[Wikipedia-l] [Foundation-l] moving forward on article validation

[Martin Walker]

Please forgive me if I'm formatting this post wrongly, this is my first 
comment to this group.

I am involved with the Wikipedia 1.0 team on en, both the Version 0.5 
project and the contact with WikiProjects.  It might help to let folks know 
what we're up to, since much of the validation work you mention goes hand 
in hand with our mission at Wikipedia 1.0.
1. We are putting together an "alpha test" version of the most important 
articles of Wikipedia (with vetting for quality), with a planned release in 
the autumn of 2006.  For this version, each article is simply nominated by 
one person, then reviewed by another from a "review team".  Anyone can sign 
up for this team, though in practice only a few who sign up seem to review 
much.
2.  We hope to go on to do further expanded versions after V0.5, but these 
will almost certainly include review by several independent reviewers, not 
merely one.
3. Oleg Alexandrov has worked miracles with a bot that uses categories, and 
this is now generating lists daily with the title "XXXX articles by 
quality," summarised here:
http://en.wikipedia.org/wiki/Wikipedia:Version_1.0_Editorial_Team/Index_of_subjects
Anything from our "core topics" list or our Version 0.5 list is tagged on 
the article talk page, the bot picks this up every night.  Delirium may be 
interested to note that the bot stores a link to the version it found on 
the day the assessment was done.  We will compare this with the current 
version, allowing us easily to check for a quality decline when we go to 
press.
4. This bot was mainly designed to help WikiProjects provide us with 
information on their articles.  This is proving a great success, with new 
projects being added to the bot's list every couple of days.  We recently 
began a "second round" of contacting projects, and this will bear fruit 
over the summer and autumn.  The Military History project, for example, now 
has over 4000 articles assessed.
http://en.wikipedia.org/wiki/Wikipedia:Version_1.0_Editorial_Team/Military_history_articles_by_quality
It is my hope that once these assessment schemes become established right 
across Wikipedia, we will end up with a large body of assessments BY 
SUBJECT EXPERTS.  People who know a subject well are much more likely to 
know, "This biography should also discuss X's work on Y."
5. This contact we are building with projects will help greatly if we 
institute a system of expert peer review (no. 2 of Delirium's list) of 
selected articles - we already have people we know in most subject areas.

Regarding Delirium's comments in detail, I don't like option 1 for the same 
reasons as others.  I think option 2 is possible, indeed many groups like 
Chemistry and Military History are already well down that road, and within 
a year I expect us to have most areas of en:Wikipedia covered.  Giving the 
work, responsibility and tools to the people who know and care about the 
particular articles is a very powerful way to do this, and extremely 
scalable.  As for option 3, as Sj points out this is partly what projects 
like Version 0.5 are doing.

Overall, I think it is crucial to distinguish between VALIDATION and 
ASSESSMENT.  Validation is often used rather loosely here, but my previous 
career in the pharmaceutical industry forces me to consider validation to 
mean, "How do we know this article is completely accurate?" It goes much 
deeper than assessment (as done at V0.5), which is merely a 10-15 minute 
scan of the article- "does it seem complete, are the sources cited, is it 
written well, etc.?"  My favourite example is an article I wrote on 
gold(III) chloride, listed as a "Good Article."  How do you KNOW that the 
magnetic susceptibility is minus 0.000112 cc/mol, i.e., can you validate 
this article?  I would like to see Wikipedia move towards having validated 
versions of articles available, articles that have been rigorously checked 
by subject experts.  I'd like to see the standard version of each article 
still fully available, but for all validated articles I'd like a tab at the 
top saying "validated" that would allow any user to see a non-editable, 
validated version of the article.  It might be necessary to create a new 
namespace on Wikipedia to do this.  This approach in effect combines 
options 1 & 2.  There is a proposal sitting on Wikipedia that suggests much 
of this - I don't agree with all of it, but it's a very good start:
http://en.wikipedia.org/wiki/User:TidyCat/Achieving_validation_on_Wikipedia

I am helping to organise a discussion on this very topic at Wikimania in 
August, I hope some of this group will be there.  I think this is a nettle 
we have to grasp if Wikipedia is to move forward and receive the respect it 
deserves.  The only way to achieve this IMHO is to get a group of people 
around a physical table (not a virtual one!) who can come to a workable 
consensus view, and to have people at that table who can also say, "I can 
write the code" and "I will authorise the changes."  Please be there!

Martin A. Walker (User:Walkerma)