Learning algorithms for the sake of it is incredibly dull, for me. I find reading about nice algorithms interesting, but I don't remember anything later unless

Learning algorithms for the sake of it is incredibly dull, for me. I find reading about nice algorithms interesting, but I don't remember anything later unless I'm solving a problem. You're better off working through a problem set like Project Euler, or trying some TopCoder Single Round Matches (SRMs, there's a huge archive). At least that way you'll see some interesting puzzles and you'll get exposed to the majority of the 'standard' algorithms. By the time you get 'good' you should be able to read a problem statement and immediately think "Ok, this is some kind of graph problem" or "List of sorted stuff? Probably need to use binary search". There are some good books like Skienna's the Algorithm Design Manual which encourages this kind of intuition.