I think we could dramatically improve the review queue process if we could auto-populate the queue with questions that are suspect for more refined reasons than currently exist. In the same vein as to the thread about the top reasons to close out a thread prima facie (e.g. the title is "statistics question"), I think we could actually work out some implementable logic for things that should get dropped in the review queue automatically.
To me, the lowest-hanging fruit are the questions for which the only tag is R (or Python or Stata, but to a much lesser extent). At least nearly all of these questions are incredibly low quality -- revolving around either a failure to read the documentation, or clueless employees not knowing how to perform their jobs.
In my imagination, this would take the form of a simple set of rules that the mods/high rep users construct which are applied to new posts. Positive rule findings populate the review queue. An example of a single rule would be ["question has 1 tag and that tag is in (R, Python, Stata)"]. Or another example: the question is tagged "mathematical-statistics and is less than $k$ words in length" for some well-chosen $k$. Or "the question title is in [statistics question, probability question] and the question contains a picture".
It seems that the discussion of my suggestion for more refined low-quality review criteria has been side-tracked -- my central question was whether or not there's a way to create such user-defined review criteria, but it seems that people are focusing on how to devise such rules. My suspicion is that it is not currently possible to customize the review criteria, so my next question is how do we go about making that possible?