My personal opinion is that they’re going about this the wrong way, with IPFS, anything blockchain, etc. They can support this entire model with a DHT (supporting PKI for identity) and torrent advertising, the latest version of BitTorrent (which has great support for cross swarm file serving by hash), and RSS or other feeds for advertising the corpus or chunks of it.

I maintain a copy in Cloudflare R2 of all IA item metadata, and would love to work on the ArchiveTeam equivalent of a Warrior appliance: appliance or container starts, consumes an RSS feed to request the least seeded items, downloads them up to disk space allowed, and then serve downloaded item files to the swarm.