Bryan Derksen wrote:
Personally, all I really want out of a sifter-type process is "this has been checked and is not blatantly vandalized or currently an active battleground, and the spelling looks okay to me." IMO a sifter like this would take a lot of stress off of editors who, rightly or wrongly, feel the need to keep a constant watch over articles and "fix" them instantly when problems crop up. It would also allow us to feel comfortable stamping ten thousand CDs without the fear that the database dump was taken at the exact moment a vandal stuck in something dreadful that is now immortalized in dimpled aluminium. All the standard Wikipedia disclaimers should still apply.
Once we have that, then maybe we can start looking at ways to produce an even more rigorously proofed versions that includes fact checking. I'm not in a hurry. :)
This sounds like exactly the right approach to me. What's more, it fits nicely with the incremental-improvement process we've used so far. Writing top-quality articles from scratch a la Nupedia turned out to be fairly hard, but incrementally improving articles has worked quite well. Incrementally verifying articles might likewise have some advantages over trying to make a leap from "this could be anything, including random vandalism" to "this has been verified by experts in the field as the Perfect Article". There's always a risk of *too* much complication, but a few simple levels of verification, starting with "this revision is not vandalized, not in the midst of an edit war, and not obviously glaringly bad", would be nice.
-Mark