I've been reading VA Classical Liberal's exploration of Atlas Shrugged, and it's left me thinking about the underpinnings of Rand's philosophy of rational selfishness.
As someone with a mathematical background, I find myself approaching the question as a matter of limits. What is the steady-state for rational selfishness? Does cooperation make sense in this framework? Does altruism? Or is selfishness the only true virtue, as Ayn Rand seems to imply?
Follow me below the fold for more.
First, some background on me and Rand. I read Anthem in middle school, and as a sci-fi junkie and a "rugged individualist", I loved it. I read The Fountainhead in high school, and while I found Rand's portrayal of formulaic individualism fundamentally flawed, again, I enjoyed the book. I never got around to reading Atlas Shrugged, and somewhere along the line - probably around when I became Catholic - I started understanding that individualism wasn't all it was cracked up to be.
So what swayed me about the 'virtues' of selfishness and individualism? A simple consideration of what values would actually generate the best outcomes for myself.
As I said, I'd like to consider the limit cases of rational selfishness. Let's say an individual had the ability to see the full ramifications of their choices, the ability to know precisely how much any decision would result in individual benefit to himself. We can consider three individual choices - (1) the individual acts to further his own interests without regard to how those acts will impact others (selfish); (2) the individual acts to further the interests of others without regard to how those acts will impact himself (altruist); or (3) the individual acts to further the interests of both himself and others, even if less net benefit will be accrued than if only one party's interests were considered (compromise). We will not consider other choices, since choices which are in no one's interest are unlikely to generate the best outcomes for the individual.
The question, then, is this. With perfect knowledge, which decision strategy maximizes individual success: selfishness, altruism, compromise, or some combination thereof?
This seems like it might be a good time to introduce an old Game Theory sawhorse - The Prisoner's Dilemma. The Dilemma is essentially a treatment of selfish vs. altruistic decision principles. Summarized, it works as follows:
Imagine that you and an accomplice rob a convenience store. During the robbery, the store clerk pulls a gun. You and your accomplice wrestle him to the ground, but in the ensuing three-man struggle, the gun goes off and kills the store clerk. Then the police arrive and arrest both you and your accomplice.
The district attorney has ample evidence to convict you both of robbery, which will mean a 1-year jail sentence. But it turns out that, with the chaos of the fight, the DA doesn't have enough evidence to pin the murder on either you or your accomplice. So the DA decides to try to cut a deal with both of you. You can either agree to testify that your accomplice murdered the clerk, or you can keep silent. Your accomplice is given the same decision.
If you choose to keep silent, your accomplice will go free on the murder charge - the DA doesn't have enough evidence to bring a case without a witness to testify. If you choose to rat out your accomplice, things are a little more complicated. Here's the breakdown of how this can fall out.
Say you both refuse to testify. Then the DA will get you both on robbery charges, but no one will be convicted of the murder.
Say one of you refuses to testify, but the other rats out his accomplice. Then the snitch will go free, and the one who doesn't talk will get saddled with the murder and serve an additional 30 years in prison.
Say you both rat on each other. Then the DA is going to make a decision on which one of you to prosecute for the murder and which one of you to cut a deal with. In this eventuality, there's a 50% chance you'll wind up convicted for the murder and the robbery, and a 50% chance you'll get off with only the robbery conviction.
So again, if you choose to keep silent, it means your accomplice won't be charged with murder - but you might. If you talk, then there's still a 50/50 chance you might get charged if your accomplice also talks.
Any analysis of the probabilities suggests that the safest move is to sell out your accomplice. If they keep silent, you'll always avoid the rap for murder. If they talk, you'll have a 50/50 chance. If you choose to keep silent, however, your probability of getting convicted based on what your accomplice does goes up. We can look at some simple probability expressions to see that this is true.
Let P(murder) be the probability you get convicted for murder. P(murder|rat) and P(murder|silent) are the probabilities you'll take the rap for killing the clerk if you (a) rat out your accomplice or (b) stay silent about what happened, respectively. P(a->rat) is the probability that your accomplice rats you out, and P(a->silent) is the probability that your accomplice keeps quiet. Then:
P(murder|rat) = 0*P(a->silent) + .5*(a->rat) P(murder|silent) = 0*P(a->silent) + 1*(a->rat)
To clarify, the probability of getting convicted of murder is always 0 if your accomplice keeps silent. If your accomplice rats you out, then you will take the rap if you kept silent. But if you ratted him out, then you only have a 50/50 chance of taking the rap yourself.
So what's the best outcome? If you both keep silent, neither of you gets convicted of murder. But what's the best choice? No matter what your accomplice does, you're safer selling him out than trying to protect him.
Does that make sense? Stop and think about it for a little while. What all this really means, is that your fate is in the hands of your accomplice and what he chooses to do. And, fundamentally, whether he believes you're willing to do what it takes to look out for him. If he believes you'll look out for him, then he won't rat on you. And if he doesn't rat on you, then you don't get convicted of murder. But if he doesn't believe you'll look out for him, your accomplice will do what's rationally selfish - sell you out to decrease the odds he'll go down for the murder.
So what do you do? Trust your accomplice to have your back, and hope you both evade murder charges? Or look out for yourself and hope that either your accomplice is more selfless or you get lucky with the DA?
This is the Prisoner's Dilemma. Does it make more sense to cooperate (possibly against your own best interests), or look out for yourself (and hope your accomplice doesn't do the same)? You can find ample discussion about this on-line, but the point I want to make is that rational selfishness - only looking out for oneself - does not necessarily produce the most favorable results.
It is possible to see a number of real-life examples of this sort of dilemma. International regulation of greenhouse gases marks one particularly good example. It is in the interest of all nations to reduce CO2 emissions, but any nation refusing to reduce emissions receives a partial benefit of the nations that do reduce emissions, as well as an individual production benefit from not regulating their own pollutants. (This is the classic 'free rider' problem)
So selfishness seems to be out as an optimal decision-making strategy. Selfishness reaps maximal gains only when the other actors with which a selfish individual interacts are not themselves selfish. If the selfish individual cannot take advantage of other, more generous actors, then the selfish individual's success is limited to only what the selfish individual is capable of achieving in isolation.
How about altruism?
A cursory inspection should reveal that pure altruism is at least as bad a strategy as selfishness, since any altruistic actor will be taken advantage of by any selfish actor. The altruistic actor will accomplish very little for himself unless he can get cooperation from others, and other actors will be inclined to betray the always-altruistic actor to maximize their individual gains.
This seems to point the way towards compromise, and most solutions of the iterative Prisoner's Dilemma indicate that a long-run compromise plan works best. Sometimes the individual looks out for himself. More often, the individual acts for the benefit of his partner, to the benefit of both actors. Specific traits that successful algorithms have in common are given on Wikipedia's Prisoner's Dilemma page (linked above).
But there is another strategy, called superrational. Doesn't that sound like a label Ayn Rand would love to apply to her characters? The superrational strategy is simple, and revolves around the idea that all superrational actors will have a single, unique superrational strategy for interactions. Somewhat simplified (admittedly for the benefit of my argument), this superrational strategy is quite basic:
If the partnered actor is known to be superrational, always cooperate.
In other words, if you know your partner knows the value of cooperation, always cooperate. If you know that your partner knows the value of altruism, be altruistic to him.
This has been my personal attitude for a long time, now - that the only problem with the always-altruistic decision strategy is that there are individuals who will betray the always-altruistic actor to reap short-term benefits. But as a whole, the superior choice is to look after the interests of others and let them look after your own interests.
Ayn Rand's philosophy of always looking out for yourself leads to an anarcho-capitalist world where corporate action is impossible because the individual can always reap greater rewards through betrayal - unless everyone else is an anarchy-capitalist, in which case you have stagnation as no one is willing to put their own success on the line and risk being taken advantage of.
On the other hand, the endpoint of superrationality leads to a structure of perfect cooperation where - yes - one bad egg (selfish actor) can reap tremendous rewards by exploiting others, but everyone is much better off for having the capacity for large-scale cooperation.
My point is this - Ayn Rand's loathing of altruism and perverse focus on the 'virtue' of selfishness is myopically misguided. In the long run, it leads to worse outcomes for everyone. Even the John Galts of the world benefit from cooperation with others, from sublimating their own selfishness and helping others, with the expectation that those others will help them in return. No individual is so accomplished that they cannot benefit from collaboration with other individuals of a similar level.
Which leaves Ayn Rand as what? Just another name in the litany of failed social theorists. A footnote to the fall of American Conservatism. And a lesson to us - that when our opponents take every opportunity to betray us, we must strive harder than ever to find common cause amongst ourselves, to demonstrate the superrationality of cooperation.