From what I've read and seen in the docs, faq, presentations, etc., it appears to me that IPFS really pulls in a lot of stuff about the network layer, because it's trying to be, well, an Interplanetary (distributed) File System.
DAT is fundamentally a portable, self-contained, data repository. Replicating DAT archives across a broad network and whatnot is definitely a problem that needs to be solved, but IMO that should be solved at a different layer, without rolling in all sorts of complecting concerns such as network ports, routing, and payments for storage and whatnot.