Should the human reviewer check every link by hand?
Think codecademy.com now first for "learn to code". I mean, is it best place to learn how to code? At the moment, I'd say it's not. Is it the third result "codeyear.com" (run by codecademy as well)? I'm sure there are better sites that could take that spot and provide more value to the user than that landing page pointing to codecademy (Nothing personal against the codecademy guys of course, big kudos for their venture so far and best wishes for the future).
The problem with today's algorithm is that it can't really tell which site is really good and which one is just popular, mainstream, consolidated, with a high PR. Think lifehacker.com second for "learn to code" (on my side of the world, on google.com and chrome incognito mode on).
Right now if every single newspaper of the world cover your site about how to build a homemade nuclear device and actually write something about the matter in the article (how it would be natural to be), use as anchor text "how to build an atomic bomb" and link to your site, your site will skyrocket to the top for that kw (and many others similar). Even if it's just a school project or a joke that made it to the news.
Where's the quality check?
Anyway I'm sure that smart people are currently working on these details and that a comment reply can't really do justice to the complexity of the issue.