TLDP and Wikipedia
Stein Gjoen, sgjoen@nyx.net
This summary was originally posted on the discuss-list 2005-08-14 at
TLDP and
turned into HTML for posting on the TARPL pages 2006-08-21.
Discussion on wikifying TLDP works started quite some time ago
and since I am working on various technical solutions for TLDP
I decided to give it a try. I have now spent more than 6 months
writing articles from scratch, editing articles, adding commenst
and feedback and participating in voting. Allthis in order to
gain sufficient experiences to enable me to make a conclusion.
So rather than letting you wade through 15 ad infested pages
before concluding with a definite possibility of a firm maybe,
I'll make my conclusion here up front:
Wikipedia in its current incarnation with software and
processes are in my opinion not suitable for the TLDP.
When I started I expected vandalism to be the largest problem but
to my surprise that was not the case. That was then but things have
changed. Vandalism is rapidly increasing and reverts are not always
able to handle this properly anymore. Some examples:
- ISO9000: large block replaced with another language which then
was
badly retranslated back to English with much cleanup remaining.
This one also features vandalism by a registered user (removal of
a cleanup tag), some comments are found in the comments tab.
- ISO9000 reverts: 2005: 2; 2004: 0 Article started 2002.
- Hubbert Peak reverts: 2005: 6; 2004: 7 Article started 2002.
- SourceForge reverts: 2005: 1; 2004: 0 Article started 2002.
Vandalims takes on several forms from large scale deletions to
insertion of controversial points of view, by anonymous users
as well as registered pseudonymous users.
While the vandalism is clearly on the rise we should also soon
expect robot spamming like bloggers have experienced earlier.
Perhaps it has already started:
link
In my opinion it is the process aspect that is the main problem
at Wikipedia and without defined processes with safeguards it
is hard to run proper quality assurance. There is however no
doubt that TLDP also has process issues. Still trading one set
of problems for another does not seem like progress. In details:
Cleanup
Occationally a page is marked with a cleanup tag,
"This article needs to be cleaned up to conform to a higher
standard of quality." Unfortunately the reason is often not
clearly stated, nor the condition for removal of the tag. Also
pseudonymous users have removed the tag entirely without due
process (ref. ISO9000).
Deletions
There is a group of self stated deletionists rampaging through
Wikipedia, attempting some kind of textual darwinism:
Deletionist
Wikipedians.
Stated objectives include "Outpace rampant inclusionism". This
has some strange implications. Example given is
The
great filter
- Requested for deletion on basis of being a "vanity page", and
strangely the vote was not executed on.
- When that failed the argument for deletion was restated as
"a simple theory of an obscure academic"
- A vote ended up as "merge and redirect", even though other
theories have their own pages with expanded explanations such
as the Drake equation.
- Then the redirect was set up but no merging was done.
Thus the article was in practice wiped and the deletionists
won the day.
Other votes
Release
Management
There were 2 votes to keep and 1 for cleanup versus 2 for transwiki
yet the decision ended up for transwiki.
Featured articles
These are deemed worthy of featuring but still of course edited
later on. Unfortunately it is hard to locate the version that
was the one that was featured.
Baselining
This is missing. Known good (enough) and featured articles should
be baselined and be easily accessible. This is important for TLDP
when running off versions for mirroring or CD inclusion.
Conclusion
Wikipedia has a lot of energy in spite of process issues. Also
it seems to me that it does not scale well with more users that
not only bring in more vandals but also makes the system slower.
On Googling for articles I often find copies at commersial sites
that takes the opportunity to fill the pages with advertising.
I have also had problems using the search engine at Wikipedia.
For all these reasons I feel the Wikipedia software and method
does not work for TLDP and the documents.
Final comment
Not all is without hope however. I have spent much time
contributing to Wikipedia in this investigation and have taken
the opportunity to add links to TLDP where relevant and not
spammy. Checking my logs I see people follow these. I hope more
can find the time to add such links where appropriate.
All this highlights the need for processes also at TLDP. We do
have a few issues ourselves we need to consider closely.