I think he has a good point. For example, DropBox has lots of special handling for image types. It has a special folder for sharing them, it has auto image uplo

I think he has a good point. For example, DropBox has lots of special handling for image types. It has a special folder for sharing them, it has auto image upload from the places phones tend to store them, etc.. Google similarly has auto-image upload and enhancement, and they have a ton of special handling for office documents that can interconvert them. So DropBox and Google Drive are both examples of cases where special things are done by knowing about file types. I can't really think of any service that doesn't. Maybe rsync run by cron or Tarsnap, but only us techies ever use that. Even then, most compression schemes can recognize compressed files vs. uncompressed and not recompress them - like ZIP utils storing files vs. compressing them when building an archive.