I think it's a good idea for the question of "what's an acceptable source" to be distinct to the rule of verifiability. The latter is central to Wikipedia; the former is a much more contentious and fuzzy issue.
People need to understand that primary sources are always acceptable. E.g. if you're referring to a mailing list archive to discuss the mailing list archive, or (for example) the text of someone's post to a mailing list, that's totally fine and what historians and journalists of computer history do all the time.
Absolutely.
Also, _almost_ any source is better than no source.
The most important thing is that the reader know where the information came from--traceability.
The second most important thing is that the reader should be able to make a rough _judgement_ about the reliability of the source. Some people may not feel that The New York Times is reliable (too left- wing or too reliant on Jayson Blair or whatever), but they know what The New York Times _is_. Similarly, if someone cites the University of Pennsylvania's website as an authority for Penn having been founded in 1740, the reader knows that a) it wasn't made up, b) there is quite likely a good argument to be made for that date, but c) Penn is not a disinterested party and universities have a motivation to exaggerate their antiquity.
The reliability of the source is relatively less important. The only real problem occurs when the "source" is, in fact, another entity like Wikipedia--one in which the identity and credentials of the contributor are not easy to assess.