Exchange Ranking
Top Exchanges
24h Volume Ranking
Popularity Ranking
Exchange BTC Balance
Proof of Reserves
Decentralized Exchanges
Funding Rate
Funding Heatmap
Liquidation Data
Max Pain
Long/Short Ratio
Whale L/S Ratio
Binance/Okex/Huobi L/S
Bitfinex Margin L/S
ETF Tracker
Solana ETF
XRP ETF
Hong Kong ETF
Bitcoin Treasuries
Crypto Reversal
Ethereum Reserves
HyperLiquid Wallet Analysis
Hyperliquid Whale Watch
Large Transactions
On-chain Movement
Bitcoin ROI
Stablecoin Market Cap
Options Analysis
News
Articles
Economic Calendar
Features
Wallet
Contract Calculator
Security
Collections
Watchlist
Following
ETH Vitalik Buterin: Adversarial governance design and artificial intelligence security
Vitalik Buterin, co-founder of Ethereum, pointed out that adversarial governance design helps solve artificial intelligence security issues, and its core is to limit collusion between advanced artificial intelligence agents. He argued that decentralization, secret voting and privacy protection tools could build a barrier against harmful collusion.
Commonality between governance structures and AI security
In a recent analysis, Butrin compared the governance system with an artificial intelligence environment involving "weaker principals" and "stronger agents." He pointed out that in both scenarios, limiting the degree of coordination between agents can improve the end result.
Specifically, Buterin describes the structural features shared between governance and AI security:
- In a governance system: Static algorithms act as principals, while humans act as more complex agents.
- In the AI security arena: Humans and weaker language models can be regarded as principals, while powerful large language models can be regarded as proxies.
It is worth noting that Buterin's focus is not on the independent actions of individuals, but on the coordination mechanism between agents. He believes that governance design can produce better results when the system can effectively limit the collusion ability between agents.
The boundary between beneficial coordination and harmful collusion
This distinction is crucial because it defines the boundary between "beneficial coordination" and "harmful collusion." It is necessary for groups to cooperate towards common goals, but the formation of certain alliances may harm the interests of non-member groups.
Buterin cited multiple cases of harmful coordination, including voting buying and selling in elections, price-fixing agreements, and a joint 51% attack by blockchain miners. He emphasized that it is often difficult to determine whether harmful coordination exists based on a single act. For example, a seller's high-priced strategy may be an independent decision or the result of collusion with other competitors.
However, rules for collusion itself can effectively curb such risks. Butrin also pointed out that buying and selling votes creates incentives to push the voting system towards "Plutocracy." This issue is closely related to cooperative game theory, which examines patterns of collaborative group action. Butrin said that because the alliance can repeatedly profit by constantly adjusting strategies, some games lack stable equilibrium results.
Decentralization as a defense
In order to deal with the risks brought by large-scale collusion, Buterin proposed a variety of defense mechanisms:
1. Decentralization and bifurcation mechanisms
Decentralization is one of the key methods to build anti-collusion barriers. In blockchain systems, if harmful alliances control the network, the fork mechanism can support "reverse coordination." By creating a competitive version, you can eliminate the influence of attack alliances while retaining most of the original rules.
2. "Skin in the Game"
Butrin emphasized that "skin pain" is another important defense. Market mechanisms can make participants individually accountable for their decisions, increasing the costs and risks of collusion.
3. Other Coordination Limiting Tools
In addition, he listed a variety of tools to limit excessive coordination, including:
- One-person, one-vote system
- Physical isolation measures
- Role-based constituency delineation
- Schelling point The application of Schelling points
- encourages defectors to expose premeditated collusion
At the same time, secret voting, privacy tools, whistleblower incentive mechanisms and internal negotiation difficulties are also seen as effective ways to block large-scale collusion.
