# Customize criteria for automatic population to review queue?

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?

• +1 This is a great idea. The single, software-only tag has always been a clear indicator of a problematic question. Others include bizarre formatting; inclusion of standard text from our site at the beginning (from a bad cut-and-paste operation); extensive typographical and spelling errors; image-only questions; and certain strange combinations of tags (such as including mathematical-statistics attached to elementary problems). – whuber Aug 4 '16 at 14:43
• I think this is a good idea, but I'm guessing we catch many of those folks as they are asking Q's for the first time. – Andy W Aug 4 '16 at 15:50
• @AndyW That's very plausible, and certainly is the motivation of the first-time poster question. Two questions (1) Do you think the marginal value of the additional filter would be worthwhile (I do) and (2) Does the first-time post queue capture the first questions from users granted 101 rep by association bonus? – Sycorax Aug 4 '16 at 16:22
• At a minimum, you'll need to add [spss] & [sas] to your list. I see 17 single-tag [matlab] threads, 9 for [jags], 2 for [c++] , 2 for [stan], 1 for [minitab], 1 for [eviews], etc. – gung - Reinstate Monica Aug 4 '16 at 20:34
• I couldn't figure out how to search for only a single tag @gung on the main site, so I wrote queries in the data explorer. This counts up the questions with a single tag, and this searches for questions with a single tag and returns the full info to see. – Andy W Aug 5 '16 at 13:10
• @AndyW, I used your query (from chat) to get the counts listed above. – gung - Reinstate Monica Aug 5 '16 at 13:22
• Individual package tags are another common culprit eg [ggplot2]. You probably also want to catch questions with multiple tags, if each tag is either a language or a package e.g. [ggplot2] with [r] is still likely to be off-topic. – Silverfish Aug 5 '16 at 21:22
• These are all good points, but I think we've assumed away the question in my post, which is whether it's even feasible for us to do this with the current software. – Sycorax Aug 5 '16 at 22:27
• I'm pretty sure this is not something we can do ourselves; it would require the SE developer team to create a custom path for populating the queue. – gung - Reinstate Monica Aug 8 '16 at 16:46