[seqfan] Re: moving to wiki

[David Wilson]

Your ideas are great. And all of them would be more effective on a Wiki 
project page than in a seqfan message. :-)

Joseph S. Myers wrote:
> On Thu, 27 Aug 2009, David Wilson wrote:
> 

> Apart from proposing and discussing new projects there will surely be 
> general policies to be agreed by the community.

Policy pages are a must.

> For example, we've 
> discussed global editing and bulk downloads; just as Wikipedia has various 
> policies on automatic and semi-automatic editing (see 
> <http://en.wikipedia.org/wiki/Wikipedia:Bots>), so I imagine the OEIS wiki 
> may also need the community to agree policies in this area (though likely 
> rather simpler than Wikipedia's); any idea someone has for some form of 
> automated check or consistency improvement seems likely to need bulk 
> downloads for prototyping purposes, and bulk editing to implement any 
> changes agreed to be a good idea.

So we create a "bulk downloading" project and a "bulk editing" project 
page to discuss any projects and policies related to those subjects. Our 
conclusions are incorporated into the policies pages or project 
requirements.

>> - Redundancy elimination. e.g, create b-files for all sequences, and use 
>> the b-file as sequence source, and generating the %STUO lines and list 
>> view from the b-file.
> 
> (In the process of doing anything like this, be careful to preserve any 
> existing comments of interest contained in b-files or on the lines linking 
> to the b-files.)
> 
> - Allow a b-file to be given when entering a new sequence as the way of 
> submitting the terms of that sequence.
> 
> - Make the "full" keyword refer to the set of terms included in the 
> b-file.

All great ideas, IMHO. My ulterior motive is, once we have localized the 
sequence data into the (hidden) b-files, we can use the b-files to 
auto-generate sequence attributes (%STUVWXO lines, some keywords, etc) 
and views (list, graph, listen, download = current b-file link).

Once we have localized the numeric data into the b-file, we can begin to 
think about, say, auto-generating the b-file itself for simpler sequences.

But everything in its order.

> - Move from the explicit "more" markers for sequences needing extension to 
> an automatically generated list of sequences in ascending order of the 
> number of terms present (including in any b-file), together with some 
> marker to indicate that a sequence should not be extended because the next 
> term would exceed some limit on the size of terms present.

Another project might be to review the current keywords. In some cases 
we can use finer attributions:

finite:	yes, no, unknown
full:	yes, no, unknown
order:	complete/eventual (entire sequence or tail ordered)
	partial/total (partial or total order)
	ascending/descending/constant/periodic

Also, a ? could be appended to certain attributes to indicate that it is 
a probable or conjectural attribute. For example:

finite: yes?

means the sequence is almost cetainly finite, but not proved so.

etc. These are just offhand ideas, which need to be beaten about the 
head and neck with a broken bottle.


>> - Independent Wiki pages for terminology, extended comments, etc.
> 
> When you have such pages to define terminology as used in OEIS, it might 
> then be useful to be able to copy text from Wikipedia into such pages 
> (which would require licensing under the Creative Commons 
> Attribution-ShareAlike License).

Yeah, I think it would be better to keep our terminology in the family. 
The OEIS should have its own terminology pages, defining standard terms 
that could be used unambiguously in OEIS sequences without cluttering 
the sequence with term explanations. Indeed, glossary pages could be 
tailored to specific sequences or groups of sequences if need be. These 
pages could reference Wikipedia or MathWorld articles, but official OEIS 
terminology should be self-contained in the site.

For example, conventions used in OEIS sequences could be recorded in 
glossary pages, e.g:

n is used either to stand for a general OEIS sequence element (as in 
"n^2+1 is a prime") or a sequence index (as in a(n) = n^2+1).

or

unseen: An unseen value is a value that has not occurred previously in a 
sequence.

We could then abbreviate

a(n) = smallest k > 0, k != a(i) (i < n), such that blah blah

to

a(n) = smallest unseen k > 0 such that blah blah

unseen would be the standard terminology (unless we settle on "new" or 
something else).

> More suggestions for projects and enhancements:
> 
> - The OEIS has a handful of concordances to sequences mentioned 
> (explicitly or implicitly) in certain books - with markers for places 
> where new sequences may need adding to OEIS.  The wiki should greatly 
> facilitate creating and improving such concordances - one person could put 
> an outline of chapters on the wiki, another pick up a particular chapter, 
> enter some sequences and make clear on the wiki page which bits of the 
> book they have or haven't checked for sequences.  Likewise for sequences 
> from journals, competitions and elsewhere in the literature, the wiki 
> should help track what sources have been searched and what was found 
> there.
> 
> - Bidirectional cross-references.  It should be easy to make two sequences 
> each reference each other with a single edit.  By processing the full 
> database it should be possible to identify cases where sequence A 
> references B but B doesn't reference A; these will be a mixture of typos 
> in cross-references, cross-references to sequences that were deleted or 
> renumbered, cross-references to more important sequences that shouldn't 
> necessarily get links back, and cases that ought to be bidirectional.

I thought it might be nice to submit a single comment

%C A?????? A123456(n) = A654321(A101010(n))

and have that comment viewable on all three sequences (with the A-num 
replaced by "a" where appropriate). It would be yet cooler if we only 
stored the comment in one place, so that edits would propagate as well.

> - Easier editing of the Index and listing sequences that aren't mentioned 
> in the Index.
> 
> - Watchlisting individual sequences of personal interest, as on Wikipedia, 
> so you can get a list of recent changes to such sequences.
> 
> - Reference and external link databases separate from the uses of those 
> references and links in individual sequences, so only one place needs 
> changing to update a widely used reference or link (e.g. changing a 
> reference to a preprint to refer to the final version, updating a link to 
> a website that has moved).

What a concept. We maybe could do the same for, say, editors, 
contributors and references and indeed OEIS entries, so that when they 
moved or changed, we need only update them at one location. The OEIS 
looks more and more like, well, a database.

> - Declaration of relations between the terms of one sequence and those of 
> another so that extending one sequence automatically extends the other and 
> inconsistencies are flagged for investigation.
>