Larry et al.
(I'm responding to several concerns in a single mail here),
I understand your point, and I agree with it to some extent: People who
write articles aren't typically the best judges of their own work. But I
think you're making a mistake by assuming that someone who would take part
in a certification project that is a subset of Wikipedia would necessarily
also be a) a contributor to the specific article/subject he certified, or
even b) a contributor to Wikipedia at all. The experts you want could well
use the system, see below.
Elian suggested a scheme where we would simply get aggregated ratings from
everyone. I am familiar with such schemes (Kuro5hin uses it), and I have
considered that option and decided to submit a different proposal, the
team system. The idea here is that teams can make their own rules, and by
selecting a team to trust, I select a whole ruleset according to which I
want to view articles.
With aggregated ratings, on the other hand, I either have to trust the
aggregation (which includes trolls and people I may not find trustworthy)
or select individuals to trust, which is cumbersome (Slashdot uses such a
system, it's very very much work to build a useful list of friends/foes).
With teams, I just have to trust the team maintainers to keep a consistent
policy and kick out people who violate the rules. [As an aside, automatic
account creation is a serious problem even in communities with a more
complex account creation process. Once you can get greater power by
creating more accounts, people will do so. Our system should not allow
For example, I could decide to trust Team Nupedia. Team Nupedia might have
a policy, defined on its team page in the [[Team:]] namespace, that
members of the team must a) be experts in the subjects they certify, b)
not be direct contributors to the articles they certify. They could also
set a high certification thresholds, e.g. 5 members of the team must rate
an article for it to be certified. This is just an example, you could make
up the rules you want. Then if Team Nupedia gets enough members, it might
provide many useful high quality evaluations of selected articles.
I think the modus operandi here would not necessarily be different from
what you envision for a separate project, but have the advantage to be
directly and visibly integrated into WP, thereby attracting more people
(even if you want an expert-centric team, you will probably get more
experts by addressing a larger sample of users).
Other teams might adopt more liberal approaches, trying to separate
obvious low quality articles from possibly high quality articles (i.e.
detecting egregious NPOV violations, spam, bad writing etc.). These teams
might produce more output and be valuable to do basic filtering (which
might go both ways -- I think it might be valuable to have negative
certification, too, to detect bad articles; I believe it was Ed who
suggested something similar). This is useful, but a separate goal from
creating a truly trustworthy encyclopaedia -- in the team system, it can
be accomplished within the same framework.
I think that such competing but yet cooperating teams with different rules
are better than either forking the whole project or forking selected
articles. I agree that one aspect of my proposal, namely that teams be
approved by the "cabal", was not very well thought out, but I do think
some selection process needs to take place, because if we tolerate troll
teams, those might create bad articles just for the purpose of certifying
them, regardless of them being replaced later; after all, it would always
be possible to view the certified version.
So much for now. To Ed: Feel free to create the page you mentioned, but
perhaps it is better to keep this discussion on the list until concerns
are addressed. Also, let's not move too fast. Wikipedia has survived well
without such a scheme for almost two years, and we should be careful with
any changes that could alienate contributors or make Wikipedia in any way
less valuable. Believe me, though, that I'm just as much in favor of
openness and massive collaboration as everyone else here -- if I had the
slightest belief that certification would undermine this in any way, I
wouldn't suggest it.