Feedback by Luca Trevisan on Opinion 61

Dear Professor Zeilberger,

you write:

"When your next paper gets 
accepted by a `prestigious'
journal, you have a right to be happy, since it is 
most likely a correct and solid 
piece of work. But if 
it gets rejected, because it is `not interesting or 
important enough', you should be even happier, since 
the probability that it is really a seminal paper is 
much higher than if it would have been accepted. Most  editors and referees prefer 
the same-old currently 
mainstream stuff, and usually don't have the vision 
and foresight to appreciate truly novel work."

We all agree that, conditioned on being seminal,
a paper is more likely to be rejected than 
accepted by a prestigious journal, but that's
different from saying, as above, that
a paper is more likely to be seminal conditioned
on being rejected than conditioned on
being accepted.

Suppose a 1/1,000 fraction of papers submitted
to prestigious journals are seminal, and
that presitigious journals accept a 1/20
fraction of overall submissions, and only
a 1/5 fraction of seminal papers.

Then

Pr[ accepted | seminal] = .2
Pr[ rejected | seminal] = .8

but

Pr[ seminal | accepted] = 1/250
Pr[ seminal | rejected] = 2/2,375

(assuming I did my long divisions right)

Of course you would be right if prestigious
journals had a higher rejection rate
for seminal papers than for random papers.
But making such an assumption would be too unfair
to editors of prestigious journals.


All the best
Luca Trevisan

Doron Zeilberger's Homepage