# “Risk” tag: no Wiki, unclear usage

The tag has no Wiki and seems to be used for several different topics, e.g.

• Bayesian risk
• risk in finance applications such as Value at Risk (VaR)
• epidemiology and accidents (human health risks)

(see the top few most upvoted questions here).

Should we write a Wiki for the tag and specify what it may and may not refer to?

• I know very little of these topics, except a little bit about financial applications, so I am not volunteering. – Richard Hardy Jun 16 '16 at 19:13
• The tag has been used 74 times, so it's just about feasible to do some kind of split, if that's the consensus. Did you pick up any sense of the relative numbers? – Glen_b Jun 16 '16 at 23:28
• We have a lot of ambiguous tags. Experience tells me that nothing happens about them until there appears a person who takes the responsibility and burden on themselves to do the whole sorting thing. We have had several such sortings, including some tags with several hundreds posts. Mostly though nobody is volunteering and nothing gets done; ambiguous tags remain ambiguous. Simply raising the issue on Meta does not tend to help at all (still, +1 and thanks for raising it). – amoeba Jun 17 '16 at 0:52
• @amoeba, I get it. I checked the top users of the tag, no one seems particularly active there, so perhaps getting their attention to this question would not help much either. – Richard Hardy Jun 17 '16 at 6:24

I think for statistical questions, only when risk is an expected loss does it make sense to tag in a question. To enhance the browsing and tagging experience, it would be useful to rename risk to risk-function.
In the other settings, "risk" tends to be specific to the field and not of particular statistical interest. I would advocate for deleting the risk tag on questions where it is application specific as it is a conflicting use of terms. In some applied questions, people have gotten away with using a tag like finance or biostatistics, so that would seem a better option instead.
• @amoeba good points again; makes my suggestion a little less obvious... my subjective preference is RR, AR, and RD are better left as separate tags since they are very different (in my narrow field) in terms of modeling approaches, and I would say the same of competing risks. If they were to be merged, time-to-event would perhaps be the right grandfather. I shall edit the answer. – AdamO Jan 5 '18 at 23:43