Is proof of space actually "greener" than proof of work? Memory and hard drives consume power, and querying them consumes power. Seems like it could be just another kind of work. It would be more ASIC resistant though.
I'm curious about this. Are you saying that relying on storage latency and volume is energetically cheaper than relying on compute?
I see a few flaws in this. First of all: all fast storage media consume energy even when idle. Secondly I think this neglects the embodied energy (energy to manufacture) of storage media. Lastly if a proof of storage mining scheme became popular you'd probably see ASICs that incorporate onboard fast memory controllers with huge caches and other approaches that would improve performance to the point that this would just become another proof of work.
Comparing full life cycle energy of different approaches to securing a cryptocurrency is actually pretty tough. It's also pretty hard to compare it to the energy requirements of more conventional approaches to currency since the energy cost of those is so spread out across society.
One could read "Beyond Hellman's Time-Memory Trade-Offs with Applications to Proofs of Space" https://eprint.iacr.org/2017/893.pdf