I attempted to download all items in an archive.org collection via torrents (>255 items, >255 torrents). Theory being: potential to get some data from peers, lighten the load on IA; built-in checksumming and repair. I've come to the conclusion that sadly this doesn't work very well. My torrent daemon struggles with so many; lots are permanently at 99%; IA torrents don't include any metadata identifying original vs derived files

I'm going to be able to partially repair this by scripting the download of the matching *files.xml metadata files and then extracting the SHA1 sums. Most of the derived files will be out-of-date, but most of the originals are probably unchanged.