uned sequences A071339 - A071341

[Frank Ellermann]

Marc LeBrun wrote:

> it's a reference database, and, moreover, one that is used by
> programs, such as superseeker, as well as humans.

That's IMHO an important point to identify new "interesting"
sequences:  If superseeker finds it, using one of the "trivial"
transformations, then it's not strictly necessary to submit it.

Of course that's only a personal rule I (try to) follow in my
own submissions.  Not asking superseeker is often a bad idea,
because either it finds something for %Y lines (references to
listed sequences), or it prevents me from submitting dupes.

I still feel free to submit a sequence found by superseeker, if
I have a good reason for it, e.g. a %D (literature) or %H (URL)
- but not even asking superseeker is wrong.

> "A numbers" are not a scarce resource.

Here I disagree.  That some of my scripts will fail for A100000
and beyond is my own homebrewn Y2K-problem, but A999999 is a
harder limit.  Really useful tools like superseeker won't work
with millions of sequences.  Not yet a real problem, but maybe
a contradiction in your statements.

> Therefore sequences should be rejected only for the very
> strongest reasons (such as errors, incomprehensible
> descriptions, etc).

ACK.  But "artificial" or "enough of this kind" are also valid.
If zeta(3)..zeta(99) are listed adding zeta(101)..zeta(10001)
is not necessarily a good idea.  I wouldn't like it.

> Indeed, there are a number of existing sequences that differ
> only by a single, arguable, boundary term, or by shifting
> offset or other simple variations.  Yet these details can
> make a critical difference in whether or not an automatic
> transform will be able to "hit" on interesting results in
> the future

Sure.  So if these minor differences are important, then there
should be a %Y to the related sequence(s).  But if a simple
pattern match excluding cross-referenced sequences finds about
652 A ~ B matches with more than 181 characters, i.e. where A
and B are identical, or all terms of A in the same order are
also terms of B, for at least 182 characters (signs, digits,
and commata), without a reference from B to A or vice versa,
then you might understand why I consider potential "dupes" as
a problem.

> It is important to respect the submitters' level of interest
> and motivation, even when we aren't as enthusiastic ourselves

ACK.  In the example discussed here I saw no problem with these
sequences.  But I tend to get annoyed very easily in the case
of "base" sequences, where the keyword "base" was omitted, or
less than ten terms with more than one error are submitted, or
the description + example are unclear, or (the worst case) the
submitter obviously did not check his terms, let alone check it
with superseeker.
                    Bye, Frank