We’re talking about at least two different notions of the word “rational”:
Robin Hanson used the definition at the top of this post, regarding believing the truth. There are social/evolutionary costs to that, partly because humans lie poorly.
The causal decision theorists’ definition that Eliezer Yudkowsky was annoyed by. CDT defines rationality to be a specific method of deciding what action to take, even though this leads to two-boxing (losing) Newcomb’s problem. Yudkowsky’s objection, summarized by the slogan “Rationalists should WIN.” was NOT a definition. It is a quality of his informal concept of rationality which the CDT definition failed to capture.
The claim “rationalists should always win” comes from taking Yudkowsky’s slogan as a definition of rationality. If that is the definition that you are using, then the claim is tautological.
Please note that I don’t endorse this misreading of Yudkowsky’s post, I’m just trying to answer your question.
As you say, defining rationality as winning and then saying rationalists always win is a tautology. But aside from your two definitions, there’s a third definition: the common definition of rationality as basing decisions on evidence, Bayes, and logic. So as I see it, supporters of “rationalists always win” need to do one of the following:
Show that the winning definition is the same as the Bayes/logic/evidence definition. Tim’s counterexample of the religious believer who’s a poor liar makes me doubt this is possible.
Stop using “rationality” to refer to things like the Twelve Virtues and Bayesian techniques, since these virtues and techniques sometimes lose and are therefore not always rational.
Abandon “rationalists always win” in favor of Robin’s “rationalists always seek the truth”. I think that definition is sufficient to demonstrate that a rationalist should one-box on Newcombe’s problem anyway. After all, if it’s true that one boxing is the better result, a seeker of truth should realize that and decide to one-box.
There are no supporters of “rationalists always win” – the slogan is “rationalists should win”. Long-term / on-average, it’s rational to expect a high correlation between rationality and success.
[1] – I’d bet that the rationalist strategy fares well against other heuristics; let’s devise a good test. There may always be an effective upper-bound to the returns to increasing rationality in any community, but reality is dangerous – I’d expect rationalists to fair better.
[2] – Winning or losing one ‘round’ isn’t sufficient grounds to declare a strategy, or particular decisions, as being non-rational. Buying lottery tickets isn’t rational because some people win. And sometimes, winning isn’t possible.
[3] – I like “rationalists always seek the truth” but would add ”… but they don’t seek all truths.”
We’re talking about at least two different notions of the word “rational”:
Robin Hanson used the definition at the top of this post, regarding believing the truth. There are social/evolutionary costs to that, partly because humans lie poorly.
The causal decision theorists’ definition that Eliezer Yudkowsky was annoyed by. CDT defines rationality to be a specific method of deciding what action to take, even though this leads to two-boxing (losing) Newcomb’s problem. Yudkowsky’s objection, summarized by the slogan “Rationalists should WIN.” was NOT a definition. It is a quality of his informal concept of rationality which the CDT definition failed to capture.
The claim “rationalists should always win” comes from taking Yudkowsky’s slogan as a definition of rationality. If that is the definition that you are using, then the claim is tautological.
Please note that I don’t endorse this misreading of Yudkowsky’s post, I’m just trying to answer your question.
Thanks, John.
As you say, defining rationality as winning and then saying rationalists always win is a tautology. But aside from your two definitions, there’s a third definition: the common definition of rationality as basing decisions on evidence, Bayes, and logic. So as I see it, supporters of “rationalists always win” need to do one of the following:
Show that the winning definition is the same as the Bayes/logic/evidence definition. Tim’s counterexample of the religious believer who’s a poor liar makes me doubt this is possible.
Stop using “rationality” to refer to things like the Twelve Virtues and Bayesian techniques, since these virtues and techniques sometimes lose and are therefore not always rational.
Abandon “rationalists always win” in favor of Robin’s “rationalists always seek the truth”. I think that definition is sufficient to demonstrate that a rationalist should one-box on Newcombe’s problem anyway. After all, if it’s true that one boxing is the better result, a seeker of truth should realize that and decide to one-box.
There are no supporters of “rationalists always win” – the slogan is “rationalists should win”. Long-term / on-average, it’s rational to expect a high correlation between rationality and success.
[1] – I’d bet that the rationalist strategy fares well against other heuristics; let’s devise a good test. There may always be an effective upper-bound to the returns to increasing rationality in any community, but reality is dangerous – I’d expect rationalists to fair better.
[2] – Winning or losing one ‘round’ isn’t sufficient grounds to declare a strategy, or particular decisions, as being non-rational. Buying lottery tickets isn’t rational because some people win. And sometimes, winning isn’t possible.
[3] – I like “rationalists always seek the truth” but would add ”… but they don’t seek all truths.”