Re: Irregular review procedures for IACR sponsored conferences



On Nov 21, 3:03 pm, Kristian Gjøsteen <kristiag+n...@xxxxxxxxxxxx>
wrote:
Markus.Dichtl.nos...@xxxxxxxxxxxxxx <Markus.Dichtl.nos...@xxxxxxxxxxxxxx> wrote:
The
fact that very bad papers are accepted is a proof that the review
procedures do not really work.I don't agree. The review procedures work, even though they are
imperfect. Rubbish is accepted and worthwhile papers are rejected:
this is known. What is unknown is how to fix that.

And since I argue that IACR, if it has published a wrong paper, a moral
obligation to publish as well a paper, which corrects the wrong one, I
had to indicate which was the wrong one.

Nah. The program committee has an obligation to produce a worthwhile
scientific program for the conference. Accepting papers that point out
that previous program committees made mistakes may run counter to that
obligation (eg. if the mistakes aren't sufficiently interesting).

Perhaps conferences should have a rump-session-like session for announcing
errors in previously accepted papers? Something like letters to the
editor in a journal? With a prize for the most significant error?

--
Kristian Gjøsteen

The main purpose of publishing corrections for wrong papers is not to
prove that the previous program committee made a mistake, but to warn
potential users. Let us look at another discipline. Medical journals
are very quick in withdrawing papers which turn out to be wrong. Of
course they publish the notice of withdrawel in the same journal, to
reach the same audience as the wrong paper. Fortunately, cryptologic
mistakes will not threaten human lifes as often as medical mistakes,
but there can be considerable financial risks. I think that it is
necessary to give the correction of the error as much publicity as the
wrong paper. So it should be in the proceedings of the same series of
conferences. If there is only a rump session like thing to anounce
errors, most potential users will not learn about the problem.
I think that, as the wrong paper will stand in the libraries for a
century or more, the correction should also be found there.

In general, I do not think that scientific interest is, or should be,
the only criterion for selecting papers at crypto conferences. If a
paper shows that a well known cryptoanalytic method breaks a well known
cryptographic algorithm, the scientific interest is low, as the method
is not new. Anyway we should learn about such attacks at conferences.

Concerning the quasigroup postprocessing of random numbers, I see a
considerable risk of people really using it. In the break after the FSE
2005 talk, I spoke with 2 or 3 persons who were really fascinated to
get a postprocessing mechanism whose output data rate is the input data
rate. Hopefully I could convince them that this just does not work,
but I am quite certain that there will be others to fall for this
mistake.

Markus Dichtl

.

Relevant Pages

  • Request for comments and criticisms about Shannons problems
    ... Request for comments and criticisms about Shannon's problems ... I have found Shannon's mistakes about one-time pad and conditional ... The papers can be found at http://www.arxiv.org. ...
    (sci.math)
  • Re: Vet Tech Journals: Unemployed again........
    ... I'm sorry you've had such a horrid time. ... mistakes are the ones who do no work. ... just that I'd never been exposed to the MLA system before. ... he graded my papers on both content and lay out. ...
    (rec.pets.cats.anecdotes)
  • Re: PDEs Involving Absolute Value
    ... problem, i.e., I already know how to do proper L1 norm minimization, ... fact, http://www.duke.edu/~sf59/SRfinal.pdf, and many other papers ... unaware of such published mistakes. ... seek out similar papers and write letters to their publishers? ...
    (sci.math.num-analysis)
  • Request for comments and criticisms about Shannons mistakes
    ... Request for comments and criticisms about Shannon's mistakes ... I have found Shannon's mistakes about one-time pad and conditional ... The papers can be found at http://www.arxiv.org. ...
    (sci.math)
  • Re: Security Analyses of One-time System
    ... Request for comments and criticisms about Shannon's mistakes ... I have found Shannon's mistakes about one-time pad and conditional ... The papers can be found at http://www.arxiv.org. ...
    (sci.crypt)
Quantcast