« Three Dialogues on Identity | Main | Decoherence »

April 21, 2008


Quite generally, your grabbing the person would be justified if this behaviour maximized ex post utility. Which is the necessary and sufficient condition for any paternalistic policy.

I'd love to read this, but for some reason my PDF viewer (evince, under Ubuntu Gutsy) renders the fonts in an unreadably unpleasant way. Is it possible that this PDF references a proprietary font that isn't included in the PDF and so won't render correctly on systems that don't have the font? If you could either ensure that the fonts needed are bundled into the PDF or use a font which all PDF readers support, that would help a lot. Cheers!

This made me realize that the usual arguments in favor of government control of medicine have to do with people not having the time or sometimes the capacity to research and think during medical emergencies, but actually medical care for non-emergencies is almost as tightly regulated.

Doctors know that most people won't think intelligently about medical matters.

What they don't recognize is that they don't think intelligently about medical matters, either. But there's no one with status and power to force them to face that.

One of the problems with this kind of story is that it attempts to compress a great deal of logic and analysis into a few words. Several questions arose for me in reading the sort of soliloquy of back-and-forth, questions and answers in the story.

First, in discussing whether you "should" stop the cliff walker, sometimes it seems like it is addressing the ethics of the action on a personal level. When should you stop someone? But some of the logic doesn't apply to considerations of whether you should grab his arm. "Well this wouldn’t justify a general policy of letting people who look like you force their advice on people like them", it is suggested. So I think that is the real issue here, what should the general policy be of society in terms of allowing people to force advice on others? Coaching things in terms of the language of personal ethics is a bit misleading, I think.

In terms of justifications for action, I would want to consider more closely the issue of majority rule. "...you should ask yourself: what is your basis for this conclusion? Is it because there are more of you, and pretty much all decisions should be made by majority rule?" Okay, suppose that is the reason. What is the counter-argument? Is it: "If so, why do you also regulate people on this superior side?" Well then I would say that we don't, that the majority-rule rule never regulates people on the "superior" side of an issue, because by definition the superior side is the one which is more numerous, and the rule which is getting enforced is the one shared by the most numerous side, and people are not regulated when a rule which they already believe in is promulgated.

And then there is the listener argument, which might apply to this case. Why do we believe that the majority are better listeners? Because if the minority were as good listeners, they would be convinced by the majority. "Well in this case you are arrogant, i.e, biased to presume your own superiority, if you do not at least consider what basis you might have for such a lopsided conclusion. Is it, again, your superior age, gender, class, or ethnicity, or the fact that you are in the majority?" It would be the latter. Again I do not see a reason why this doesn't work.

Now I can understand why it is not practical to develop a response to every single possible objection, with all possible responses to the responses, etc. The parable would become an encyclopedia. But at the same time, this points to a problem with this format of argument. Its admirable brevity makes it all too easy for the reader to raise objections that seemingly negate its conclusions.

Lastly I found the final section with the results on warnings vs bans to be quite confusing. "Assume that for most risk levels, the activity level the advisor prefers is close to, but not quite the same as, the doer’s. For example, if an all-knowing cliff walker’s ideal cliff distance were ten feet, an all-knowing advisor might prefer eleven feet instead. So if the advisor would be exactly believed by an ignorant doer, that advisor would be tempted to give slightly distorted advice." I don't understand the last sentence. If the advisor would be exactly believed, why not give accurate advice? He should tell the walker to stay at eleven feet, and then presumably the walker would do so, since he would exactly believe the advice. The rest of this section then develops from this seeming need to distort advice, with distortions of distortions, making the whole thing incomprehensible since this foundation point was left obscure.

Hal, if the majority votes a rule to keep people from the cliff, that rule applies to members of that majority. So they are seen as superior rule makers, yet still inferior regarding their own cliff-walking choices. Regarding warn or ban, the 11ft advice is distorted relative to the doer's preferences; once the doer realizes this he will no longer want to believe everything the advisor says.

I was wondering if you could tell me the name of the painting used as the header of your blog with the man tied on the boat. Its lovely and I would like to know who painted it.



Paul Crowley, the fonts display fine for me using Dapper and Evince 0.5.2. Hmmm examining the document properties the fonts are Microsoft TrueType fonts: ArialMT, Arial-BoldMt, TimesNewRomanPSMT, and TimesNewRomanPS-ItalicMT.

These can installed in Ubuntu, you might violate a few laws by doing so. I'm no lawyer and I don't know or care. The package is msttcorefonts. Anyone that cares about freedom would not use such fonts but in all likelihood Robin Hanson gave it no thought. I would hope he does care about freedom and from his post here I would think that he does.

Every action we take impacts everyone else, one way or another. That is not the same as establishing benefit or harm.

How does it harm anyone if the cliffwalker should fall and die? It really doesn't. Other people's feelings about his state are irrelevant - having the world not match your desires is not a harm. Neither is having the world match your desires a benefit. If people wanted him to fall, should we shove him off the cliff? No, because their feelings do not enter into the harm-benefit calculation.

The world would be a lot better off if people would just mind their own business and only that, instead of trying to declare everyone else's business their own too.

Rachel, I believe it's Ulysses and the Sirens.


How does it harm anyone if the cliffwalker should fall and die? It really doesn't.
I've heard arguments that negative externalities from events like this (to the walker's family, or outraged onlookers) constitute harm and therefore justify intervention. I think the problem with this is that without some ethical viewpoint, we can't distinguish between the negative externalities which justify action and negative externalities which don't. Externalities are completely arbitrary and subjective value judgments, after all. For this reason, it isn't clear to me why we'd want policies which maximize utility, when utility itself (if it were even measurable) is a mutable thing.


so let's say your cognitive abilities have been impaired for a few hours. Maybe you hit your head on the doorframe. Maybe the cook put something funny in the salad you ate.

Are you trying to tell me that you don't want me to grab you by the arm? Rather mind my own business? Really?

As in really, really?

If you came to my door and rang the doorbell because you wanted a chat, and no one answered, and you broke down the door and found me lying in my kitchen having a stroke, and your forcing your way into my home was what made it possible for me to receive lifesaving treatment... does that in retrospect validate your action?

If you kick a blind beggar in the head, and his sight returns, should he hail you as a benefactor?

In the cliff example, what counts as an allowable intervention depends a great deal on what knowledge you have and what actions are available to you. In most of the hypothesized cases, you have no grounds to interrupt the walker at all.

What if the cliff walker is kin? Do I have a greater right to intervene if the person is my child or a sibling? If so, how much less do I have to know about the state of the individual before action is permissible or obligatory?

The expected harm of grabbing the walker when he's aware of what he's doing is small, provided you let go when he asks you to. The expected harm of not grabbing him when you should have is pretty big. That makes a difference. It's not exactly burglary or battery.

Robin Hanson is asking interesting questions about our rights to be protected from the utility calculations of the majority (or even from the selfishness of the majority). I especially like that the analysis is posed as a check on our moral intuitions (as John Leslie's Doomsday argument is intended as a check on our feelings of permanence).

Nagel (Private Rights and Public Space) posits that certain rights should be guaranteed without any inquiry into utility. The right not to be tortured, for instance, should be absolute: even if torturing *this one* person would result in saving *all these other* people from torture, we don't have the right to do it.

I'm not sure if "the right to walk off a cliff" is one of those - it rather appears not. But, as others point out, even if utility is our basis for action, there are problems estimating the happiness my intervention in another person's life will bring him or others. If we follow LemmusLemmus' line of thinking, Christian sects that believe all non-believers are going to Hell have an ethical duty to convert the rest of us - by force, deception, or other harmful tactics, if necessary. It's just a utility calculation, after all, and they're sure they're right.

Sister Y - it's important to distinguish subjective, objective, and evidence-based versions of utilitarianism. Only the subjective version says we have a duty to do whatever we believe (however unreasonably) would be best.

Part of why I am confused here is that the story doesn't seem to clearly distinguish between factual information and preferences. In decision theory I have a probability distribution over possible facts about the world (or we might say, possible worlds), and a utility function which ranks these possible worlds. Then I choose the action whose probability-weighted utility of the possible outcomes is best. But in the story, it's not clear whether the advice and/or grabbing/banning is due to disagreements about factual information, or to dislike of the cliff-walker's utility function. Do we regulate because we think people are misjudging factual situations? Or because we accept that they may have the facts straight, but we don't agree with their tastes and preferences in terms of what they like and dislike? These seem like quite distinct kinds of reasons, but in this story they tend to intertwine.

Hal, part of my point was that interesting things happen when we both have different info and different preferences. The effects of small amounts of differing preferences can be magnified by differing info.

Jeffrey Friedman has some interesting critiques of majority rule you can find from here. Where he differs from Bryan Caplan is that he believes the better-informed elites are dogmatic ideologues and we can not confidently say it would be best to leave policy up to them.

I don't get it. Your paper doesn't seem to actually say much of anything. It just asks many questions and proposes very few answers. By the time you reach your "conclusion", I don't see substantial points that you've presented to support it -- you just seem to pose a lot of questions and then decide that you're right. If you're so sure your position is correct, why don't you establish the argument directly?

The very fact that you're able to grab him justifies your doing so, because a sober and harm-free suicide attempt would have no surprise witnesses.

Personally, I'd want to make sure this kind of suicidal thought process doesn't become a precedent. So I'd grab the guy and beat him to death before throwing him over the cliff. Wouldn't want him to survive, have a change of heart, and start pumping those silly ideas into the gene pool.

Instinct isn't paternalism, so I'm not fond of the analogy. But in all honesty, I'd grab the guy and try to talk him round. I'm not issuing a decree outlawing suicide here, I'm having a natural human reaction and trying to save a life. I have prior experience of having had irrational intentions, and been grateful to someone for talking some sense into me. With that in mind, the pros of asking the guy if he'd like to chat about what's on his mind outweigh the cons - both at first glance and on reflection.

Ping, my conclusion is mainly that his is a a subtle topic, and deserves careful consideration.

Thomas and Ben, by his words and deeds he isn't obviously trying to commit suicide - he is just taking more of a risk than perhaps you think appropriate.

But what if, aside from the whole cliff thing, he seems no crazier or immoral than most? What if his action mainly affected only him?

Apologies Robin, I misinterpreted this bit as a 'rational-looking suicide attempt'.

Thomas and Ben, by his words and deeds he isn't obviously trying to commit suicide - he is just taking more of a risk than perhaps you think appropriate.

Appropriate? How is it appropriate to apply one's own preferences to random people and situations about which little is known?

There have been societies in which it was asserted that the State owned its citizens and possessed a right to their labor, such that attempting to kill yourself was considered a crime , a treason against the State. The most common penalty for that crime was death.

That says a great deal about the sorts of thinking you're implicitly advocating.

Richard, yes, but I think part of Robin's point is that it's easy for an agent to exaggerate his objectivity and access to better evidence, perhaps without even meaning to do so.

I wonder what is the summum bonum of society as a whole (if, unlike former PM Thatcher, you allow yourself to postulate such collective entities), and who is to bear the cost for the reckless choices of individuals aside from the individual in question? Doesn't moral hazard enter into the analysis? Or is the characterization of another's behavior as "reckless" an arrogant, insupportable assertion on its own?

Imo a more extensive study might well delve into not only public arguments but also ulterior motives of "advisors" and study cases of bad "advising", such as (I presume) the recent debacle with the polugamous sect in Texas. Although sometimes paternalism is well motivated in principle (say, anti-trust legislation intended to prevent domination by a monopoly, and regulation of medical practice is intended to protect the vulnerable patients) yet the devil lies in the details.

The comments to this entry are closed.

Less Wrong (sister site)

May 2009

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30