reviewing

[Juergen Schmidhuber]

A few comments on the recent discussion.

The situation in theoretical physics heralds the things to come in
other fields such as ours. In the early 1990s they were the first
to institutionalize electronic publishing - astonishingly, computer
science itself is a late-comer in this area. In theoretical physics,
priority in the digital public archive has become pretty much the
only thing that counts. Leading journals were forced to shorten the
subsequent peer review process down to 2-3 months (!), otherwise most
citations would go to digital preprints instead of journal papers.

To a certain extent the "bidding-for-papers process" suggested by
various contributors to this list is already evolving. More and more
frequently, journal editors are approaching authors of interesting
preprints, encouraging them to submit a version to their journal,
listing rapid review among the incentives.

How important is the peer review system anyway?  Rustum Roy & James
R. Ashburn (co-author of the 1:2:3 superconductor paper) recently
wrote (Nature 414:6862, p394, Nov 2001): >>>...many leaders [...] such
as Nobel laureates [...] regard peer review as a great hindrance to
good science [...] An enormous amount of the best science has been and
is run without the benefit of this rubric, as is the worldwide patent
system [...] Everyone except the true believers know that it is your
nearest competitors who often `peer' review your paper [...] The
enormous waste of scientists' time, and the absolute, ineluctable bias
against innovation, are its worst offences. `Review by competitors' is
an all-too-accurate description of this system, wreaking devastation
on papers and proposals [...]  ... should not repeat the old canards
such as:" despite the problems thrown up by peer review, no serious
alternative has yet been proposed." Nonsense. They have not only been
proposed but have been in regular use worldwide for a very long
time. The users include the world's largest research agency [...] and
industrial research worldwide.<<< I omitted many statements - do read
the full letter.

Fortunately, none of this criticism applies to connectionism and
machine learning, where all reviewers are completely objective and
unbiased :-)

-------------------------------------------------
Juergen Schmidhuber                      director
IDSIA, Galleria 2, 6928 Manno-Lugano, Switzerland
juergen at idsia.ch     http://www.idsia.ch/~juergen