Really insightful comment.
>what makes a bad question on Slant?
The primary criteria is that it's scoped enough that the Viewpoints are useful to compare. The one missing feature that helps this (and your other points) is a "topic description" that will provide some more context to the title, this is in the works (and is desperately needed)
>That's all fine and good if the point is to list pros and cons of the options. Maybe I should view each viewpoint as an analysis in and of itself. The viewpoint on Raspbian will say it's good for general-purpose stuff, etc.
That is the general idea. As you pointed out the topic doesn't provide enough information, so the Viewpoints are self-contained analyses on the use-case & pro/con for the different options.
>But then what are the upvotes actually measuring? It's not fitness for a specific purpose, it's just popularity, right? And how do I choose something as "best", when I myself might argue that three options are "best" given what I'm trying to do?
Bingo. This is something we are experimenting with. When the question is scoped tightly enough, the votes do correspond to the best option. When its not, as in the Raspberry Pi question, its just a rough indicator of popularity.
But as you pointed out just because the question isn't tightly scoped doesn't mean its not a great discussion. If you just bought a rasberry pi, didn't have a specific use-case in mind for it and just wanted to fiddle around, a high-level comparison of the different OS's is handy.
Edit: Missed the last question:
>And how do I choose something as "best", when I myself might argue that three options are "best" given what I'm trying to do?
Might have to get a little theoretical here, bear with me.
So depending on the scope of the question, the tradeoff analysis either takes place in the mind of the reader or in the writing of the Viewpoints. Back to the Raspberry pi question, as the topic isn't tightly scoped, the Viewpoints are a "analysis in and of itself" and you mentally then pick the option that has pros/cons that best match the use-case you have in mind.
Again this backs up your point about the voting not being aligned to the "best" for YOU, as people will just vote for what they personally like. We have a couple ideas for this and will have to learn and experiment over time.