Verifiable Delay Functions

Dan Boneh described these today at Grincon, and they seem like something useful to follow to make an extremely fair POW/consensus method that does not scale with mining farms (or something).

https://crypto.stanford.edu/~dabo/pubs/abstracts/VDF.html

I haven’t seen the talk yet, though I’ve skimmed the paper. What’s a quick overview on how they could be used in Grin, exactly?