Interesting that you pick that, it would be more meaningful if TLS/SSL was not vulnerable to man in the middle and certificate spoofing (respectively). The security is useful, but is it sufficient? Sure searching for hash collisions provides a use, but would it not be interesting to devise a way to provide proof of work for a problem that had more useful results. A good example would be solving satisfiability problems posed in previous blocks, it is easy to calculate the difficulty of the problem and the solving solution is not much better then trying random solutions and hoping and the results is instantly verifiable. It would let people encode meaningful problems to be used as proof of work.