This seems to not be a problem on every site. From Gaming, we have 98 open results for "should" in the title, consisting of items like the following. There are 10 closed, just for statistical completion.
- What race should I genocide? - What races you could genocide is helpful, but not as helpful as knowing what races you should genocide to make things easier for yourself.
- When should I start drinking? - Barring the obvious appeal of the title, the question itself is "When should I start using potions", which again, there is strategic advantage towards when it is a good time or not.
- When should I focus my unit's fire? - I think the trend of timing/selection playing a key point in gameplay strategy should be clear by now.
- What items should be saved for turn-ins? - As such, let's pick a different example. This is an example in which the highlight is in what is most profitable, moreso than all that is possible. Hence, should.
I could list a lot more. A lot of it falls into a basic practice - sometimes what someone should do is far more important and has a pretty logical and gameplay reasoning behind it, that far outweighs merely asking what one "can" or "may" do. Sometimes it comes from a matter of the timing of engaging in a strategy, it could matter with the actual structure of the approach, or it may simply be a matter of understanding what is effective.
Some of these could be rephrased with a different word, but that feels extremely unnecessary, and quite discomforting.
I imagine that RPG probably has a similar relationship with this word. It's a vital style of question in a few cases, and I'd rather not have to force an exemption on some sites in order to accomodate having this filter as a baseline. It's not a block, but getting warned for a valid question (especially with a statistically greater number of valid questions) is pretty aggravating.