And here are the actually interesting parts:

https://docs.datproject.org/how-dat-works

https://github.com/datproject/docs/blob/master/docs/hyperdri...

Seems very similar to BitTorrent, but without trackers, with Content Defined Chunking (instead of Fixed Size Chunking) and utilizing Merkle trees + public key cryptography to enable append-only operations on shared data sets (rather than 'read-only once published').