That's definitly the right direction. Our software stack is getting too complicated, and is carrying so much stuff that is almost never used (e.g. obscure entries in terminfo's db). But of course, real world does not give us much time to rethink/refactor our computing structures.
Maybe people will start to refactor the worst offenders. It takes more than a slight improvement to displace software that's widely used, but I'm sure there are things that we use that could be an order of magnitude better (according to some metric). Gmail was able to win converts because it was an order of magnitude better than the competition.
