46 total views
DeFi and cryptocurrency are not trustless, they only provide trust redistribution.
Original title: ” DeFi Trust Law: V God’s trust model is suspicious and fails to consider the metaTrust issue ”
Written by: EJ Spode
In a recent article , Vitalki Buterin made an interesting analysis of trust. Although I think his article is a good starting point for discussion, I am skeptical about his definition of trust and his views on future trust.
What is trust?
According to Vitalik’s introduction, we should understand trust as follows (for Vitalik’s view of trust, we call it vTrust).
vTrust believes that trust is the use of any assumption about the behavior of others.
For some applications, this is the perfect definition of “trust”, but it is not universally applicable, and it does not seem to apply to cryptocurrency and blockchain technology.
The problem in principle is that anticipation of someone’s behavior is not the same as trust in their behavior. For example, I expected that everyone who read this article had been to the bathroom today, but it would be strange to say that I trust them to have been to the bathroom. This is because when we trust people to do something, we not only want them to do something. Our expectation is that even if there is a lack of external motivation (and possibly even external motivation), they will follow the given norms (for example, will not run off).
For example, let’s take a recent famous DeFi case, in which a developer named Chef Nomi withdrew from his agreement Sushiswap with $14 million in developer funds (he later returned the funds). Until this exit, many people probably believed that Chef Nomi was doing the right thing, despite the fact that he was interested in cashing out ($14 million is a lot of money!). Our expectation was that he would take actions contrary to incentives. why? Because this is trust.
Let us formalize the definition of trust and call it kTrust (for Kantian trust, because the core idea is basically Kantian).
kTrust: Trust means that the agent will act in accordance with given guidelines even if he has the motivation to act (for example, he will not run away).
We will find it useful to think about the level of trust. We may believe that a developer runs for ten thousand dollars, but if external incentives increase to one million or one billion dollars, we may lose trust. When we say that we trust someone “to a certain extent,” we mean that despite the opposite external motives, we still trust that person to follow the guidelines, but there are limitations. Maybe all trust is conditional. We only trust others to a certain extent.
Given that people can and do violate trust, it is not surprising that we seek to design around the needs of others’ trust. For example, being told that a new encryption application is “trustless.” But this is only wishful thinking. Elsewhere, I think that these new agreements do not eliminate the need for trust (kTrust), they just redistribute trust. So, for example, we no longer trust bankers to verify our remittances, but we still need to trust code reviewers to reliably assess and report the integrity of the code in DeFi applications. Even talented developers must rely on auditors to verify the reliability of their code, and will not adulterate during the evaluation process.
So what does Vitalik’s view of trust tell us? The problem with Vitalik’s vision is that even if it can’t be done, it tries to design around trust. We must also be prepared to assess the ethical characteristics of the people with whom we do business. This is not an engineering problem. Or in any case, if it is an engineering problem, then it is a problem of “complete artificial intelligence”-to solve this problem, you must first solve all problems in artificial intelligence.
This brings us to Vitalik’s concept of trust model. Vitalik believes that the real problem is not to eliminate trust, but to minimize the amount of trust necessary for the system to function properly. Ideally, zero trust points will be required. In the worst case, a person depends on a person, and a person must trust that person (potential Bernie Madoff scenario). Generally, as shown in the figure below, the smaller the agent we have to trust, the better our condition.
In Vitalik’s view, we can’t actually eliminate trust, but we can guide ourselves into the green area in the picture above, thereby reducing our dependence on trust to a certain extent.
What can we say about this? I believe that the failure of Vitalik’s analysis lies in the metaTrust problem. We may have a system that requires 0 of N trusted agents, but how do we know that this is actually part of our system? We rely on code reviewers and technical friends to provide these assessments to be trustworthy. If we are not sure whether we are in such a system, a system that only requires one out of a million agents to use is useless. Therefore, we trust the auditor to confirm that our system is designed this way.
In short, this is the problem: if we don’t believe that we are in such a system, it doesn’t matter as long as we believe that the system we are in requires only one millionth of the trusted participants. Moreover, we usually only know that we are in such a system (if known) because one or a few auditors have confirmed that this is indeed the case. Indeed, we may find that the system being audited is so complicated that we actually need N of the N auditors to determine whether the system is reliable, so that we know we are in a reliable state. Therefore, the metaTrust necessary for the normal operation of the entire system will be located in the red area of the Vitalik chart.
In fact, we have cleared away the need for trust, which is difficult to see here, but it still exists, and now it is more dangerous because it is difficult to see.
Cultivate a kTrust culture
If the need for kTrust cannot be eliminated and it cannot be designed, is there anything that can be done? Yes!
Traditional finance has not failed because it has too many trust points. But it failed again because it cultivated a culture of “greed is good, trust is for schlubs”. “If you are burned to death, it is your fault.”
The problem for us is that too many people in the DeFi field have adopted this attitude. “The code is the law. If you don’t see the cash out, it is the law.” But kTrust cannot be eliminated. Generally, DeFi and cryptocurrencies are not trustless, they only provide trust redistribution.
Today’s DeFi community members have the opportunity to show that, unlike traditional finance, they value trustworthiness and plan to inculcate trustworthiness as an important criterion. The entire community must use it as a norm.
If this is not possible, then DeFi will definitely fail like current centralized finance.