Atiq
> -----Original Message-----
> From: tccc-bounces@lists.cs.columbia.edu [mailto:tccc-
> bounces@lists.cs.columbia.edu] On Behalf Of Petri Mähönen
> Sent: Monday, December 14, 2009 4:30 PM
> To: Constantine Dovrolis
> Cc: tccc@lists.cs.columbia.edu; Joe Touch
> Subject: Re: [Tccc] Improving submissions and all that...
>
>
> I did sometime ago, by a coincidence, a look of some conferences (more
> in
> wireless side). This is not statistically reliable sampling, so I
> should look again
> more carefully. But half-anecdotally and half-statistically: There
> are *some* people
> who are submitting a lot to the same conference with collaborators and/
> or students,
> but still it is a relatively small number of papers. Interestingly
> enough also many of
> those people seem to get many of those papers through (I would not
> dare to say that
> there is quality correlation there, but still that is what my numbers
> quickly showed).
>
> But in the conferences I did look, it was not even a close producing
> the flood and
> burden for the reviewers (perhaps differently: we tend to sometimes
> hear in
> coffee breaks complaints towards TPC like; "Why you did allow X papers
> from such and
> such group into the same conference?". This is issue if there has been
> institutional bias,
> but *if* (big if) there was a near perfect ranking, should we thus
> drop better papers to
> make space for wider author spread?).
>
> I did have a discussion with some of my postdocs and students about a
> week ago
> (promoted by this thread), where I made a theoretical suggestion for
> different
> "new world orders":
>
> * In the future for a single conference, one can submit only 1 or 2
> papers from the
> same group (well, author if you want). And only X papers from the same
> department
> are allowed.
>
> * The conferences will be based mostly by keynote type of talks,
> limited number of
> "student/postdoc" papers, and very extended poster sessions (only
> abstracts will
> be published). None of these will have value for tenure track, but
> will be taken
> into account as positive signs for Ph.D. thesis. The won time is used
> for faster
> reviewing for journals, and focusing on long-term, high impact work.
>
> In short, they though that I have had too much coffee or something
> else. The same
> arguments were obviously generated; synchronization between students,
> unfairness
> process of deciding who is submitting where, how to enforce things,
> knowing that
> if small number of schools change the rules are the others following
> etc.
>
> But I have to say that it was hilarious and thought provoking lunch
> break.
>
> -- Petri
>
>
> > Hi Joe,
> >
> >> Before we go ahead with this suggestion, do we have any evidence it
> >> would substantially change the number of submissions?
> >>
> >
> > I will look at the statistics from a couple of conferences I was
> > involved in as program co-chair. I invite others to do the same.
> >
> > Anecdotally speaking though, I think we all know that it is common
> > for people in our community to submit multiple papers (say almost
> > one with each student?) to the major conferences..
> >
> >> Also, how would it affect each of us who have more than 2 graduate
> >> students whose work synchronizes?
> >>
> >
> > This may be a slightly different issue, but I think we push our
> > students hard to publish too many papers. It is typical for
> > a fresh PhD graduate these days to have 10 publications or so!
> > This has become an arms race.. I am trying to convince my students
> > that they only need a couple of super good papers to graduate -
> > they don't gain much from minor papers in second-tier places.
> >
> > Regarding the synchronization issue, some randomization always
> > helps. Noone will get "injured" if instead of submitting to
> > Sigcomm, submits to say Conext or Sigmetrics or Infocom or IMC or
> > ICNP or Mobicom or (your favorite conference) every now and then..
> >
> > Constantine
> >
> >
> >> Joe
> >>
> >> Constantine Dovrolis wrote:
> >>> I like the idea of limiting the number of papers submitted
> >>> by an author (or co-author) to a conference.
> >>> The National Science Foundation in the US has enforced such
> >>> a limit on the number of proposals that a PI can submit to
> >>> a given NSF program. I think we need to do the same with
> >>> conferences.
> >>> This simple change can go a long way:
> >>> 1. reduced reviewing load,
> >>> 2. authors will focus on quality instead of quantity,
> >>> 3. less paper "recycling",
> >>> 4. and hopefully, fewer conferences!
> >>>
> >>> How about we all agree: no more than 2 sigcomm'10 submissions
> >>> for any co-author?
> >>>
> >>> Constantine
> >>>
> >>>
> >>> Petri Mähönen wrote:
> >>>> Like Tony Ephremides already mentioned the quality of conference
> >>>> have
> >>>> deteriorated on average, although there are (of course) also good
> >>>> ones
> >>>> left.
> >>>> One could try to obviously limit the flood of the papers by
> >>>> introducing some constrained resource (token) algorithm into it,
> as
> >>>> mentioned in various emails:
> >>>>
> >>>> * Pay paper submission fee (non-refundable)
> >>>> * Limit submission to X papers by group/person/... (which
> allegedly
> >>>> could solve some *claimed* institutional bias)
> >>>>
> >>>> One could say a lot about fairness of any such solution.
> >>>>
> >>>> However, I think that generally the problem is not only that
> >>>> there are
> >>>> (just) more good papers flooding to the conferences. It is the
> >>>> point
> >>>> that there are so many
> >>>> bad papers, or paper that are not tackling worthwhile problems
> >>>> submitted into conferences. This continues to be the case as long
> >>>> as
> >>>> the conference papers
> >>>> are seen to be a means for career, prestige, traveling to exotic
> >>>> places etc. As a TPC member and chair, I can tell you that a lot
> >>>> of
> >>>> time does not go to excellent papers, there is too much time used
> >>>> for
> >>>> papers that any decent self-critique should have stopped, and of
> >>>> course a tough cases of borderline paper decisions.
> >>>>
> >>>> FWIW, there are also fields where (a) conference papers are not
> >>>> given
> >>>> any value in tenure track etc. games; (b) virtually all graduate
> >>>> students can get
> >>>> poster presented in the conferences (but only poster abstract is
> >>>> published); (c) most of the talks are given by senior people or
> >>>> occasionally graduate
> >>>> students on exceptionally interesting new findings (and again
> these
> >>>> may or may not be published in proceedings, but those are not
> >>>> seen as
> >>>> "career path
> >>>> papers"). One can be a lot of opinions of that approach, but at
> >>>> least
> >>>> conferences tend to be places to go to hear interesting news, good
> >>>> talks from senior
> >>>> and some younger people, and there is a lot of time for
> discussions
> >>>> (and yes, people tend to sit in the sessions and they show up).
> >>>>
> >>>> In the present competitive situation I fail to see easy way (at
> >>>> least
> >>>> on TCCC level) to solve the problem, especially by asking same
> time
> >>>> (i) fewer papers,
> >>>> (ii) higher quality, (iii) more conferences, (iv) less
> conferences,
> >>>> (v) total balance between TPC backgrounds, (vi) perfect reviews
> >>>> etc.
> >>>>
> >>>> I think it leaves also for me an opportunity to try to find a
> >>>> bottle
> >>>> of wine...and not trying to solve this NP-hard problem.
> >>>>
> >>>> Only slightly more seriously, we should also be realistic what re-
> >>>> engineering we can do for conferences/quality statements/etc. I
> >>>> think
> >>>> Joe has been doing
> >>>> a good work on this thread trying to stay his feet on ground and
> >>>> pointing out realities and taking into account different points of
> >>>> views.
> >>>>
> >>>> Petri
> >>>>
> >>>>
> >>>>>> I'm not picking on you, but we can't solve the problem without
> >>>>>> acknowledging that something has to give, namely we can either
> >>>>>> have:
> >>>>>> - fewer papers accepted/submitted OR
> >>>>>> - more conferences OR
> >>>>>> - larger conferences.
> >>>>> Clearly, we have more submissions than good reviewers, much like
> >>>>> we
> >>>>> have more e-mail than time to read it. The solution is that
> there
> >>>>> must
> >>>>> be some constrained resource attached to submissions.
> >>>>>
> >>>>> For example:
> >>>>> (i) one could limit submissions from any one author, or
> >>>>> (ii) one could require a certain review/submission ratio. Poor
> >>>>> reviewers will not be asked to review again, and will naturally
> >>>>> stop
> >>>>> being able to submit papers.
> >>>>>
> >>>>> best,
> >>>>> -Ari
> >>>>> _______________________________________________
> >>>>> Tccc mailing list
> >>>>> Tccc@lists.cs.columbia.edu
> >>>>> https://lists.cs.columbia.edu/cucslists/listinfo/tccc
> >>>> _______________________________________________
> >>>> Tccc mailing list
> >>>> Tccc@lists.cs.columbia.edu
> >>>> https://lists.cs.columbia.edu/cucslists/listinfo/tccc
> >> -----BEGIN PGP SIGNATURE-----
> >> Version: GnuPG v1.4.9 (MingW32)
> >>
> >> iEYEARECAAYFAksmpP0ACgkQE5f5cImnZruQIQCg8hZShzZFe01Ms0enZ8iu7m6q
> >> cigAmwQ2OUoBKSkmGA/oiRhkOTNmxoro
> >> =6hKB
> >> -----END PGP SIGNATURE-----
> >
> > --
> > Constantine
> >
> > --------------------------------------------------------------
> > Constantine Dovrolis | 3346 KACB | 404-385-4205
> > Associate Professor | Networking and Telecommunications Group
> > College of Computing | Georgia Institute of Technology
> > dovrolis@cc.gatech.edu
> > http://www.cc.gatech.edu/~dovrolis/
> >
> > _______________________________________________
> > Tccc mailing list
> > Tccc@lists.cs.columbia.edu
> > https://lists.cs.columbia.edu/cucslists/listinfo/tccc
>
>
> _______________________________________________
> Tccc mailing list
> Tccc@lists.cs.columbia.edu
> https://lists.cs.columbia.edu/cucslists/listinfo/tccc
_______________________________________________
Tccc mailing list
Tccc@lists.cs.columbia.edu
https://lists.cs.columbia.edu/cucslists/listinfo/tccc
Posts
No comments:
Post a Comment