Quantity of followed links is mostly a list building issue. Like most link building methods, there are crappy ways to do it that will bring you no value and there are quality ways to do it that will bring you value. That's why you'll have 100 threads on BHW telling people to do X and then 100 more threads telling people not to do X/saying it doesn't work. There's a LOT of variables. If you take the time to build a good list you can get lots of followed links.
Here is a fairly old thread on getting followed wiki links. That said, I don't mind links that aren't followed because using only links that are looks unnatural.
Relevancy is why I like wikis, because I control most of the page content.
(WARNING: conjecture ahead!)
I think Google primarily devalues links based on page relevancy, not domain relevancy. Not because they don't have the ability to do that, but because a huge quantity of natural backlinks that G supposedly loves are from one-off pages on someone's blog, a news site, forum, bookmarking/sharing/social network site, etc. These are not sites that devote their whole existence to dog training or blue widgets. Telling the whole web that their links are of no value unless they restrict their domain to a specific subject is, IMO, contrary to G's best interests. In short, to do that would be to promote almost exclusively unnatural linking schemes (like ours).
Of course you'll get more value from a relevant domain, but I think that is not because of devaluing, but primarily because that domain and those pages are
also backlinked with relevant keywords. Hence the power of tiered structures. A content-orphan can be made to look relevant when others link to it with relevant content/anchors.
