Thanks for the reply. Point by point:
Not even close. a tiny fraction of human misbehavior is legibly punished.
The more powerful can cause more damage. Also, some human interactions can cause more damage than others. We try to surround interactions where more damage is possible with identification and logs. If the entity is more powerful we should protect more (or all) interactions with identity and logs.
How? Even assuming the client WANTS to prove anything, how would it do so?
In the post, I use zkPassport as an example of identity that could be required. It’s not sci-fi, some internet services already require zkPassport to be used (one example was the token sale of a crypto project called Aztec).
The pieces you list exist and are described in the posts I linked: identity (zkPassport), adjudication (the Vitalik post, Kleros and others are implementing it).
Finally, no protocol on the internet assumes a client WANTS to do anything. They MUST do it to interact or they don’t interact.
If you can actually solve, or even slightly improve, ANY of these, great. Otherwise, you’re not really furthering the discussion.
I do appreciate you taking the time to explain the downvote, but I still don’t understand why the post is being downvoted:
httpi even if possible would not work
httpi if possible would be great (but you think it’s not possible)
one sentence pattern-matched to “crank bullshit” and the whole post got classified as such
The first part of your comment pointed to 3. This last sentence points to 2.
I think deciding between 1 and 2 is already furthering the discussion. There is no point tackling the hard challenges if we think it’s 1, but we should look seriously into this if we think it’s 2. If it’s 3, that is a problem with one sentence, not with the idea.
much clearer, thank you!
On 1.
I am not saying those pieces are solved. I am claiming they are the right pieces (identification and logs).
Government-provided identity is the most widely deployed identity primitive we have, and we already use it for many of the interactions we consider most consequential. My thought is that something like zkPassport could provide a privacy-preserving identity/attribute layer for a protocol.
I do think semi-automatic adjudication is the inevitable future for many high frequency interactions. The level of automation is being researched; I think Vitalik’s post points in a good direction.
On enforcement, I don’t have a strong claim about the adoption mechanism. Some protocols became widespread without a government mandate (e.g. HTTPS); others depend heavily on mandates (e.g. KYC in banking). I think both paths are plausible, and I don’t think the proposal needs to assume one or the other yet.
On 2. the main risk this tries to mitigate is giving more power to misbehaving agents. The more assets (compute, robots, etc) are behind a layer of identity and logs the easier it becomes to 1) identify misbehaving agents and 2) reduce their power.
0. I think it is useful to name early ideas/concepts.
3. Everyone acknowledges we are far from solving these problems. but we have to start somewhere.