Flagged revisions is not going to solve much more than obvious vandalism. If we flag a good proportion of article, then we will need lots of reviewers, and the level will be set at sysop of lower - the job will be tedious and done by the lazy with an eye on edit count. The problem is that subtle attempt to insert credible untruths, half-truths, or facts spun to create an imbalanced biased picture of a person will almost certainly walk through this.
Only what is obvious to the average lazy reviewer will be prevented - but what is obvious to the reviewer is not harmful, because it is also obvious to the reader. Hence, general flagging will not solve the BLP problem, it will not really even help.
We won't dent this until we start to take maintainability into consideration as well as verifiability. Sure, any individual BLPs /can/ be written in good way, but, taken together, our wiki-structure /will not/ maintain this level of BLPs without an unacceptable level of harmful articles. Eventualism does not work here - because shitty biased BLPS in the meantime are not acceptable.
We have two choices: 1) delete a large proportion of our lower notability (=less watched by knowledgable people) BLPs. OR 2) tweek the structures so that those motivated to be doing the quality control (and that includes clued readers) are able to maintain more articles.
The second option means looking at: 1) Spot banning anyone pushing negative POVs on a BLP. We should not waste resources arguing with such people. 2) Permanently semi-protecting any article where there's been a previous harmful BLP violation that's not been reverted within a few hours. These are the articles that our open editing has failed once - the subject should not be open to it again. 3) *Insisting on sourcing*. Yes, the patroler /could/ google and check the verifiability of the thing for himself. But we simply DO NOT have enough clued patroler to do this. We must put the onus on the editor giving the information to "show his working" - so that the partoler (or the casual reader) will be quicker to see any problems with the sourcing.
Why should unsourced BLPs not be tolerated? Because we cannot maintain any level of quality control as long as we keep making the checker do all the work. You want it in? You source it - otherwise NO.