Sounds hard to define specific metrics to define how many tags are too many.
I think most tags should pass two basic tests:
- Reading the tag alone, am I able to identify some specific topic which the question will be targeting.
- There should not be another tag that already exists which covers the same exact same topic.
For example, in the Substrate StackExchange, we have tags [FRAME]
and [pallet]
. While these topics are very similar, I think that these two tags do have distinct topics they are trying to cover.
[FRAME]
can include questions which are specific to the underlying macros that we create that help users write pallets, while [pallet]
can be used to talk about existing pallets created by the community, or broader pallet development questions.
Another example is [polkadot-js]
and [polkadot-js-api]
. In this case, I think these two tags represent the same thing, and ideally an administrator would go and merge these two tags.
Finally, lets look at [api]
or [curiosity-killed-the-cat]
. The [api]
tag is just way too broad. You cannot easily identify what a question would cover with this tag, and probably it shouldn't exist.
Similarly, [curiosity-killed-the-cat]
is not a real topic or category of questions, and probably needs a more appropriate tag name or to be removed all together.