Concept: Decentralized Reputation Initiation Protocol (DRIP)

[sanity]

This is a preliminary discussion for brainstorming purposes, not even close to a complete proposal.

DRIP (Decentralized Reputation Initiation Protocol) is a mechanism for initializing reputations in a decentralized system similar to the "web of trust" used by PGP and later Freenet, based on the prisoner's dilemma problem in game theory. DRIP is designed to encourage cooperation between existing users and discourage Sybil attacks, while also achieving some other important goals:

• Being more egalitarian, because it benefits trustworthiness rather than wealth
• Avoiding pointlessly burning energy, unlike proof-of-work systems
• DRIP is unique in that it uses "proof-of-trust" as an alternative source of scarcity, rather than relying on economic incentives like proof-of-stake systems or energy consumption like proof-of-work systems. This means that users who are trusted by others in the system have a higher likelihood of being able to create new identities, regardless of their wealth or other resources.

Here's how it works: Two existing users (sponsors) participate in a trust exercise to "sponsor" a new user. They each have a "cooldown period" after sponsoring a user, which is a period of time during which they're unable to participate in the trust exercise again. In addition, there is a "co-sponsorship cooldown period" which is a period of time during which any two sponsors are unable to co-sponsor with each other. The co-sponsorship cooldown period is much longer than the per-sponsor cooldown period to limit the rate at which any two sponsors can co-sponsor with each other. Both the per-sponsor cooldown period and the co-sponsorship cooldown period are specified by a global decentralized system, which monitors the rate at which identities are created and adjusts these cooldown periods to limit this rate.

In addition, there is an "inbreeding prevention" mechanism that prevents sponsors who are too close on the "family tree" from co-sponsoring a new user together. This helps to maintain the diversity and robustness of the system.

Each sponsor secretly decides whether to cooperate or defect, and they commit to their decision by exchanging hashes and signatures. Then, they reveal their cooperate/defect decisions and the protocol determines the outcome based on their decisions:

• If both sponsors cooperate, their cooldown periods are of medium duration and they each get the right to create a new user.
• If one sponsor defects and the other doesn't, the one who defects has a short cooldown period and the one who cooperates has a long cooldown period.

[AshtonSnapp]

What exactly does the cooperate/defect system exist for in this proposal? What problem does it solve?

Also, as a side note, it reminds me of the Prisoner's Dilemma: cooperating is better for the group, but defecting is better for the individual assuming only one side defects.

[sanity]

Yes, it's based on the prisoners dilemma. The idea is to provide a decentralized way to create identities, but make it hard enough to avoid people creating lots of puppet identities for nefarious reasons (ie. Sybil attack). Does that help clarify?

[AshtonSnapp]

That does help clarify, thank you!

One question then: what should happen if both sponsors decide to defect? I'm assuming this is a possibility since, like in the Prisoner's Dilemma, they don't know what the other picked until after they both make their decisions. Would it result in the sponsorship failing?

[sanity]

That's a good question, and the answer will depend on exactly how we want the incentives to work.

One possibility if both sponsors defect is that both are punished with a long cooldown period, and then perhaps also with the sponsorship failing.

A higher cost for mutual defection increases the trust required for sponsorship.

I also just realized I forgot to mention that any two given sponsors can only co-sponsor together a restricted number of times - perhaps just once at one extreme, or perhaps with a co-sponsorship cooldown period much longer than the normal cooldown period. Will edit original.

[ple1n]

PoW/HashCash is too inefficient and stupid. And this is not a cryptographic problem (at least not directly solved by it)

[ple1n]

It has no use beyond cyber arms race. No one other than miners would participate in this non-sense. Yes it it totally okay to use it for cryptocurrency arms race, but even eth is transistioning now. In locutus it makes absolutely no sense, because mobile users exist and miners' hardware outperform regular PCs by orders of magnitude.

And 'a trillion dollar' isn't a valid argument. People invent stupid games and people participate in it and that's it. We have other arms races already, nuclear etc. Furthermore, reasonable solutions do exist which I have mentioned (ZKP + ML for biometric proof of personhood) and web of trust.

[sanity]

@TheRook

You are approaching one of the most difficult problems in modern cryptography and acting like I should already know that you solved the problem.

I expect nothing other than that if something is unclear you seek clarification before jumping to conclusions. This is an early brainstorming discussion, not a finished proposal.

I don't see any rigor here

That's why the title starts with the word "Concept", rigor comes later - I'm just trying to figure out of this approach is worth pursuing.

and pretending like you don't need to prove your solution to the community isn't in the best interest of any solution or any community

I didn't "pretend" anything of the sort.

I am on the top 10 of all time for security and cryptography on stackoverflow, if I can't make heads or tails of your solution - then who will?

I'm sure you're very smart, but when you see something you don't understand, you should ask for clarification rather than assuming the worst.

[ple1n]

Hmm you are still thinking about this. Apparently, the ultimate goal is to solve sybil attacks. The 'word' sybil presupposes the existence of real identities, which by definition are tied to real humans. For ensuring real identities I suggest Proof of Personhood. Down this rabbithole I found ZKP with ML which I talked about. PoP is not the complete answer tho.

Even with complete real humans, they can still be bribed/hired to attack the network. And to reach the end goal, PoP is not even a must. We can view each identity as an individual actor, and it doesn't need to be uniquely tied to a human. An ideal actor of Locutus network executes the protocol 'correctly' which is hard to define but usually means 'lack of malicious human intervention'. We can therefore call it good-will of an actor. An actor can be controlled by a human or not, and a human doesn't necessarily have good-will. What is 'good-will' is defined by the network.

I would point out that the meaning of 'trust' kinda shifted. It became a means to enforce will, which is not a bad thing.

It's possible to create an egalitarian blockchain with PoP identities and democratic block production (voting with random sampling).

[ple1n]

I also suggest not thinking too much about web of trust. It would work but it's only a means and is not going to yield anything magical

[sanity]

Is there a good, decentralized solution for proof-of-personhood? That seems more like outsourcing the problem rather than solving it.

[ple1n]

@TheRook

what is 'loci' ? I am not sure what you meant

voters can be created on demand

PoP would prevent that.

[sanity]

🤔 This is a preliminary concept discussion - we're not seeking anyone's recommendation at this stage.

[sanity]

@TheRook Get over yourself.

[ple1n]

You are not solving it either. What have you proposed ? PoW ? It was considered before the project started.

[sanity]

@TheRook You seem to think it's our job to impress you, take a break and come back when you're ready to contribute constructively.