"false hits are a fast growing problem"

[Joerg Arndt]

* Edwin Clark <eclark at math.usf.edu> [Dec 15. 2006 13:56]:
> On Fri, 15 Dec 2006, Joerg Arndt wrote:
> 
> > Why should it be useful?
> > It is especially one that will give false hits, i.e.
> > it will make the database LESS useful.
> > 
> > I USE the database a lot and false hits are a fast growing problem.
> > 
> 
> Yes, just imagine a  Borges like database containing ALL integer
> sequences (I know, I know. It's not possible!). Then every finite
> sequence a(1),...,a(1000) would appear as the initial sequence of
> uncountably many sequences in the database--thus rendering the
> database essentially useless. :-(   Is this the direction that
> sequences derived for the sake of sequences is leading?

Exactly.

> 
> One extreme criterion for a new sequence might be that the sequence
> is based on published mathematics. Is that unreasonable?
> 
> --Edwin Clark
> 

Reasonable but cruel.  Folks like me (dumb computationalist)
couldn't submit sequences like (Gray code for "Pell-words",
using dots for zeros, columns are the words)

  .................111111111111111112222222
  .......11111112222221111111..............
  ...11122111............11122111......1112
  .1221....1221..1221..1221....1221..1221..

and similar.  This _is_ a sequence of sufficient interest.
Similarly (Gray code through sparse signed binary words,
M=-1, P=+1, dot=zero):

   PPPPPPPPPPPMMMMMMMMMMM.....................
   .................................MMMMMPPPPP
   PPPMMM..........MMMPPPPPPMMM...............
   .........MPPM..................MPPM......MP
   PM..MPPM......MPPM..MPPM..MPPM......MPPM...

(Will submit when I have a proper internet connection, i.e.
only next year).

Also sequences that are hard to compute but mathematically
quite uninteresting are often useful entries.

Also many sequences could be interesting if the submitter would care
enough to give a proper (math terms, and correctness) definition,
examples, and comments.  The total lack of all of these would seem to
indicate to better reject a sequence ("do your homework, then try
again").

Random mixes of sigma(), tau(), lambda(), sumOfDigits(), and
MyFooBarFunFunctionVer163() will not create anything useful or
interesting.  If such a sequence should ever be of any interest a
reason should be given (with corner cases: ask the list).

(May have to leave soon, my further answers may be delayed to early
next week.  If so: nice weekend to everybody!)