I've seen the idea before and I'm not really convinced. Sure, it sounds like a great idea, but let me offer some thoughts.
1. I have never been to 11 of the 12 sites mentioned in the article. Before I read this article, I knew exactly one site where lazy registration is actually practised: Stack Overflow. The idea has been floating around for some one or two years now. It hasn't really catched on. To me this suggests that many authors find it difficult to implement.
While difficulty in itself is not a problem, the time and effort invested in something needs to pay off, and it needs to pay off more than investing it into another feature. Opportunity costs apply here too.
2. How many potential users really care about registration? Does a "registration reddit-style" really put off that many people? In my experience, people will just pick their normal username and mash a laughably simple password into the appropriate box. There isn't much cognitive overhead there. This will, in my opinion, really only be a problem to people who have never registered an account before, anywhere.
Sure, you need to convince them that your product is useful. They won't explore further if they aren't convinced that it may provide some value to them. But it's probably a lot easier to put a tour on your website than to implement lazy registration.
3. If it does indeed give more users, do the extra users remain users for the same amount of time, or do they switch more easily to competitors? What happens to the users that you would've gotten with registration as well? Do they switch more easily? It seems to me that that's a very real possibility.
Currently, my intuition tells me to just use a registration form if you need it. In places where you don't need your users to be logged in, don't require them to be. Just let them do as much as possible without an account, but I wouldn't go for "unregistered accounts". That's too much effort for something of which you don't know how it will turn out.