Rationality is Self-Defeating in Permissionless Systems

Bryan Ford, Rainer BöhmearXiv 2019 · OpenReview

TL;DR

This paper argues that designing permissionless decentralized systems (like Bitcoin) to be secure against "rational" attackers is fundamentally flawed. A rational attacker can exploit the system's openness and external financial markets to profit from attacking it, even if the attack itself appears irrational within the system's context. This means such systems are only secure if they can withstand fully Byzantine (arbitrary) attacks, rendering rationality assumptions useless for security.

What Problem Does This Paper Address?

Many decentralized systems, particularly in the blockchain and cryptocurrency space, are designed with the assumption that participants will act rationally, meaning they will deviate from protocol rules only if it benefits them economically. This is a weaker assumption than the "Byzantine" model, where participants can act arbitrarily, even maliciously, without regard for their own economic well-being. The problem is that while the Byzantine model is notoriously difficult to design for, the rationality assumption is often seen as a more tractable path to security and efficiency. However, this paper questions whether this assumption truly provides security in open, permissionless systems. The authors highlight a gap in understanding how the external economic environment can incentivize seemingly irrational behavior within a system, thereby undermining its security.

Methods

The paper presents a "metacircular argument" to demonstrate the self-defeating nature of rationality assumptions in permissionless systems. The core of the argument relies on three key assumptions:

  1. Openness and Permissionlessness: Anyone can join the system with weak, anonymous identities.
  2. External Economic Power: The system does not control a majority of global economic power; external actors can muster comparable resources.
  3. Leveraged Assumptions: The system's security relies on specific behavioral assumptions (like rationality), and there exists a strategy that violates these assumptions, causing observable failure.

The argument then posits a rational attacker who, leveraging external financial markets (like derivatives or short-selling on another cryptocurrency), can profit from causing the target system to fail. This attack is rational from the attacker's perspective in the broader "open world" but appears irrational within the confines of the target system's assumed threat model.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Rationality is a Weak Assumption in Open Systems: The paper's central thesis is that assuming rationality for attackers in open, permissionless systems is a false sense of security.
  • The "Open World" Advantage: External financial markets and systems provide a crucial avenue for rational attackers to profit from attacking a target system, even if the attack itself is economically detrimental within the target system.
  • Metacircular Self-Defeat: The argument is "metacircular" because the system's reliance on rationality assumptions creates the very conditions for a rational attacker to exploit those assumptions.
  • Byzantine is the Only Real Security: For open, permissionless systems, true security can only be achieved by designing them to withstand arbitrary (Byzantine) behavior, as rationality assumptions are ultimately rendered moot.
  • Real-World Parallels: The authors draw parallels to the 2007-2008 financial crisis and the use of credit default swaps, illustrating how external financial instruments can be used to profit from systemic failures.

Key Results

The paper doesn't present quantitative results in the traditional sense, as it's a theoretical argument. The key "result" is the logical demonstration that:

  • Any open, permissionless system relying on rationality assumptions for security is inherently vulnerable.
  • A rational attacker can always find an economically rational strategy to profit from attacking such a system by leveraging external markets.
  • This implies that systems must either be closed (with strong identities and legal recourse) or designed to be Byzantine-resilient to be truly secure.

Limitations and Future Work

The authors acknowledge that the argument is presented as a "rough and informal sketch" and that formalization would require significant effort. They also note that the argument does not apply to all systems:

  • Permissioned Systems: Systems with strong identities and legal frameworks (like national cryptocurrencies or corporate systems) might be able to enforce rationality through legal means.
  • Weak Rationality Assumptions: Systems that use rationality primarily for initial growth (e.g., mining rewards) but then rely on other mechanisms for internal security might be less affected.

The paper suggests that future work could involve formalizing this metacircular argument. It also implies that the practical implication is to either embrace Byzantine fault tolerance or restrict system openness and identity requirements.

Summary updated

Abstract

We outline a metacircular argument explaining why it is rational to be irrational when attacking open-world decentralized systems, and why systems whose security depend on rationality assumptions are insecure.
BibTeX
@misc{ford2019rationality,
  author = {Bryan Ford and Rainer Böhme},
  title = {{Rationality is Self-Defeating in Permissionless Systems}},
  year = {2019},
  eprint = {1910.08820},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1910.08820},
}

Loading related content...

Figure