Thursday 4 April 2013

Thought experiments and taking offence

I love Brad DeLong's academic work. He's way smarter than me and, more importantly, clearly works much much harder than I do. And he tackles interesting questions. But every time I check his blog, I get an awful "Everyone in the world is evil or stupid or both except Brad and a few of his friends" vibe.

I'd not been there for a while, but switching to The Old Reader (for now) from Google Reader messed up my RSS habits. And so, there I was, looking at Brad DeLong saying that Steven Landsburg is the stupidest man alive and that "the University of Rochester has a big problem" - presumably Landsburg's continued employment there.

What was the cause? A piece at Gawker taking great offence at a thought experiment Landsburg proposed. I'll link Gawker at the end so you're not tempted to read it before reading the original Landsburg piece.

In grad school, we had a lot of fun with thought experiments of this sort. The classic one is Nozick's experience machine. Step into the machine and you'll experience a simulated life much better than the one you'd otherwise live; moreover, you'll never remember that you're actually in the machine. If you stay out of the machine, there has to be something that matters more to you than experienced utility. 

Tyler Cowen liked to ask a variant on it in our Economics & Philosophy class: World B is identical to World A, as far as you are ever able to observe, but in World B, your wife has been cheating on you for years and you never ever knew it, nor will you ever know it. The worlds are identical except for the unknown-to-you fact of your wife's infidelity. Are you worse off in World B? If so, clearly state exactly how, and make that consistent with other things you believe about utility. 

So, what was Landsburg's offensive thought experiment? Recall that the rules of thought experiment club are that you don't add in auxiliary assumptions but stick to what's stated in the thought experiment. Landsburg asked a series of three questions, then wanted to know why our answers to 1 and 2 might differ from our answer to 3. Here they are.
Farnsworth McCrankypants just hates the idea that someone, somewhere might be looking at pornography. It’s not that he thinks porn causes bad behavior; it’s just the idea of other people’s viewing habits that causes him deep psychic distress. Ought Farnsworth’s preferences be weighed in the balance when we make public policy? In other words, is the psychic harm to Farnsworth an argument for discouraging pornography through, say, taxation or regulation?
That's scenario 1. Most economists just ignore that psychic harm - the world's essentially impossible to evaluate when we add this kind of thing in. Further, Farnsworth could pay other people not to watch pornography if he really cared about it that much. But we could assume that away to stick within the proper confines of the thought experiment: say transactions costs prevent it. Is pyschic harm of this sort admissible in the utilitarian calculus? Here's scenario 2:
Granola McMustardseed just hates the idea that someone, somewhere might be altering the natural state of a wilderness area. It’s not that Granola ever plans to visit that area or to derive any other direct benefits from it; it’s just the idea of wilderness desecration that causes her deep psychic distress. Ought Granola’s preferences be weighed in the balance when we make public policy? In other words, is the psychic harm to Granola an argument for discouraging, say, oil drilling in Alaska, either through taxes or regulation?
Actually, policy does weigh Granola's concerns. It's counted as existence value over and above option value or use value. Sound analyses don't put much weight on it, but it does sometimes count. If I get existence value from thinking heroic Randian thoughts about oil derricks and man's mastery of nature, that gets ignored in the cost-benefit analysis for some reason. But, again, all these kinds of pyschic distress are treated pretty dismissively in economics. And now the third and controversial question:
Let’s suppose that you, or I, or someone we love, or someone we care about from afar, is raped while unconscious in a way that causes no direct physical harm — no injury, no pregnancy, no disease transmission. (Note: The Steubenville rape victim, according to all the accounts I’ve read, was not even aware that she’d been sexually assaulted until she learned about it from the Internet some days later.) Despite the lack of physical damage, we are shocked, appalled and horrified at the thought of being treated in this way, and suffer deep trauma as a result. Ought the law discourage such acts of rape? Should they be illegal?
If we take this as a parallel thought experiment, the only trauma here allowable is the psychic distress, which we otherwise typically ignore.

It is a hard question and so a good one. All our intuitions tell us to condemn the third scenario while dismissing the psychic harms in the first two. But if we stick within the confines of the thought experiment, it's hard to distinguish the cases. We can say the psychic harm is worse in the third case, and it would be in the real world, but it's not hard to have a thought experiment Granola McMustardseed who gets more psychic harm from oil drilling than from being in Scenario 3. Or a Scenario 3 victim who never learns that it happened - a case pretty close to Cowen's World A vs World B.

I don't have any great answer other than that when we step away from the thought experiment and into the real world, a rule allowing Scenario 3 that imposes psychic harm no greater than that imposed in Scenarios 1 & 2 would also necessarily allow much much greater harm because we cannot set rules only allowing Scenario 3. But that's a cop-out, because even if we could do it in the real world, I'd still want it banned - in the same way that I think I'm worse off in Cowen's World B and that I wouldn't want to step into Nozick's machine. My maximand isn't just experienced utility.

Meanwhile, Gawker turns it into a story about how Landsburg thinks rape is OK and DeLong signs onto their interpretation. The contrast between the quality of comments at Landsburg's post and DeLong's is interesting too... Landsburg's commenters wrestle with a difficult thought experiment; DeLong's want Landsburg fired.


  1. I guess the only real difference is that there is an actual physical violation of the body of the victim in Scenario 3, regardless of whether he/she was aware of it or if any physical harm ensued. If we are to treat it solely as a thought experiment then that can be the only distinguishing factor which could determine the level of outrage. Non-consensual sexual contact is quite rightly held to be a violation of the individual regardless of their ability to grant consent or to be aware of the act.

    I guess some folk would get a lot more hot under the collar knowing that porn was being viewed somewhere, but personally I'm more inclined to be concerned about environmental desecration - there is actual damage occurring in this case as there is no reasonable case you could make which could convince me that a forest consented to be chopped down, for example. Whereas it is not unreasonable to imagine a certain level of consent in the porn industry.

  2. Landsburg points out that the law says nothing about my shining a flashlight at you and photons entering your body. Or my breathing near you and you breathing my air.

    Note that I agree with you though! It points to some of the inadequacies with some versions of utilitarianism.

  3. I'm glad to see you picked this up. I stumbled across it today, too. My reactions have been:

    - I thought it was a provocative thought experiment, much like the 'can you sell yourself into slavery' question, but others are clearly offended by it. Is it a function of my privilege as a white male that I have the one reaction and not the other?

    - The violation of the bodily boundaries -- and not on a sub-atomic level -- is clearly germane. It makes a different to how I/we react to the situation. Why is that?

    - It seems like a Kantian issue. That is, we don't want people treating others as means, not ends. I wouldn't be allow to create pornography starring Farnsworth, for example, even though he'll never see it.

  4. I just try to imagine myself as the victim in such cases and ignore the privilege stuff; you can't think if you're always second-guessing whether you're allowed to think, etc.

    Alternative thought experiment:

    A: You're rich. Someone steals a loaf of bread from you and consequently survives; he otherwise would have died. You never notice the loss. Are you worse off?

    B: You have two kidneys. Someone steals one from you and consequently survives; he otherwise would have died. You never notice the loss - it was done while you were undergoing other surgery, you never had any adverse side effects. Are you worse off?

    C: Someone who never would have had sexual contact with another person rapes you in your sleep; he consequently regains the will to live, doesn't kill himself, and has a great life. You never find out it's happened and you suffer no adverse consequence. Are you worse off?

  5. A - yes, I paid for that loaf of bread dammit!!! ;)

    B - effectively no, although I would have appreciated being asked first, chances are I would have said yes to the kidney.

    C - again, effectively no, although unlike B above I probably would not have consented. I'm strongly heterosexual after all :)

  6. Thought experiments usually ask, Why are these things different?

    Correct interpretation: why are these things different?
    Idiot interpretation: these things aren't different.

  7. To the question of why people take offence, I think it may be due to Landsburg's style more than his use of thought experiments as such. The use of names such as "Granola McMustardseed" in the first two cases gives the blogpost a ridiculing tone that grates with the use of the actual Steubenville rape case. Whether it was Landsburg intention or not, I think this makes it easy to read the post as though he is downplaying the harm to the victim in this case.

  8. What I don't understand is how this "thought experiment" could be made without any consideration about the *biological* factors being involved, at which point it becomes blatantly obvious what the difference is between 1 & 2 vs 3. In case 3, the rapist has given themselves a chance of impregnating the victim. Even in this case, if the victim does not become pregnant, if this was not prosecuted because of the "lack" of effect it had on the victim (which also isn't true, due to factors around sperm competition (i.e. if the victim was attempting to become pregnant with their husband, the rapist's sperm could block or kill or otherwise immobilize the husband's sperm so she doesn't become pregnant to the husband either), it leaves the rapist to attempt to impregnate the victim again in the future. If the only way the rapist could be prosecuted in this situation was if the victim became pregnant, then that *MASSIVELY* changes the cost/benefit analysis for the rapist (and that's even if she was on the pill, or was considered "infertile", etc... Pills fail, people who think they are infertile find out they aren't, etc). Society has a *huge* interest in maximizing the cost and minimising the benefit for rapists so it isn't used as a default method of fathering children in competition with methods that are much more socially constructive.

  9. I agree with all of that, but that isn't what the thought experiment is trying to assess. It's trying to see why we object to 3 even when we assume all your considerations away, when we don't object to 1/2.

  10. "chance of impregnating the victim."

    Landsburg doesn't mention it, but the "rape" in the Steubenville case was digital, not penile, so couldn't have impregnated the victim. One of the things I find odd about that case is the blurring of two different offenses, both wrong but one, both for consequential reasons and in terms of human attitudes, substantially worse than the other.

  11. My reaction was similar to yours—I was angry at Brad, not Steve. My basic conclusion was that the people who reacted as he did were either stupid or evil—stupid if they didn't see that Steve was raising an interesting legal and moral puzzle, evil if they saw it but took it as an opportunity to attack someone they disliked or disapproved of by pretending not to.

    My daughter's response was that I am more inclined than most people to see such questions in terms of ideas, to view a persuasive argument for an ugly conclusion as an intellectually interesting puzzle rather than a scary attack. I suspect the same is true of Steve. Whether it's true of Brad and he is just pretending I don't know, not knowing him.

  12. But the biological considerations indicate that the premise of the thought experiment is flawed. First of all, you cannot assume the considerations away, even if no pregnancy is caused or disease is transmitted, sexual contact and especially intercourse is not a null event (benefits and consequences eventuate even without pregnancy or disease). But more importantly, the main consideration of the society around rape is not about psychological distress (although everyone gives it lip service). Consider Scenario 3 with a female rapist and a male victim. Studies have been done on the victim of female on male rape that show that the male victims have the same feelings of psychological distress due to loss of control, etc. However, there is no way that the Steubenville rape case would have been prosecuted with the same amount of vigor (if it was prosecuted at all) if it was female on male. That's because rape by females on males simply does not have anywhere close to the same amount of threat to society. First of all, it is assumed that any woman who actually wanted to become pregnant (without infertility being a factor) could find someone willing to immediately make her pregnant (without offering financial support), so it's not seen as a woman gaining a child that she wouldn't have had, instead the only gain that she could really get would be financial (gaining financial support through using specific genetic material), but even then, it would be hard for a female rapist to then force financial support from the raped man through the courts (I guess I'll search to see if there are any known cases, but I figure by that point, it would be more in society's interest to prosecute her). Second of all, the woman could only "gain" once from this method every 10-12 months (given the time needed to recover from childbirth). In comparison, if even a relatively small proportion of male rapists impregnates a much larger number of women over that same 10-12 month period, it would seriously tilt "the playing field" against the men not committing rape at which point all stability evaporates from said society.

    People are offended by Langsburg's experiment because it reveals the unpleasant truth that psychological distress is not the driving factor behind the persecution of (male on female) rape. Society likes to pretend that it is "nicer" and more "considerate" than it really is,

  13. It comes back to society having an interest in maximizing the cost and absolutely minimizing the benefits of non-consensual behavior. If someone has gotten into the position of digitally penetrating the victim, then they wouldn't have had to "work" much harder to gain penile penetration of the victim and possible impregnation. Society doesn't want the rapist to even be able to get on the continuum toward possible impregnation, so they criminalize any possible steps or any "failed attempts" towards the truely objectionable action that they can. Even if the rapist had no interest in impregnation at that particular time, it still means that the rapist has gained experience that they could use to non-consensually impregnate someone in the future.

  14. DeLong's comment on it at Pileus I found a bit enlightening ...

  15. Landsburg's post was perhaps a little inflammatory and flippant given the subject matter, but it raised interesting philosophical questions which were the heart and purpose of the post. But DeLong's response on the other pileusblog post:

    "But Landsburg doesn’t think it’s a perverse utility function, does he?
    He thinks it’s a normal utility function–in fact, it’s the post’s
    author’s utility function, isn’t it? Is it your utility function too?"

    Am I missing something or did he just accuse Landsburg (and you??) of being the sort of person who would enjoy carrying out sexual assault? I really need to understand this: from my memories of micro, higher utility simply means "would be chosen". It's not a prescription that the person is necessarily made better off by higher utility, as people may make wrong choices. In that case, you can say that rape increases the rapists' utility by just noting the fact that they performed the terrible actions they did.

    But in my mind this conflicts with the idea of a perverse utility function. What exactly is a perverse utility function? Is it simply one that allows increases when people do terrible things, and such actions are not included in a non-perverse utility function? If this is the case, and DeLong is right in saying Landsburg has only a normal utility function in mind, then all this would mean is that Landsburg has an opinion on the theory of value: we have to accept that terrible actions that people (sadists, rapists etc.) carry out give them well-being. This may or may not be true. It's a difficult philosophical topic, and a real issue the utilitarian has to face in describing the correct theory of value. But it is no way justifies the leap to "In fact, it's the post's author's utility function, isn't it?"

    I hope I've misread DeLong somewhere, because otherwise...

  16. Is it even surprising these days that our moral intuitions don't match any normative moral philosophy? If they did then the entire field of moral philosophy would hardly need to exist because our intuitions would already be internally consistent. As Seamus often says, nobody is a true utilitarian.

    Frankly, I agree with @olejr:disqus that posts like Landsburg's are a cheap stunt to get press time. It doesn't add to our understanding of morality, nor does it change the fact that rape is universally considered to be wrong under all circumstances. He must have known what the reaction would be and yet he did it anyway.

  17. I don't know Landsburg, but who would seek that kind of publicity?

  18. 1. Agree we can use preference simply to mean "is chosen";
    2. I used "perverse" to mean a utility function where terrible acts enter on net positively. That's a judgment call though, and I would not say that it's the basis for discounting individuals' utility.
    3. I'm pretty sure that DeLong was saying that Landsburg and I likely would enjoy raping women because we think it's possible that others, whose preferences I find repugnant, enjoy doing so. I'm not sure that that's a constructive addition to the discussion.

  19. James. It is not surprising, and as you say, that is why the field of moral philosophy needs to exist. But that is what makes the question interesting. It is possibly the case that Landsburg likes being controversial, but I wouldn't dismiss the post as a cheap stunt that doesn't add to our understanding of morality. I certainly found the discussion on his blog (that subset that engaged with the thought experiment) useful, and I certainly found David Friedman's engagement with the question insightful.

    Furthermore, of course it doesn't change the fact that rape is universally considered to be wrong under all circumstances. That is precisely why he used that example (other than the fact that it occurred to him while listening to a news broadcast). Imagine, as per Eric's later post, he had used psychic distress caused by hovering over a mountain. It is unlikely that those with a libertarian streak would have found themselves dismissing psychic distress in the first two cases but not in the third and finding the need to ask what is the essential difference.

  20. I'm sure that the engagement with moral questions could happen without grossly offending so many people. Stunts can be fun to pull off and spectacular to witness but I don't tend to feel too sorry for the stuntman when he gets a few burns.