The database of the site now contains a large record of millions of game plays of players trying to go from one Wikipedia link to another. See here for some interesting academic research that has been done on the site's dataset: https://docs.google.com/document/d/1jSGFRZYrJnlDUBhGbQrO9e-n...
I know what I'll be doing next weekend! :)
OTOH, I learned a lot that afternoon. Watching TV programs has never been the same.
An interesting side note: if you repeatedly click the first link at the beginning of any wikipedia article (except links in parenthesis) you will always end up at Philosophy.
Ignoring this backlink, though, it would work:
Mathematics > Quantity > Magnitude (mathematics) > Mathematical object > Abstract and concrete > Referent > Linguistics > Science > Knowledge > Fact > Verificationism > Philosophy
But if the algorithm should ignore the already visited links then it would probably work because the pages usually describe something using the higher level concepts first. The question is how much larger the pool of "always found" pages becomes.
That was fun!
http://www.imdb.com/title/tt0133093/quotes?item=qt0324307
which cites Lewis Carrol's Alice in Wonderland:
http://www.urbandictionary.com/define.php?term=Rabbit%20Hole
There is a word for it, which I forgot, when you look something up on Wikipedia, the article contains a link to another article, and you go, "Oooh, that sounds interesting", open it in another tab, then, when reading the second article, you come across two or three more of such links, and before you know what is going on, you have dozens of tabs open. The only limit is your patience and your computer's RAM.
Eventually you'll end up reading articles that are not even remotely related to your initial inquiry, but highly interesting nonetheless.