« Formative Youth | Main | Markets are Anti-Inductive »

February 25, 2009



Let's try an unpack it; there seem to be several forces at work. First, the religious exception; that's there because it's completely impossible, in practice, to try and ban certain religions or seriously control the religious (as opposed to the political) piece of it.

Putting that to the side, the difference seems rather clear: deception is punished, self deception is not; statement of narrow impact are punished ("I saw her stab the guy!"), statements of diffuse impact are not ("Drug prohibition is a unqualified social good"); anything where "reasonable people can disagree" (social measure) is not punished. Different standards are demanded depending in what capacity the speaker is speaking - a journalist reporting, a witness testifying, a scientist publishing, are held to higher standards than the same person in the pub (and, as a consequence, are believed more). Finally, the harm caused (to single individuals, not to groups) is somewhat factored in (the president of News Corporation spreading a vicious lie does not have the same impact as Sam the cleaner doing so).

The blackmail case arises from a different direction; either the blackmail is about a crime, or it is about something not criminal but socially unacceptable. If it's a crime, the state wants to know about it, and apply the usual binary solution of a trial: be proved guilty, or take the issue no further.

The other case is more interesting; allowing blackmail would be adding an extra method of enforcing social norms (the profits to be made by blackmail would draw more people to investigate). I could see some arguments being made for allowing this (especially by some who feel that hypocrisy is particularly bad) but there are many good reasons that laws are enforced while social norms are not. The stereotypical behaviour of blackmail victims (warning: partially fictional evidence) also seems to imply that blackmail payers are mistaken in their preference for paying rather than letting the secret out (immediate impact bias).

Do you have a personal alternative that you would prefer to have implemented, Robin?

There are more than just the truth claims at stake. I think it should be pointed out that blackmail doesn’t need to be true in order to be effective. Libel (and to a lesser extent slander) protects small businesses, individuals and organizations from unfair competition or harassment by larger businesses- this can be seen as a form of protection for the free market at the product level. Without some form of libel it would be almost impossible to launch a product that could challenge the interests of a media giant. Libel and slander can also prove to be dangerous. Religious claims are an exception because they are unfalsifiable in nature and, at least in the US, our separation of church and state makes dealing with even clearly dangerous charlatans a wooly issue. Medical claims require some form of expertise- we can’t expect everyone to have a doctorate- and could undermine the public trust in hospitals. I guess in a nutshell, there are externalities in play and these things impact more than the speaker and the listener. If blackmail was legal, being upwardly mobil or a public figure would be a mine field. If libel was legal being an entrepreneur could... well you get the idea.

Truth, in isolation, is all but irrelevant to this discussion. Selective truth-telling is just has harmful as outright lies, if not more so.

At least in the USA, I don't think it's an "odd mix" of the two--it's pretty clear that America both implicitly and explicity prefers "Listener Beware", but has accepted an ad-hoc collection of exceptions for cases where that doesn't seem to work, basically along the lines Stuart Armstrong gives.

Also, note that libel and slander aren't necessarily falsehoods, just intended to damage a reputation.

It's about simplifying people's lives. Similar to the way that massive amounts of choice can paralyse us, having to deal with every single detail of every transaction we make can similarly paralyse us.

So, for instance, if I had to check the label on all of my food to see if it said "May Contain Poison" then I wouldn't be able to do my shopping in five minutes. Outsourcing the checking to a central agency saves me time - and because it's something _everyone_ has to do, this is worthwhile.

So it's a balance between centralising something (and the inefficiencies that come with that) and personal respnsibilities (and the inefficiences that come with that)- and the balance usually depends on how often something needs to be checked, and the education level of the average person.

When two companies are signing contracts it's a rare occurrence with a high value, so they can be assumed to check things in great detail themselves, and check the reputation of the other party.

When I buy over-the-counter medicine, it's a low-value, high-frequency operation and I don't have the ability to check the claims of the supplier.

Clearly the first should be left up to the "inividuals", while the latter should be done by a central clearing house.

I guess that factually false statements which are directly and immediately applicable to someone and can harm them are the ones which are not allowed - namely libel, slander and blackmail. These statements can cause harm directly and therefore the cost of these being believed, when said, is more immediate. You can see the costs of their effects easily and therefore they come first in the order of things needing mediation, and therefore law. Most of the examples opposite to this view are more long term and the harm caused is fuzzier. This then can be left to people to decipher and believe/disbelieve in their own time.

Laws are after all evolved and have been modified incessantly - closing loopholes and trying to stay ahead of events. This being the case it makes sense that the statements that cause direct harm and most easily fact-checkable are outlawed while the rest are left alone.

"Bryan and I find it easier to understand two extreme positions than the actual intermediate mixture we have."

Why do you find easier to understand extreme positions than a moderate one? Is it a general rule for you or does it apply only in this case? (I almost universally prefer a compromise over extremes and, possibly because of my fear of fanaticism, I have strong dislike for desires to apply some particular rule without exceptions. This is why I find your statement surprising.)

Current law hews pretty closely (although not exactly) to a principle that it is permissible to say something false, so long as you believe it. The legal line, therefore, tends to lie between errors and lies. (There is first a requirement that an assertion be sufficiently verifiable as to be worth regulating; this is the fact/opinion dichotomy of American First Amendment law.)

I defend the error/lie distinction on social epistemic grounds here. The capsule summary is that lies involve epistemic harms that errors do not: (1) An individual's belief in a proposition is important evidence of that proposition's truth, but it can be very hard to verify belief states in the presence of deception, (2) lies typically involve the strategic creation of false evidence designed to resist falsification. Accordingly I think that most harms flowing from honest errors can be mitigated by mandating disclosures of underlying data (and by promoting social institutions that promote the views of those with relavitvely error-free track records), while lies tend to require stronger regulatory medicine.

Blackmail is a separate issue; truth v. falsity isn't the key issue there. Rather, the law seems to be driven by a combination of privacy concerns, when the threatened disclosure involves lawful but embarassing activities (i.e. we don't want people threatened by the disclosure of facts that they would prefer to keep secret) and misprision concerns, when the disclosure would relate to unlawful activity (i.e. we don't want to give people incentives to conceal the facts of a crime once they become aware of it).

Barry: I agree, but selective truth-telling probably represents deception in its own right; most speech-settings involve a norm of disclosing enough information to make your contributions non-misleading. (The most prominent counter-example being the trial process, which sometimes encourages lawyers to selectively reveal information so as to create false impressions.)

Souless Automaton: Modern American law, at least, tends to limit defamation lawsuits to cases involving either knowing or negligent falsehood.

I think Mark has a good point, though I would probably approach this issue from an econ and law perspective: the lowest-cost investigators of the truth of a claim are the people against whom it is directed.

Thus, the authorities can stay back from judging the truth of claims until they are discovered by the people bringing suits; this simultaneously minimizes investigative/transaction costs and prevents socially-disruptive lies from freely propagating.

We should also remember where the laws come from. I would guess that these civil laws are designed to provide a healthy environment for commerce.

If a religion wants to make outlandish claims, well, the only suckers who will far for it are its adherents.

Medical claims are much murkier; so much commerce depends on an implicit government certification...before we had FDA approval, you had to decide for yourself which snake oil was actually effective. Without the ability to punish false medical claims, the entire pharmaceutical industry would fall apart.

Stuart, our society is an outlier re religion; most regulate it. Blackmail typically increases total punishment, even of criminal activity.

Like prase, I am not surprised by a moderate outcome. Perhaps you could convince me that the outcome is an "odd mix" by listing potential examples comparable to the existing exceptions.

Transaction costs seems to me like a pretty good explanation of the exceptions: in slander/libel and a trial there are interested parties to pursue the issue. The main question I see is why more cartels have not followed the medical precedent.

Good systems have a built in method to change. If you make a law to disallow political criticism society can't anymore fight that law. You get feedback loops that in turn change the laws to increase the power of those people in the government.
For that reason political speech deserves additional protection.

It's also the human right to determine your own identity. People define themselves through religion and their political beliefs.
If you tell a lie about another person you infringe that persons ability to define their identity because he has to defend himself against that lie.

Our law is centered on the idea of human rights.
Truth has no rights. Ideas also have no rights.

The idea of the independent human being that has the right to life and liberty entails the right to define your identity.

Authorities are also not allowed to torture people to bring truth to the table because that would infringe someones self integrity (at least they used to be).

To simplify, in the US we have "listener beware" for political and religious claims and "punish falsehood" for other claims.

This could be explained by the fact that most disagreements in politics and religion are dishonest (as per Robin's paper) and therefore the government has a particular inability to determine what actually is the truth.

People are less biased when investigating fraudulent products, so it is conceivable that a government punishing falsehood outside the realm of politics and religion could do more good than harm.

Don't take this as rude, but there seems to be a lot of rationalisation of the status quo going on in the comments here. My "analysis" included, but at least my opinions have changed as a result of Robin bringing this up - I understand the banning of blackmail better, and (with the exception of witnesses at trials and scientists) would now favour less libel laws than I did before.

Have others had their opinions changed (even slightly) by this blog post?

In the US, alcohol companies may not buy TV ads truthfully saying most studies find people who drink more are healthier, and trial witnesses may not truthfully tell rumors they've heard about the accused.
You disappoint me a bit here, Robin. The reasons for this are obvious, and unrelated to your main points - the first is because alcohol companies are allowed to advertise only on sufference anyway, and the second is because repeating hearsay in court is very pernicious to determining a correct sentence (especially if this is instiutionalised).

Blackmail typically increases total punishment, even of criminal activity.
Of course; so blackmailing about crimes is a mixture of vigilanticism with concealment of the crime. All the more reason for the state to hate it.

Most democracies do not have criminal defamation laws; rather these issues are raised in civil courts. In those cases they must be able to demonstrate that there was a damage created by a malicious intent - think negative-advertising where I just tell lies about your product loudly and frequently enough that many people believe it. Similarly as an individual I might attempt to discredit someone's reputation at work to get them fired so I can get their job.

I don't understand why people think words cannot create real harms or culpability.

Tom P. said: To simplify, in the US we have "listener beware" for political and religious claims and "punish falsehood" for other claims.

I appreciate to simplify, but I think this summary is misleading. We have, in fact, four primary regimes:

1. Listener Beware: Various non-verifiable statements, including religious and normative assertions. This applies to some political discourse, but not all; many political claims are in essence factual and can be the subject of defamation liability.

2. Punish Insincerity: This applies to factual assertions about political or other public figures, and (in essence, if not as a matter of formal doctrine) to most other speech on matters of public concern. (I would group the various "negligent misstatement" torts in this category for simplicity's sake, only because it will be rare for people to sincerely believe what they have no reason to believe, at least with respect to the verifiable matters that fall outside of the "listener beware" rule.) So understood, this rule also applies to perjury prosecutions, to almost all fraud claims and to many malpractice lawsuits premised on incorrect expert advice.

3. Punish Falsehoods (even if sincere): We apply this rule only to "commercial speech," a hard-to-define category that most often involves advertising or attempts to negotiate commercial transactions. Even here, there aren't many cases in which this rule is applied to its fullest extent (that is, to cases involving honest errors by a commercial speaker).

4. Mandate Disclosure of Underlying Data: we use this rule in the trial process and in securities law, for example. We often use it in conjunction with another rule; for instance, we apply rules 2 and 4 during trials (lies are punished as is the withholding of certain underlying information), while not punishing errors that involve neither a failure to disclose or an attempt to deceive.

So Robin is right to say we mostly have a blended regime. Our pluralism about normative and religious matters---combined with a desire not to stifle public debate on these issues---probably explains why we have category one. Mostly, we use category 2, which makes good sense if you think that (a) punishing honest errors will deter too much truthful speech, both directly (because even truthful speakers will decline to express themselves for fear of liability) and indirectly (because deterring false speech will also deter the truthful counterspeech that might better educate an audience), while also thinking that (b) deliberate lies cause more harm the honest errors (because they are harder to cure with effective counterspeech) but banning them will deter less truthful speech than would the "punish falsehoods" rule. In other words, our middle ground is designed to strike a balance between promoting as much truthful speech as possible, on the one hand, and deterring as much false speech as possible, on the other.

As to category 3: our use of this for commercial speech mostly stems from an empirically dubious assumption that commercial speakers are less likely to be deterred by liability than other speakers. I think this assumption is wrong (indeed, commercial speakers may be more likely to rationally weigh the costs and benefits of liability before speaking), so I generally advocate getting rid of the "punish falsehoods" rule except in a few narrow circumstances where the harms of false speech are very high and rebuttal speech is unlikely to occur. And even in those circumstances, a combination of 2 and 4 should normally work better than 3.

"why are libel, slander, and blackmail illegal"

Why is this a puzzle? From either a monkey or a political theory perspective?

It's obvious from a monkey-tribe outlook - libel/slander speech acts (which by definition are untrue) attack & disrupt the social hierarchy and status machinery, a prized property of monkey society. If you lie about some element of my status, you've harmed one of my most important pieces of monkey capital and injured society overall by making it more difficult to rank status "correctly."

As for Rothbard, it is likewise obvious, yes? Rothbard's axiom is about violence & property. He often states it in a form similar to "the fundamental axiom of libertarian theory is that no one may threaten or commit violence ("aggress") against another man's person or property."

Until the invention of the consumer credit rating, which really was artifact of the 1960s as credit cards became widespread, one of a man's most important possessions was his public commercial reputation. Libel & slander laws - defamation laws - date back to Roman times and were created precisely to protect a man's property in his good public reputation.

Until the industrial revolution, most Western men had little or no tangible property - no land, no houses - perhaps only small livestock, some small furniture, and his chattel wife and children. In this environment, his intangible public reputation was actually his most valuable piece of property.

Thus it seems clear that defamation laws were an important piece of property protection, and remain so even to this day. Truth has long been considered a defense against defamation claims in the US.

Understanding this as background, Listener Beware seems appropriate when the economic burden would fall on the Listener, or when there is an ability to access accepted objective information, such as a credit score. Punish Falsehood seems appropriate when overall social harm is possible, the economic burden would fall on the person defamed, or on innocent third parties and there is no way to objectively check information.

there seems to be a lot of rationalisation of the status quo going on in the comments here

That is what was initially requested. Recall what Robin initially wrote:

Surely there are some implicit bias theories behind these rules, but until these theories are made more explicit it remains hard to evaluate how much sense these rules make.

Robin was asking for people to make explicit the implicit bias theories behind the status quo rules. By definition, those theories necessarily rationalize the status quo.

Blackmail seems to me rather simple to explain in terms of the incentives of powerful people. There's not much use in blackmailing Joe Average; even if you can prove he cheated on his wife, he doesn't have a lot of money to give you to avoid the consequences. So blackmail is an activity that, by and large, transfers money from rich (and/or powerful) people to poorer ones.

That said, it also seems to me that blackmail is a rather unusual sort of transaction, because it is very difficult to enforce. Suppose we agree that you will pay me $100 a year, and I will remain quiet about that time you got drunk at my pool party. How can this contract be enforced? As soon as the facts are laid before a judge, the secret is out! Even if the details are kept confidential, as Eliezer pointed out the other day, "the most important part of a secret is that a secret exists". Just knowing that you have a secret you are willing to pay to protect will cause most of the difficulties you are trying to avoid by protecting it, because people will insist on speculating. (And the speculations will likely be worse than your actual behaviour at the pool party!)

Now, if I could credibly and verifiably commit to actually forgetting about the pool party, in exchange for a one-time payment - editing my memories - that would be different.

Mark Spottswood: Yes, I deliberately did not specify American law in that statement. It's my understanding (with only moderate confidence here), that US law is the outlier in this regard, and that in most European countries slandering someone with deliberate and malicious intent is potentially illegal, even if no falsehoods are stated (though it may strengthen the defendant's case). In contrast, in the USA factually true statements, as well as statements of personal opinion, are essentially immune. Or, at least, that's my understanding.

Pointed direct harm to an individual is most relevant. One can certainly make all sorts of claims without punishment. 'Vaccines cause autism' has been one. Only identifying which vaccine would be actionable. Harm against an individual relying on it is less, but can come into play especially when there are commercial considerations. 'Laetril cures cancer' is alright for personal expression, but would not be tolerated commercially. There aren't, or shouldn't be, markets in libel, slander, and blackmail, but they have market implications. Punish falsehood applies very much to markets to reduce fraud and reinforce honesty. Opinions take place beyond markets and listener beware would typically apply. Opinions stated in court would take on aspects of libel and slander even if unintentional. Of course people have been sued for opinions even when they are presumably free to express them.

The added examples are very good.

Still, they are in two areas that have already been identified as special (law and medicine). In particular, if we view the rules of medical speech as regulatory capture by a cartel, we should not expect those rules to be well-explained by avoiding harm to the general public.

But they aren't special at all, but common to all market transactions. Misrepresentation is fraud. They apply equally to financial claims such as Madoff to selling tilapia as red snapper. If legal process is considered the search for truth, then falsehood cannot be tolerated, and as market disputes are settled legally, falsehood couldn't be tolerated there either or the we would have the defense 'the sucker believed me'. Science as the search for truth won't tolerate it either although the process is not a legal one, though the law will draw on science as a resource in adjudicating its disputes. Libel and slander are frauds on the market.

Deceptions or attempt at deception is seen as immoral or unethical. People punish virtually all types of deception, even if it is just a lost of trust. Deception is seen as a public moral issue yet it is extremely common, and the consequences can be from negligible to very high. For authorities to regulate even a fraction of it would be impractical. So we usually see high consequences and easily identified deception regulated by law, like medical claims and truth in advertising. A more difficult identified from of deception in courts is also regulated because of the high consequences, courts usually have the power to imprison or expropriate assets according to testimony.

That is what was initially requested. Recall what Robin initially wrote:

Surely there are some implicit bias theories behind these rules, but until these theories are made more explicit it remains hard to evaluate how much sense these rules make.

Ah yes. Me bad.

There is a LOT of bad legal analysis going on here. Most of it is simply wrong. An example in the post itself is that you CAN be paid not to tell the truth, e.g. you can contract away your right to disclose that you were paid by X Corporation for your injuries. You can sign a non-disclosure agreement as to almost anything. Also, trial character witnesses may, in fact, testify to someone's general reputation in the community; so, even if it's true that a witness can't say, "Simon told me the Defendant kicked a puppy," he can say, "Defendant has a reputation for violence against animals."

I think you and Brian need a Con law expert, who will provide the real constitutional/First Amendment limits, both criminal and civil, for libel, slander, and blackmail. I think a great debate topic would be on "group libel." The Supreme Court has strongly indicated that a group cannot sue for libel, i.e. Jews cannot sue if someone publishes an article on how the blood libel is becoming popular again. See, e.g., http://supreme.justia.com/constitution/amendment-01/44-group-libel-and-hate-speech.html. Is that a good rule?

I strongly agree with UlrichRoarke: this post (including the comments) has lots and lots of bad law.

It's a good thing it's not illegal to say untrue things about the law in a blog post, or a lot of people participating in this thread would be in trouble.

For instance, a commenter says: "in the USA factually true statements, as well as statements of personal opinion, are essentially immune." No. Truth is a defense to libel and slander, but there are other torts you can be liable for even based on making true statements about someone. If I procure your medical records and publish them in a newsletter, I may well have probably broken the law; you'd need to research the applicable state tort law.

More fundamentally: the post itself is assuming the law must be based on "bias." ("Surely there are some implicit bias theories behind these rules...") You're saying that because that's the theme of this blog. But there are innumerable pages worth of judicial reasoning in American case law explaining why we draw the lines where we do. Now, that reasoning may be open to criticism -- I'm not saying we have to go along with it. But why not give one single example of a Supreme Court case (or a law review article, etc.) on a specific 1st Amendment issue? There's been so much written about why you give special protection to political speech, why it's OK to heavily regulate commercial speech, etc.

When the blogger says, "until these theories are made more explicit it remains hard to evaluate how much sense these rules make," I wonder: "more explicit" THAN WHAT? Do you realize that when a 1st Amendment issue gets decided, it's frequently accompanied by a fairly nuanced written opinion going through the relevant concerns, often at great length? And even if you think judicial opinions are too focused on specific legal issues (though they can include lots of tangential material), what about scholarly articles, treatises, and law school casebooks? Have you seriously examined these before concluding that the theories underlying the regulation of speech needs to be "more explicit"?

Again, I'm not endorsing everything those theorists have written -- they disagree amongst each other sometimes, so they're not always right. But a more illuminating post would pick out something they've written and explain what it gets wrong, not simply assume that no one has given an adequate explanation.

Institutions almost always punish falsehood although the manner differs. Religion punishes through doctrinal councils and excommunication, politics through parties and elections, science through peer review and rejection, the law through courts and penalties, markets through law and judgments, though there exist unsettled areas within them where listener beware operates until settled. Only inter institutional and interpersonal relations open themselves to arbitration in the absence of which listener beware rules.

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