> The idea of a 'canonical answer' is probably more descriptive than the
> traditional 'frequently asked questions (with answers)', but the
> latter is, well, traditional, and people are primed to look for it.

That they might be, though they depressingly infrequently do so.  But that's not my point.  

To me, canonical questions are different from FAQs in one important respect: they're one level up.  That is, canonical questions tend to be quite *general*, and the questions that get closed as duplicates of them are often *specific*.  Moreover, the specifics of the question aren't handled in the canonical answers, and this sometimes gets questioners' backs up.  They feel their question isn't a duplicate because, say, the [ipv4 subnetting examples][1] are all based on RFC1918 addresses, but they want an answer about some `114.` addresses.

I have to explain that, as I understand it, **canonical questions cover broad areas where to the extent that the individual's question is interesting, it isn't specific to their usage case; and to the extent that it's specific, it isn't interesting**.  So we've decided, as a community, to solicit answers that answer in the general sense all the generalisable bits of any given question on that subject, and then declined to provide any further specific answers on any aspect of that topic.

And to that extent, which I find important, I don't think canonical questions are the same as FAQs.


  [1]: http://serverfault.com/questions/49765/how-does-ipv4-subnetting-work