Pardon my subject - it is only slightly exaggerated.
Obviously [[WP:V]] and (to a lesser extent) [[WP:RS]] are absolutely vital policies that cannot be discarded. On the other hand, in their current form they are abominations that fundamentally undermine key aspects of Wikipedia's mission.
The problems with the pages are threefold.
1) They actively encourage removal of material that is accurate. 2) Their definitions of acceptable material were written with an eye towards only a handful of Wikipedia's articles, and render large portions of the site functionally un-editable. 3) They increase the burden of responsibility far beyond that which can reasonably be asked of the casual editor who does most of the heavy lifting for Wikipedia.
In order.
1) They actively encourage removal of material that is accurate
Admittedly, our standard for inclusion is "verifiability, not truth." We ought not, however, fall into the trap of deciding that we are therefore against truth. Our goal is to offer the sum total of human knowledge. If information is true and significant, we ought be trying to find a way to get it in.
Both policies direly need a basic common sense check - don't challenge material you don't actually doubt the accuracy of. Editors should read skeptically and critically, but they should not go removing things for the sake of removing them. There should, of course, remain the exception for BLP. We should be clear that this exception is for legal reasons, and out of a desire to be respectful citizens of the world we're offering Wikipedia to.
But to allow and endorse this sort of hatchet work as a general case is far too much of a gift to the clueless. We need to codify that our sourcing policies exist to improve Wikipedia - not to give the most obsessive of us a job gutting Wikipedia.
2) Their definitions of acceptable material were written with an eye towards only a handful of Wikipedia's articles, and render large portions of the site functionally un-editable.
The majority of our policies in this area, and [[WP:RS]] is by far the worse offender here, were clearly written to provide us needed protection against nutjobs on our more pathological articles. They're excellent policies for keeping the Israel/Palestine articles sane, keeping the LaRouchies and Scientologists at bay, and telling Gene Ray that he should take nature's four-sided harmonious time cube elsewhere.
Unfortunately, that's a fraction of our articles. Most of our articles do not need that kind of protection, and that kind of protection is gravely dangerous to them. My favorite example is [[Spoo]] - a featured article that we ran on the frontpage on 4/1/06. Spoo is a fictional food source from the television show Babylon 5. The thing about B5 is that there's a huge amount of information availbale. J. Michael Straczynski, the creator of B5, has made a great deal of stuff public about the show. The only problem is that he did it on Usenet. Now, there is nobody in the world that doubts that those Usenet postings were made by JMS. It's dead-on fact. But [[WP:RS]] rules out such posts in all cases. Which means, by a rigid definition of [[WP:RS]], we have to go scrag [[Spoo]].
This is not the only case of this. I remember a nasty fight over [[Able and Baker]] regarding whether the website for Dayfree Press, a webcomics syndicate, was a reliable source on Dayfree Press. Or, and this is possibly the most ludicrous thing I've ever seen argued on Wikipedia, the assertion that it is not verifiable that a post was made to a talk page on Wikipedia. Not even raising the question of the identity of the poster (Which would have been a sensible debate to have) - someone was asserting that the existence of the post itself was not verifiable.
The problem here is that the reliability of sources is a tremendously subtle issue. When dealing with Babylon 5, Usenet posts are tremendously reliable. When dealing with Scientology, a personal website is one of the most important sources in existence. J. Michael Stracyznski's news posts are reliable. Other people's aren't. It's not a clear-cut matter, and we can't write policy that's meant to be applied like a sledgehammer to cover it. The fact that blog posts suck as sources on topic A does not mean they suck on topic B.
Every other policy we have on Wikipedia explicitly gives editors a wide berth to interpret it on a case-by-case basis. This is, for instance, the whole reason [[WP:NPOV]] works. But increasingly, [[WP:V]] and [[WP:RS]] are written to be applied by machine. We need to return to trusting that editors will be able to work out what a reliable source is on a topic by topic basis, given a set of general guidelines on the matter. I have in the past recommended basing these guidelines on _The Craft of Research_, a book published by UChicago Press. This has long been one of the standard academic research manuals. It's flexible, sensible, and respected - exactly what we want and need.
3) They increase the burden of responsibility far beyond that which can reasonably be asked of the casual editor who does most of the heavy lifting for Wikipedia.
Wikipedia is supposed to be editable by anyone, and we have repeatedly stressed the importance of keeping it open. Unfortunately, with our zeal for sourcing everything and placing ridiculously high demands on our sources, we are increasingly threatening that ideal. It is not, under the rules as written, possible to edit Wikipedia without access to a substantial reference library. Someone who stumbles upon an article and sees an omission is no longer encouraged to just add it - they're encouraged to go make a research trip to add it.
This is, of course, not how it actually works. 99% of our content has been and will always be people working from personal knowledge. It was how Wikipedia was built, and it's how it will always remain, no matter how we rewrite the rules. We need to remember that as editors, though. The goal should not be to scrap 99% of the contributions. It should be to remove things that are too troublesome to include without reliable sources (As defined on a case by case basis), and to try to find reliable sources for everything else.
In essence, we have written a set of policies that fail to reflect how we do work, should work, or could possibly work. And, due to the frighteningly large number of contributors who, given a piece of bad policy, will follow it rigidly without thinking about it, this is a solidly dangerous thing. (Something to remember: IAR is our most ignored rule.)
What can we do to repair these policies to better reflect and engage reality? Obviously, as I said, we cannot abandon the policies - [[WP:V]] is essential, and [[WP:RS]] has the potential to be a vital guideline. But the current approach clearly does not work. So we need to rethink it.
-Phil