Edit: here's an article on it. It's called "sudden termination". http://www.tuaw.com/2009/09/03/mac-10-6-comes-with-license-t...
I'm not so sure. Continuous saving is one of those ideas that sounds great until you actually do it, but then you realise that saving isn't just about persistence, it's also somewhat about marking a meaningful checkpoint in your work.
Take a look at Google Docs, for example. Open a spreadsheet, edit a few things, hey, great, I can go back to every previous edit in the history of the document. Two months and thousands of edits later, all those early edits are completely and utterly useless, but it puts a horrible drain on the UI because you don't have (for example) separate copies that you saved from when you distributed draft 1 of the sheet and when you distributed draft 2, which are the only milestones you really care about by now.
The idea isn't even that great in the short term, because any usefulness in being able to jump back to work from a couple of days ago is swamped by the awkwardness of having to navigate through every single edit you've made since then to get there. Give me a real document control system and my Ctrl+S reflex any day.
This is a very interesting question. I'm pretty sure it's pure "technical debt"; we're still so conditioned by extremely slow storage devices (the floppy is still the canonical "save" icon, remember?) that nobody questioned this sad state of affairs.
It became suddenly obvious that all and every application should continuously save its state, so that you would need to name your files only to better find them. A whole new paradigm waiting to be implemented, fully orthogonal persistence ( http://en.wikipedia.org/wiki/Persistence_%28computer_science... ).