

Not saying that it’s necessarily a bad option, but my biggest issue with delta chat is that it does not offer forward secrecy (if a user’s private key is compromised, past messages can be revealed); Signal does. Delta no question beats signal in decentralization, though email is less decentralized than it seems–how many people do you know who still use gmail? Delta also inherently leaks metadata on whom you’re communicating with to the email host (that’s just imap/smtp). Signal can mitigate this somewhat with Sealed Sender (which gives one-way anonymity), though it can be broken with statistical analysis, and signal metadata is more identifying due to requiring a phone number.
Roughly a truncated cone with diameters ~7 nuts and ~9 nuts, and the cup is ~12 nuts high (loose guesses, it is hard to tell due perspective and nuts of different sizes). Throw in an extra layer to account for the heap at the top (which is a dome taller than 1 hazelnut, but treating it as a shorter but full layer should give some error cancellation) to give a height of 13. The volume of a truncated cone of those dimensions is ~657 cubic hazelnut diameters. Random sphere packing is 64% space-efficient (though wall effects should decrease this number) giving a total of 420 nuts (nice).
Multiple edits for clarity and typos.
Answer
This ends up being about 5% lower than the true answer. I’m surprised it’s that close. This is in the opposite direction from what I expected given wall effects (which would decrease the real number relative to my estimate). Perturbing one of the base diameters by 1 nut causes a swing of ~50, so measurement error is quite important.