Decision Theory
Acausal
Trade
How rational agents can coordinate without communicating - through shared reasoning patterns that transcend causality.
The Core Insight
"If two perfect reasoners face identical situations, they will reach identical conclusions. This logical correlation can substitute for causal connection - enabling coordination, cooperation, and even 'trade' between agents who never communicate."
You and a stranger are placed in separate rooms. No communication allowed. You each must choose: cooperate or defect. If both cooperate, you both win. If one defects while the other cooperates, the defector wins big and the cooperator loses. If both defect, you both lose a little.
Standard game theory says you should defect. Your choice doesn't affect theirs, so defecting gives you the best outcome regardless of what they do. This logic is airtight - unless you know something special about your opponent.
What if your opponent is your identical twin? Your clone? A copy of your brain running on different hardware? Now the logic changes. Whatever reasoning leads you to your decision will lead them to the same decision. The outcomes "I cooperate, they defect" and "I defect, they cooperate" are no longer reachable.
This is the foundation of acausal trade.
Playing Against Yourself
The clearest example of acausal trade is the Prisoner's Dilemma played against a perfect clone. Experience it yourself - and notice how the logic of your decision changes when you know your opponent reasons exactly like you.
The Setup
You are playing the Prisoner's Dilemma against a perfect clone of yourself. Your clone has your exact brain structure, memories, and reasoning patterns. You cannot communicate, but you know your clone will reason identically to you.
Whatever logical process leads you to your decision will lead your clone to the same decision.
Payoff Matrix
Two Kinds of Correlation
The key to understanding acausal trade is distinguishing between causal and logical correlation. Causal correlation flows through physical channels. Logical correlation arises from shared computation.
Causal Correlation
Standard causation: A sends a message to B, B receives it and responds. Information propagates through space at finite speed. If A and B are separated, there's a delay. A's action causes B's reaction.
Hofstadter's Superrationality
Douglas Hofstadter introduced the concept of "superrationality" in 1983. A superrational agent recognizes that other superrational agents will reason identically, allowing coordination on mutually beneficial outcomes.
Douglas Hofstadter's Superrationality
If all players are rational and know all players are rational, they recognize that their reasoning processes are equivalent. A superrational agent asks: "What should agents like me do?" rather than "What should I do given what others might do?"
Configure the Game
Total payoff: 6 points
Timeless Decision Theory
Eliezer Yudkowsky developed Timeless Decision Theory (TDT) to formalize acausal reasoning. TDT says: decide as if you're choosing the output of your decision algorithm for all instances of that algorithm, including simulations and predictions.
Newcomb's Problem
Omega, a near-perfect predictor, has put $1000 in a transparent box. It has also either put $1M in an opaque box or left it empty, based on its prediction of your choice.
Coordination Without Communication
Acausal trade becomes powerful in multi-agent settings. When many agents need to coordinate on a threshold - like a public goods game - acausal reasoning can solve problems that defeat classical game theory.
The Public Goods Game
N agents must decide whether to contribute to a public project. If at least K agents contribute, everyone benefits. Each agent pays a cost to contribute. Without communication, how do they coordinate?
The Decision Theory Landscape
Different decision theories give different answers to acausal scenarios. Understanding this landscape is crucial for evaluating when acausal trade makes sense.
Causal Decision Theory
Core Principle
Choose the action with the best expected causal consequences.
On Clone PD
Defect - your choice doesn't causally affect your clone.
On Newcomb's
Two-box - taking both boxes can't causally affect box contents.
Known Problem
Loses against predictors, gets mutual defection against clones.
Why This Matters for Acausal Trade
Acausal trade relies on decision theories like TDT or FDT that recognize logical correlations between agents. CDT cannot support acausal trade because it only considers causal consequences. The debate over which decision theory is correct is ongoing in philosophy and AI safety research.
Objections and Responses
Acausal trade is controversial. Critics argue it relies on dubious assumptions about decision theory and agent similarity. Explore the main objections.
Acausal trade is controversial. Explore the main objections and how proponents respond.
Real-World Applications
Acausal trade has implications beyond toy problems:
AI Alignment
Future AIs might coordinate with each other - or with us - through shared reasoning patterns
Simulated Beings
If we live in a simulation, we might "trade" with our simulators through their predictions of us
Moral Philosophy
"Act only as you would will all similar agents to act" - a decision-theoretic Kantian imperative
Game Theory
Programs that share source code can "inspect" each other and achieve mutual cooperation
Whether these applications pan out depends on unresolved questions in decision theory and the nature of agency.
"The universe doesn't care about the timing of when you made your decision. All that matters is the logical structure of your choice - and that structure can correlate with other instances of itself, across space and time."
Explore Related Topics
Acausal trade connects to other deep ideas in decision theory and philosophy.
References: Hofstadter (1983), Yudkowsky (2010), Soares & Fallenstein (2017)