OEIS gripes

[David W. Wilson]

Olivier Gerard wrote:
> 
> On Fri, Jul 06, 2001 at 02:02:26PM -0400, David W. Wilson wrote:
> >
> > I was looking at some of your recent sequences.  Many are incorrect,
> > duplicates, or unjustifiably short sequences (easily filled out).
> 
> Your reaction is to my opinion most welcome, but the "your" is a bit
> misleading as the first 8 sequences you quote are from someone who
> is not on the seqfan list (Mr Amarnath Murthy).

I started out intending this note for Neil.  Later on, I decided
seqfan was the proper venue, since Neil is there as well.  Hence "your"
would have originally been to Neil.  Sorry.

> > As much as I love the OEIS, it looks as if it is becoming a dumping
> > ground for some very poor-quality stuff in terms of correctness and
> > relevancy.  The OEIS would certainly benefit from a review process.
> 
> Neil is the final authority on what goes into the EIS, but in the same
> time, he cannot do everything without slowing many aspects of this
> database in the process or spending so much time on it that he cannot
> continue just breathing, eating, doing his personal research...
>
> The purpose of the seqfan mailing list is precisely to help interested
> and devoted people work together to enhance and edit the EIS.
> Many on this list have long mail threads with Neil while improving many
> sequences every week.

With regard to Neil:

As a long-time contributor to the EIS, I have developed a high regard
for Neil. He has personally acknowledged almost every contribution I have
made to the EIS, however trivial.  I, in turn, have always acknowledged
Neil's ownership of and authority over the EIS.  I also recognize that
Neil has many obligations apart from the EIS.  I should think it goes
without saying, but nevertheless: any criticisms I might make of the EIS
or of Neil's policy with regard to the EIS must not be construed as
criticism of Neil himself or disregard for his efforts or responsibilities.
(I assume all this goes for every seqfan member.  A seqfan charter to make
these issues clear would be nice).

I have been authoring/extending/correcting/corrupting sequences in the
EIS since 1994, alone in the pre-seqfan days, now as a member of seqfan.
As a defender of the EIS, I feel compelled to opine that the quality of
the EIS has been declining in recent years.

In the early days of working on the EIS, I could characterize the mistakes
I found as few and honest.  Lately, however, I am finding an overwhelming
number of sequences that are irrelevant, indecipherable, and innaccurate
due to just plain carelessness.

In my last message, I critiqued the sequences on the recent sequence
list in order to illustrate my concern.  Note that these sequences have
already entered the database.  Among them, I found several that were just
plain wrong, many of these, when corrected, were already in the database.
There were others that had only few elements, most of these could easily
have been extended by the author to full length; instead they were
submitted with a "more" keyword in expectation that someone else would
later finish the sequence.  I did not even bother to critique sequences
for legibility and EIS-worthiness.

This notice is found on the EIS submission page
(http://www.research.att.com/~njas/sequences/Submit.html):

  IMPORTANT: Thousands of people use the sequence database every day.
  Please take great care that the terms you send are absolutely correct.
  The standards are those of a mathematics reference work. 

If I could put it mildly, I would: this is a joke.  Most recent sequences
would not stand up to a ghost of a review given to, say, a JIS article.
A JIS article must be accurate, articulate, original, and enlightening.
A good percentage of recent EIS submissions do not meet those criteria.

I think a large part of the problem is Neil's open door policy with
regard to submissions.  This policy served well in the early days of the
OEIS, when Neil could critique all the sequences that came his way.  But
that day has passed, and today Neil and seqfan together have more than
they can do to stay on top of the deluge of incoming sequences.

I feel that the only real way to restore the former quality of the EIS
is to implement some sort of web-based review process. Each submitted
sequence would have to pass under the eyes of a reviewer, not necessarily
Neil.  The reviewer would have a checklist of criteria for admitting
each sequence (accuracy, legibility, relevancy, etc).  If a sequence
has problems, the reviewer has the option (not obligation) to address
the issues.  He also has the option to return the sequence to the
submitter for repairs, or to reject the submission entirely.  Previously
submitted sequences could also be subjected to the review process for
correction or rejection.

In the short term, such a process would stem the flood of poor sequences
into the database, and offload a lot of work for Neil and seqfan.
In the long term, it could trim a lot of existing fat from the database
and improve its overall quality.

Does anyone share or disagree with my opinions on this issue?