On Fri, Feb 13, 2009 at 8:20 AM, Lars Aronsson <lars(a)aronsson.se> wrote:
Emanuele Casadio wrote:
This would be a great idea but, wait a minute...
will this procedure
When we renamed [[en:Angola/History]] to [[en:History of Angola]]
(after it was decided that subpages was a bad idea), we copied the
text, because back then there was no way to rename a page. In the
revision history, it looks as if I (user:LA2) created this page on
February 25, 2002, but I probably only contributed the initial
line. The oldest versions from 2001 can be found in the revision
history of [[Angola/History]], but this was never documented.
We're not discussing the same thing here.
Emanuele is talking about intentionally "archiving" the old edit
history of an article (to avoid reaching 5,000 revisions at the same
You are talking about creating a new article with information cut from
the "==History==" section of the main article (to avoid letting it
grow too big to load)
Either of these could be considered GFDL-compliant by Wikipedia
standards as long as the page titles and link syntax are standardized
enough to be machine-readable (so at the current time, no).
If I wish to mirror the [[History of Angola]] article and credit all
editors who may have directly or indirectly contributed to the current
content of this article it would be easy to scrape the edit history of
[[History of Angola]] and the upload log for each image in the
article. But it would be much harder to get a list of users who have
edited the [[Angola#History]] section. Even if they have never edited
[[History of Angola]] they may have contributed material which was
cut-n-pasted from the former to the latter.
Perhaps we add something like this on the toolserver, some way examine
all edits to an article and get a list of users who have made changes
to a section with a certain title. This could be integrated with some
kind of "extended attribution" tool as long as it knows which rocks to
There would always be a number of false positives (credit where none
is necessarily due) to this approach. However this is already the case
when you consider vandalism and reverts, or edit wars or any other
case where no unique revision is added. I guess one could get more
sophisticated with this, by considering only the earliest revision
when duplicates are found, then get a list of unique users
contributing to the short list of revisions.
False negatives on the other hand would be harder to correct.