Look ma, it scales
For what it's worth, below is what I learned by studying how Low Quality review queue works at Stack Overflow. Per my understanding of recent feature changes, NAA and VLQ flags push post into this queue unless it has been already reviewed there.
Side note it's not the first time that I am active in this queue but things feel really different compared to my past experience.
Oblivious of meta
Review in this queue is supposed to run without any meta guidance.
This is a matter of simple math: there are about 30,000 eligible reviewers (2Kers), while top viewed flags question at MSE has only about 18K views - including views from lower rep users, as well as views from users of other SE sites and from all the internet at all.
No diamonds
Review in this queue is supposed to run without diamond moderator involvement.
This is a matter of simple timing: in straightforward cases, deletion takes only a few minutes. If it was expected that moderators control such a fast paced activity, they would have to be on a 24x7 permanent watch, which is certainly not the case.
Mob rules, balanced
At first glance, it seems that just six-seven 2K users can in theory create a death ring and remove 20 any answers a day through flagging and consequent voting in LQ queue (20 is review limit when queue has less than 1000 items).
But system is indeed designed to prevent anything like this from happening.
To start with, if post author ever notices deletion they can simply undelete it solely at their discretion. And this is not all, system appears to be designed to either dissipate into disputed flag or alert moderators any time when things deviate from happy path ("happy path" here being unambiguous decision of reviewers and voters that post isn't wort keeping, followed by total negligence of its author).
For the sake of completeness, moderators have means to block repetitive abusive flagging (flag suspensions). But if you think of it, system is designed so that most bad flags are expected to be blocked before they even bother a moderator.
When there are 30,000 mostly reasonable reviewers in the queue, chances for a substantially mis-flagged post to pass through without being stopped by regular reviewers are pretty slim.
First round: light as the breeze
Overall, it looks like no risk and no pain are involved for all the parties in the initial phase of dealing with VLQ/NAA flags. Worst case for flagger is they get painless "disputed" flag. Worst case for author who noticed deletion of their answer is they click undelete.
Bad answers are easy and quick to delete, and when deletion mistakes happen, it's easy to revert for post author.
- Interesting that per my observations, SO hot questions recently look like carrying less of trash answers than it was before. Not having 10k over there, I can only guess, but it's quite possible that garbage simply is quickly cleaned up via queue (and if the posters undelete, moderators get notified and can intervene).
How does it feel like for a (responsible) flagger?
In a word, great.
Best thing is, one doesn't need to study cumbersome rules of the game anymore. You just flag answers you feel better be deleted and regularly check amount of your disputed flags.
When you feel like it's higher that you would want it to be, you visit posts with your disputed flags and try to figure what could make your peers - regular 2K users - disagree with your flag.
The system feels simple, transparent and fair.
What does this mean for Workplace?
Hard to tell, because amount of posts, flags, reviewers and moderators differs so dramatically. Taking into account site specific quality requirements like Back It Up / Don't Repeat Others adds to the complication.
Above approach works at Stack Overflow but they simply don't have an alternative due to large scale.
At smaller sites, it is technically possible to have more mod control over review queues, for example like in Programmers close queue (where moderators tend to intervene when it looks like there's too much stuff in there) or even a really tight control like in Low Quality queues at Programmers and Workplace.
Related reading:
