I agree with much of what Dr. Connolley says, with one key exception.
We must take into account the possibilty of POLITICAL bias on the part of journal editors. Since scientists and those who publish their findings are human beings, there is at least the theoretical possibility that they have human failings. One of these is the desire to be "considered right". Another is the desire to keep a steady paycheck coming in. Then there's always the desire to use science to advance certain causes.
The system of publishing results in referreed journals is by and large an excellent one, but we would do the entire world a dis-service if we were to ENDORSE the process as incapable of error or bias.
There have been enough cases of both of these problems to justify holding off on such an endorsement, even if OBJECTIVITY were Wikipedia's only concern.
Topics such as total world annihilation via an all-out nuclear way between superpowers - or destruction of the environment (via over-heating the atmosphere - are naturally of concern to scientists. They are not mere automatons, nor are they mere disinterested seekers, seeking to satisfy their curiosity. Science is also done on purpose to affect technology and policy.
"If I say this, the policymakers will choose that." Who can resist such a temptation? Is there any evidence that scientists better than the rest of us, and immune to ordinary human failings? (And is Wikipedia prepare to amend its NPOV policy to endorse this view of scientific objectivity and infallability?)
I think we should stick to our original commitment. Just say that "certain scientists" say X about Y. And don't try to endorse any particular view on current scientific controversies as "representing a consensus" which Wikipedia then is committing to endorsing.
Ed Poor, aka Uncle Ed Bureaucrat Member of Mediation Committee Self-described "chief exponent and defender of NPOV policy"