Vitalik Buterin: Tradeoffs on Blockchain Coordination
Editor's Note: This article comes fromBabbitt Information (ID: bitcoin8btc), Author: Vitalik Buterin, Compiler: Free and Easy, released with authorization.
Editor's Note: This article comes from
Babbitt Information (ID: bitcoin8btc)
Babbitt Information (ID: bitcoin8btc)
, Author: Vitalik Buterin, Compiler: Free and Easy, released with authorization.；
Special thanks to Karl Floersch and Jinglan Wang for their feedback and comments；
On Collusion
Engineering Safety on Coordination Problems
trust model
The meaning of decentralization
Coordination, that is, a large group of actors working together for a common good, is one of the most powerful forces in the universe. For example, if we work together to stop global temperatures from rising, the temperature rises will be much smaller. It is this factor that makes corporations, nations, and any social organization potentially larger than a few individuals.
Coordination can be improved in many ways: faster information dissemination, better norms (to identify what behavior is classified as cheating) and more effective punishment, stronger organization, such as smart contracts that allow interactions with reduced levels of trust tools, governance techniques (voting, stocks, decision markets...) and more. In fact, we as a species have gotten better at all of these things over the past decade.
However, there is also a very counterintuitive dark side to coordination. While it is true that "everyone coordinates with everyone" results in far better results than "everyone coordinates for itself," it does not mean that every step everyone takes toward more coordination is It must be beneficial. If coordination is improved unbalanced, the results can easily be detrimental.
We can think of it as a map, but actually maps have billions of "dimensions" instead of two:
"Everyone for himself" in the lower left corner is a place we don't want to go. The "full coordination" in the upper right corner is idealized, and it may not be realized. And in the middle ground, there are plenty of reasonably safe and efficient places that might be the best places to stay, and there are plenty of deep, dark caves to avoid.
Now, what are these dangerous forms of local coordination, and when does the coordination of some individuals lead to a deep black hole? This is best illustrated by example:
Citizens of a country who heroically sacrifice themselves for the national interest in war... when that country becomes Germany or Japan in World War II;
A lobbyist bribes a politician in exchange for the politician adopting the lobbyist's favored policies;
All sellers of a product in the market collude to raise prices simultaneously;
Large blockchain miners conspire to launch a 51% attack;
In all of the above cases, we see a group of people coming together in solidarity to help each other, but this does a lot of damage to some people outside the coordinated circle, which in turn damages the net benefit of the world as a whole. In the first case, all the victims of the aggression of the above-mentioned countries were outside the scope of the coordination and suffered a heavy blow as a result. In the second and third cases it is the people affected by the decisions made by corrupt voters and politicians, in the fourth case the customers, in the fifth case non-participating miners and the blockchain user. It is not an individual who betrays the group, but a group who betrays a wider group (often the entire world).
This kind of local coordination is often referred to as "collusion," but it's important to note that we're talking about a fairly wide range of behaviors. In normal language, the term "collusion" tends to be more commonly used to describe relatively symmetrical relationships, but in the above cases, there are many examples that are strongly asymmetric. In this sense, even an extortionate relationship (“vote for my favoritism policy, or I’ll reveal something about you publicly”) is a form of collusion. In the rest of this article, we will generally use "collusion" to refer to "unwanted coordination".
Evaluate intent, not behavior (!!)
An important feature of the milder conspiracy cases is that one cannot tell whether an act is part of an unwelcome conspiracy simply by looking at the act itself. The reason is that the actions taken by an individual are a combination of the individual's internal knowledge, goals, and preferences, as well as the incentives imposed on him externally, so that actions taken by people, when they collude, are coordinated voluntarily (or in a benign way) with people ) actions often overlap.
For example, consider collusion between sellers (a violation of antitrust laws). If operating independently, each of the three sellers could price a product between $5 and $10. The difference in range reflects hard-to-understand factors such as the seller's internal costs, their own willingness to work at different wages, supply chain issues, etc. However, if the sellers collude, they may set the price between $8 and $13. Again, the range reflects different possibilities regarding internal costs and other hard-to-see factors. If you see someone selling that product for $8.75, are they doing something wrong? You can't tell without knowing if they coordinate with other sellers! Making a law that says it would be a bad idea to sell the product for more than $8 might have a valid reason why the current price would have to be high. But having a law against collusion, and successfully enforcing it, leads to the desired outcome - if the price has to be high enough to cover the seller's costs, you can get $8.75, but if the factors driving the price up Naturally low, you won't get that price.
This also applies to bribery and vote-buying cases: it is likely that some people voted Orange legally, while others voted Orange because they were paid to do so. From the point of view of those who determine the rules of the voting mechanism, they do not know in advance whether the Orange Party is good or bad. But what they do know is that a vote where people vote based on what they really feel works pretty well, but a vote where voters can buy and sell their votes works pretty badly. This is because selling votes has a tragedy of the commons: each voter only gets a small benefit from voting correctly, but if they vote the way the briber wants, they will get the full bribe. So voting, which allows the sale of votes, can quickly become a plutocracy game.
secondary titleLearn about game theoryWe can zoom out even further and look at it from a game-theoretic perspective. In versions of game theory that focus on individual choice, that is, that assume that each player makes a decision independently, and that do not allow for the possibility of groups of agents working in mutual interest for each other, there is mathematical evidence that, in any game There must be at least one stable Nash equilibrium in . In reality, the mechanic designer has a lot of freedom to "design" the game to achieve a specific outcome. But in a version of game theory that allows coalitions to cooperate (i.e. "collusion"), which we call cooperative game theory, we can show that there are many classes of games without any stable outcome (called "cores") from which coalitions cannot gain profit.
This fact, the instability of the majority game under cooperative game theory, is considered a highly underestimated simplified universal mathematical model, which explains why there is likely no "end of history" in politics, nor a system proven to be totally satisfactory. For example, I personally think it is better than the more famous Arrow's impossibility theorem (
Arrow's theorem
)it works.MACINote again that the core dichotomy here is not "individual vs group". For mechanism designers, "individuals vs groups" is very easy to deal with, it's "groups vs broader groups" that poses the challenge.
secondary title
Decentralization as anti-collusion
Following this line of thought, we have another, smarter and more actionable conclusion: if we want to build stable mechanisms, then we know that an important part of doing so is finding ways to make collusion, especially at scale Collusion is more difficult to occur and maintain. In the case of voting, we have secret ballots - a mechanism that ensures voters have no way to prove to a third party how they voted, even if they wanted to (
is a project that attempts to use cryptography to extend the secret ballot principle to a networked environment). This undermines trust between voters and bribers, greatly limiting the potential for unwelcome collusion. In cases such as antitrust, we often rely on whistleblowers, and even reward them, thereby explicitly incentivizing participants in harmful collusion to defect. In terms of broader public infrastructure, we have a very important concept: decentralization.
A naive view is that decentralization is valuable because it reduces the risk of single points of technical failure. In traditional "enterprise" distributed systems, this is often true, but in many other cases, we know it's not enough to explain what's going on. Looking at the blockchain from here is instructive, and a large mining pool publicly demonstrating how they distribute nodes and network dependencies internally doesn't quell community members' fears of mining centralization. The picture shows some of the people who controlled 90% of the Bitcoin hashrate at the time, at a roundtable at the same conference, and it really freaked people out:
So why is this picture scary? From a "decentralization is fault tolerance" point of view, it doesn't hurt for large miners to be able to talk to each other. But if we think of "decentralization" as barriers to harmful collusion, the picture becomes quite scary, as it shows that these barriers are not as strong as we think. The fact that these miners can easily coordinate technically, and that they may all be in the same WeChat group, actually means that Bitcoin is "not much better than a centralized company."
So what obstacles remain to collusion? Some major ones include:
Moral barriers, as Bruce Schneier reminds us in his book Liars and Outsiders, many "safety systems" also serve a moral function, alerting potential misbehavers that they are about to commit Serious offense, if they want to be a good person, they shouldn't be doing it. Decentralization arguably helps with this function.
Anti-coordination, the fact that the system is decentralized, makes it easy for participants not involved in the collusion to make a fork, drive out the colluding attacker, and continue the system from there, with low barriers for users to join the fork.
The risk of defection is that it is much harder for five companies to band together to do something that is widely viewed as bad than it is for a non-controversial or benign purpose. The five companies don't know each other well, so there is a risk that one of the companies refuses to participate and quickly whistles, and it is difficult for participants to judge the risk. In addition, individual employees within the company may also report.
Taken together, these hurdles are indeed formidable — often enough to deter a potential attack, even if all five companies are simultaneously fully capable of coordinating quickly to do legitimate things. For example, Ethereum blockchain miners are perfectly capable of coordinating gas cap increases, but that doesn't mean they can easily collude to attack the blockchain.
Experience with blockchains has shown that designing protocols as decentralized institutions is often a very valuable thing to do, even if it is known in advance that most of the activity will be dominated by a small number of companies. This idea is not limited to blockchains, it can be applied in other contexts as well (see here for antitrust applications, for example).
secondary title
Fork is anti-coordination
We can use two core operating principles to achieve this goal: (1) enabling anti-coordination and (2) risk sharing. The idea behind anti-coordination is this: we know we cannot design systems that are passively robust to collusion, in large part because there are so many ways for organizations to collude and there are no passive mechanisms that can detect them, But what we can do is actively address collusion and fight back.
In digital systems such as blockchains (and this can also apply to more mainstream systems such as DNS), a major and crucial form of anti-coordination is forking.
If a system is taken over by a harmful coalition, dissidents can get together and create an alternate version of the system that has (mostly) the same rules, except that it removes the need for the attacking coalition to control the system that power. Forking is very easy in an open source software environment. The main challenge in creating a successful fork is usually gathering legitimacy (a form of "common sense" in game theory) to get everyone who disagrees with the direction of the main coalition to follow you.
secondary title
Market and Risk Sharing
Forking is a powerful form of anti-coordination precisely because it introduces a risk-sharing mechanism. In Hive, the community fork of Steem gave up the hostile takeover attempt, and those coins that voted for it were deleted in the fork, which means that the key people involved in the attack suffered losses.
secondary title
Structural coordination
All of this gives us an interesting look at the behavior of the people who build social systems. One of the goals of building an efficient social system largely determines the structure of coordination: which groups of people and which configurations can come together to achieve their group goals, and which groups can't?
image description
Different Coordination Structures, Different Results
Sometimes better coordination is a good thing: people can work together to solve their problems together. But at other times, more coordination is dangerous: some players can coordinate to disenfranchise everyone else. At other times, more coordination is needed for another reason: to enable the wider community to "fight back" to prevent collusion to attack the system.
In all three cases, different mechanisms are available for these purposes. Of course, it is very difficult to block communication directly, and it is very difficult to make coordination perfect. However, there are many choices in between that can have powerful effects.
Here are some possible coordination structure techniques:
Privacy protection technology and norms;
Technical means that make it difficult to prove your actions (secret ballots, MACI and similar techniques);
deliberate decentralization, assigning control of certain mechanisms to a large group of people who all know they are not coordinating well;
Decentralization of physical space, separating different functions (or different shares of the same function) into different locations (see, for example, Samo Burja on the link between urban and political decentralization);
Separation of different functions (or different shares of the same function) to different types of participants (e.g. "core developers", "miners", "coin holders", "application developers" in a blockchain ","user");
Schelling points, which allow large numbers of people to coordinate rapidly around a single path;
