I use Jungle Disk with S3 for storing media that are important to me like photo's. Other stuff i backup locally, but there could be a fire, theft etc. For code i use Darcs and Mercurial repositories and Jungle Disk frequently backups the repos to S3.

Man, good idea... unfortunately I didn't even think about personal use for S3 and I already signed up for an online backup service. I kept thinking that an online backup system would work great with S3 but not for a second I thought of using it for myself :(

There are some tools available for doing backups directly to S3; but they all have limitations because of S3's rather weak API. In addition, the pricing structure of S3 (bandwidth + storage + a significant per-request fee) means that certain workarounds (e.g., adding indirection) end up being absurdly expensive.

S3 can be an excellent backing store for an online backup service, but only if there's some intelligence sitting in between client systems and S3.

I think if amazon marketed it properly, it could potentially weed out a significant share of the shitty hosts that are lurking around. But then again, it's probably easier to just FTP to a shitty host.

Good idea. I think now we are really heading somewhere. This is could be a start to fixing all the clutter and mess on the Web.

If the S3 auth access can be abstracted to use your "normal" amazon login, that will go a long way in making it user friendly.