[sigcomm] considerations for reviewing extended papers

[Jennifer Rexford]

I think there are two issues getting mixed together a bit here, that 
would be useful to split apart.  First, should we require the longer 
paper to be a non-trivial delta over the short paper to be considered at 
a conference?  Second, if the long paper is being considered, should it 
be evaluated based only on the delta or on the entire body of work.

In my mind, if the long paper doesn't have a non-trivial delta (e.g., 
the addition of an evaluation section, prototype, etc.) it's unlikely to 
be in serious contention at a competitive venue anyway.  So, the first 
question might be viewed as moot in practice.  On the second question, I 
think we should judge the longer paper in its own right -- i.e., giving 
it credit for whatever spark of an idea that the shorter paper already 
articulated.  In my mind, that's the crux of the question.  If we judge 
the paper only on the part that differs, it discourages work that 
fleshes out a clever idea, or discourages publishing the spark of the 
idea in a workshop beforehand.

-- Jen

> 
> Fred Douglis wrote:
> 
>>Vern,
>>
>>I agree that this is indeed a contentious issue, and one with which many
>>conferences struggle, with or without a formal policy in place.
>>
>>Also, your proposal to treat workshops like TRs suggests that everyone has
>>some uniform view of TRs.  I was assuming you basically ignore TRs when
>>considering originality... right?  I think that was more implicit than
>>explicit in your description, but maybe I missed it.
>>
>>I come down on the side of the current SIGCOMM PC chairs, I guess.  I
>>think a workshop paper is to a conference paper as a conference paper is
>>to a journal paper.  Most journals won't republish a conference paper
>>verbatim, but expect some increment, which seems to vary depending on the
>>journal.  
> 
> 
> A slightly different variant occurs when a number of workshop papers is
> integrated; in that case, the contribution of the conference paper is to
> put all the work in a single context. The question is whether
> substantial new work needs to occur for that submission to be
> considered, and whether it is the total of the work or just the new
> contribution that is measured.
> 
> There is also the challenge of how to cite ones own work in a
> double-blind* review process (*- though the review process is oddly
> inverted here, where authors are anonymous during submission but reviews
> are sometimes published with attribution afterwards, but that's another
> issue). I.e., this problem is amplified by blind authorship.
> 
> 
>>I certainly think that if an author publishes at HotNets, say,
>>and then sends something to the annual SIGCOMM conference that is not a
>>"significant" improvement over the HotNets paper, then it should without
>>question be rejected.
> 
> 
> The ACM and IEEE require augmentation to publish work again. The issue
> is whether the augmentation needs to differentiate the submission -- the
> ACM and IEEE minimum requirement, which goes beyond wordsmithing but
> measures the contribution of the work as a whole -- or sufficient _on
> its own_ to warrant inclusion (which seems to have been the test by a
> few PCs recently).
> 
> Joe
> _______________________________________________
> sigcomm mailing list
> sigcomm at postel.org
> http://www.postel.org/mailman/listinfo/sigcomm