I read that I have no control over who I send my data to (to what notary, in Holochain), as my data hash is random, and therefore my data will be sent to a random set of notaries. But what if I construct the following attack:
I create as many sybil identities as possible or adaptively corrupt a good number of real notaries.
Instead of immediately sending out my data, I grind through hashes until I get a hash that matches me with my controlled notaries.
I imagine this can be stopped if app validation rules make it so that, by the laws of chance, I have to grind through so many hashes to get n=0 on any honest notaries (for example the notaries checking data are always 300 random peers in a network of 50,000, and I have corrupted 10,000), that this ‘PoW’ like work of mine has a chance of succeeding that is so small that it would amount to breaking cryptography. Maybe this is not realistic.
However, what if I manage to grind through hashes so that I get to send my data to 270 corrupted notaries, and 30 are honest (some number where I control many notaries and a DDoS against the honest notaries becomes feasible). I DDoS the honest notaries so that they have a liveness break and get validated by the 270 corrupted validators?