The general idea is that you precompute a bunch of proofs and store them on disk. Challenges are issued and if you have a matching proof, you can claim the token for that challenge.
The part that I find interesting isn't the tokens. It's the concept that an algorithm can prove you've committed some disk space to the network. The reason I find it interesting is because, for small amounts of disk space, the incremental cost of participation is $0. If I plot proofs on unused disk space and only try to match proofs when my computer is on anyway, it basically costs me nothing.
Compare that to someone trying to farm (aka mine) on an industrial scale and they have a real, non-zero cost starting with the first byte of storage they buy. I find it interesting because it's different than normal economies of scale where a large operator eventually outperforms all the small operators (economically).
Unmatched, but valid proofs, are a byproduct of mining the tokens and those are what proves you have some idle capacity. Since normal PC owners pretty much have to buy excess capacity for something like a hard disk, participating in a system like that doesn't have much downside. It's that byproduct that I think is interesting.
I'm not sure if there are any pragmatic use cases, since the 500GB of free space a normal user has might not exceed the value of pretending to be a real person, but being able to say "I have $50 of idle capacity and I can prove it" could be a novel way to combat large scale networks of bots and bad actors.