Conversation

I’m really interested in this conversation, because I think the biggest problem in crypto privacy tech right now is that 99% of people think that “mixing” is the only possible kind of privacy, but I think “mixing” can never succeed at providing privacy.
2
23
To me, “mixing” means trying to hide the patterns of your choices in among the patterns of other people’s choices. Computers can already see through that, and they’re getting better at seeing through it faster than we’re getting better at “mixing”. It’s never gonna work. BUT…
1
18
At the risk of rambling on, Imma say more. The fundamental distinction has to do with things on the order of 2^32 possibilities vs things on the order of 2^128 possibilities. Let me explain …
1
8
Suppose you are one of 4 billion (2^32) users, and you let an AI watch the actions of all 4 billion of you, and then you ask it to identify you uniquely among all of them based on your patterns of behavior. Will it be able to do so? …
1
10
Answer: yes, and it requires surprisingly little information. Just let it see a few bits of data about your behavior, and the AI will be able to identify and track you uniquely out of all 4bn (2^32) users. …
1
14
Now, suppose you have a spending key which is 128 bits long — eg a seed phrase which is maybe 12 words. Can a computer “brute force” guess your seed phrase out of all possible seed phrase? …
1
5
This is why attempts to “anonymize” your crypto by “moving it through” something typically fail. That action that you take is itself emitting information, and the haystack of other information that you’re trying to hide in is generously on the order of 2^32. …
1
15
Show replies