[seqfan] Re: On editors-authors discussions

Marc LeBrun mlb at well.com
Fri Sep 9 03:14:10 CEST 2011


I think the editors are doing an excellent job with a very challenging task!
Kudos and appreciation!


A quick suggestion:

>> In other words, when an editor asks "Why is this interesting?", this is not

>> an invitation to reply via a message, but to add something to the sequence

>> to answer the question.



> Yes!  Precisely!  I've been looking for a wording that will suggest
that to
> people.  Often I write that and often people react defensively
to it.

A possible way to phrase it might be as a concrete request, rather then an
open-ended question.  For example, instead of asking just

  "Why is this interesting?"

Make a concrete request

  "Could you please clarify with some brief notes what about this sequence
interested you, how you came to study it, or the like?"


Some quick responses:

> I wish A166746
counted n-"digit" representations rather than those below 10^n

No need to wish, compute the desired sequence and submit it too!  Or, if you
can't do so right away, you might put a suggestion in the Comments like "XXX
would also be interesting" so it's not utterly forgotten.

More generally: maybe it's already present and I just haven't found it on
the Wiki yet, but if not I think it would be an EXCELLENT idea if someone
were to start a section for "Suggested Projects".

These could range from big complex efforts (maybe even fundable research?)
to a list of these kind of miscellaneous ToDo's.  (Heh.  Of course I really
should do it myself, rather than just making suggestions but it's been on MY
ToDo list far too long!)

> is A189408 actually interesting?

Definitely!  It was interesting enough for *someone* to compute, write up in
the referenced papers, and for *someone* (ahem) to take the trouble to craft
an submit and OEIS entry for, so ipso facto it is interesting (at least to
*someones* somewheres) right?

Because of this phenomenon (sort of like the presumption of innocence in
law) in the OEIS it's important to keep in mind that ALL submissions should
be viewed as having a small positive initial presumed interest.  Rejection
should require a preponderance of evidence that including it would reduce
the NET value of the OEIS by offsetting this with sufficient inutility.
Borderline submissions should always be given the benefit of the doubt.  The
important line to hold is preventing the OEIS from being disfigured by
mathematical graffiti--which is easy to spot!  Don't sweat the small stuff!!


> And then
there are sequences that come too close to an existing
> sequence.
Should A141768 have been approved, given that A090659
> was already in
the database?

Absolutely!  The fact that someone's even questioning this worries me a
little.  The *sequences* are clearly very different, although their
*definitions* appear to be strongly related.  The goals of the OEIS are NOT
very much like the goals of Bourbaki; the point of the OEIS is NOT to create
some kind of minimalist structure.  Heh, otherwise we could just have only
A000027 and leave everything else as an exercise for the user!  No, an
encyclopedia should be encyclopedic.  As long as their relationship isn't
utterly stupid, leaving out sister sequences diminishes rather than improves
the value of the OEIS.

Please don't be stingy with sister sequences, there's plenty of precedent:
for example, not only do we have A000027 but we also have A001477!  And keep
in mind that even what appear to be very trivial differences may be vastly
significant in some contexts, and can be greatly magnified under automation.
For example merely shifting its offset by 1 radically changes a sequence's
Mobius transform.  

So please keep in mind that blacking out a sequence in the database will
cast shadows all over the search space.  I agree, balance is a challenge.





More information about the SeqFan mailing list