Trying to articulate the general rules (my opinions only, of course):
- Can the question be answered usefully? (not lacking necessary details, etc)
- Will the likely answers to this question be useful to others?
- Is the question asked in a way where it's likely somebody with a similar question in the future will find it via google?
- Corollary: If any of those is a problem, can the question be improved to fix those issues while still preserving the original intent?
Or more succinctly: Google hits on our questions/answers are the lifeblood of the site; any question that doesn't have a possibility of contributing to that should be considered a candidate for closure.
I think there's not a lot of value to a "What's an orange tire I can buy" type question. It may be specific enough that nobody but the original questioner will ever find the answer useful. A year or two from now the answer will be wrong because the manufacturer stops making that model (or that particular shop drops it). It's basically a shopping answer that might be better answered with links to the websites of tire manufacturers that make tires with a stripe of color. I didn't close the question when I first saw it, but if I'd seen a vote or two to close I might've.
At least in theory, answers to questions about putting a Trek ABP bike on a trainer could apply to other rear suspension bikes being put onto a trainer.
I wouldn't want to close the more general "part X in Y color" type questions systematically, but unless the color is silver or black it is likely to be an answer that's not much use to many other people. So depending on specifics my opinion could go either way on questions like that. Something a little more general like "part X in different color choices" would be a bit better, especially if it led to answers with multiple choices.
It always depends on the specifics, but I would think that a question like "I have X derailer, will Y shifters work" would be a good question that I'd want to see left open. The ideal answer wouldn't be "yes", "no" or "if you add Z part", it would tell how to tell what shifter would work with that type of derailer. Even better would be to change the question to that slightly more general case: "I have X derailer, how do I tell what shifter will work with it?".
I'm inclined to try to steer away from too much straight out product recommendations. If product recommendations have to happen I'd prefer to see a selection instead of a single one.
I love the idea of questions about how to build a bike, though. And I get that there's details where some specific product recommendations are hard to avoid. The questions that are about trying to learn instead of what the best part is are going to be much better, though.
Coincidentally, there was a recent post on the Stackoverflow blog about this kind of question:
Q&A is Hard, Let's Go Shopping!
Now, I think some of the issues with the shopping type questions aren't as applicable to bicycles because the technology isn't changing as fast as computers and electronics. However, specific product recommendations are still unlikely to be useful for more than a couple years, since all the bikes and parts seem to change every year or two (even if just to nearly equivalent new model).
I do think there can be a bit more value to some bicycle product recommendations than, say, electronics product recommendations: there are bike parts under $50 (such as tires) that can make a huge difference and are very hard to find any kind of reviews or details, even who makes parts like what you're looking for. Most of the reviewing seems to go to the big ticket items.