I volunteer a good amount of time each week helping Wikipedia and I closely followed the rather lengthy discussion that the community had a bit back about possible changes - flagged protections was just one of many schemes considered. A common argument during that discussion was the flagged protection actually increases editing access.

The pages that you are talking about are called semi-protected articles, which are mostly higher-profile articles that were prone vandalism in the past. Right now, autoconfirmed users can edit them freely, but they are locked to edits from people without autoconfirmed status. Autoconfirmed status is automatically given after you fulfill those conditions you mentioned of waiting 4 days and making 10 edits to unprotected pages.

Under the flagged protection system, non-autoconfirmed users, be they anonymous IPs or newly created accounts, can now submit a change to a semi-protected article, it just won't be live immediately. It will be incorporated into the live version once a "trusted" user asserts that it is legitimate. While the definition of "trusted" isn't pinned down yet, discussion indicates that bar will be pretty low, probably that any autoconfirmed user is "trusted" for those purposes. After all, an autoconfirmed user could make any change they want already, so why not shouldn't they be able to approve a pending change from a non-autoconfirmed user as well?

This means that unlike now, where a non-autoconfirmed user can't submit any change at all to a semi-protected page, under flagged protections they can, it just isn't committed right away. Since semi-protected articles tend to also be highly watched by regular editors, the time delay for an approval of a legitimate edit should be pretty short.

Of course, flagged protections only effects edits by non-autoconfirmed users on semi-protected pages; everything else stays the same, and in fact, the system is adding more ability to edit, not taking it away.

Great point -- it will be interesting to see how one becomes a "trusted" reviewer -- mainly, whether requires knighting by an admin or an algorithmic judgment a la auto-confirmation.

In the article I wrote about Wikipedia last year, I mentioned the interesting model at Helium for refereeing content. It wouldn't work for Wikipedia, but I believe they do measure authority there by subject matter -- i.e., you're invited to referee submissions on subjects similar to what you've written about previously.

actually I wouldn't mind if there were more protections in place. I've been scanning through some page histories lately and there seems to be a large number of vandalism changes. Pretty obvious stuff to the human eye, but it must be annoying to fix these. Usually they're made by IP (anonymous) editors. If just those changes are held back pending moderation, I think it would help legitimate editors get more done.