Princeton.EDU Mon Oct 31 09:49:15 2005 From: jrex at CS.Princeton.EDU (Jennifer Rexford) Date: Mon, 31 Oct 2005 12:49:15 -0500 Subject: [sigcomm] attendance policies for SIGCOMM-affiliated events In-Reply-To: <20051031172130.4A267375FA2@lawyers.icir.org> References: <20051031172130.4A267375FA2@lawyers.icir.org> Message-ID: <4366591B.5040409@cs.princeton.edu> Hi folks, FYI, I checked with ACM SIG Services on ACM's policies about restricted attendance events. The policy is rather vague, though such events are indeed permitted. See below... I've asked for some clarifications but expect that the SIGs are permitted a fair amount of discretion to decide whether to hold such meetings, and what policies (if any) to have on how attendance is determined. -- Jen ACM Policy on Restricted-Attendance Meetings Restricted-Attendance Meetings, as an extension of ACM conferences or workshops, and for circumstances in which it is clearly desirable to hold attendance down to the number of people who will be expected to contribute to the results of the session, shall have a place in ACM activities. In all cases, plans for such meetings shall be publicized and at least the opportunity to request permission to attend shall be extended to all members of cognizant ACM subgroups (such as SIGs and Chapters). For meetings of potentially broad general interest, such opportunity shall be extended to the entire membership of the ACM.