Monday 5 December 2011

Democracy is not a truth machine

In a democracy people are free to express their opinions and question those of others. This is an important personal freedom, and also essential to the very idea of government by discussion. But it has also been held to be instrumentally important because in open public debate true ideas will conquer false ones by their merit, and the people will see the truth for themselves. In other words, democracy has an epistemic function as a kind of truth machine. From this it follows that in a democracy there should be no dogma: no knowledge protected from public challenge and debate. Yet this whole argument is founded on embarrassing misconceptions of  the nature of truth and of the working of democracy.

The case for seeing freedom of expression as a public as well as a private good was made most eloquently and famously by J. S. Mill in On Liberty.

[T]he peculiar evil of silencing the expression of an opinion is, that it is robbing the human race; posterity as well as the existing generation; those who dissent from the opinion, still more than those who hold it. If the opinion is right, they are deprived of the opportunity of exchanging error for truth: if wrong, they lose, what is almost as great a benefit, the clearer perception and livelier impression of truth, produced by its collision with error.

Mill's argument is a rhetorical tour de force in defence of pluralism and individual freedom. But it is only partly right. In particular Mill acknowledges no distinction between moral, religious & political opinions, and opinions about facts & rational truths. He also fails to distinguish the processes of discovery and evaluation.

Mill is quite right to defend people's freedom to form, express, and debate their own opinions about religion, politics, and morality. These are subjects on which anyone can have an opinion; on which in a free society everyone has the right to have an opinion; and on which the very legitimacy of opinions requires their formation in a particular way (through non-coercive persuasion and critical self-reflection). Liberalism itself is founded on a respect for individual autonomy on these issues, in the sense that every person is considered to develop their own 'conception of the good': a kind of personal moral idiolect governing our judgements about value (opinions) that is built up from our cumulative interpretations and re-interpretations of our own experiences.

Mill's particular contribution here is the counter-intuitive point that we can each learn from engaging with unconventional opinions even if we still conclude that they are bad and wrong, because they may be interestingly wrong. Books like Plato's Republic, Machiavelli's Prince, Hobbes' Leviathan, or Mandeville's Fable of the Bees offend nearly everyone and persuade almost none, but they act like grit in an oyster by forcing us to rethink the justifications for our conventional beliefs. Allowing such dangerous ideas to be aired thus supports rather than undermines individual intellectual autonomy, which is why many undergraduates are required to read them.

Mill's concept of an opinion also fits neatly with the public reasoning he recommends for democracy. Though held by private individuals, opinions have an inter-subjective aspect that allows them to be justified, debated, and critiqued in public. We can discuss and try to change each others' minds about what we should value and what should be done on issues as diverse as Who should win American Idol? or Should we invade Syria? In democratic politics, after talking like that for a while, we participate in a social choice exercise that transforms those individual opinions into a collective decision - we vote to decide which opinions are most agreeable to most of us.

But ethical precepts like 'torture is a crime against humanity', religious beliefs like 'Jesus loves you', literary judgements like 'Ulysses is the best book ever written', or explicitly political views like 'a republic is better than a monarchy' have the peculiar character of being fundamentally a matter of opinion. They really are up to us to decide. They are therefore quite different from objective truths, whether rational truths (as produced by rational enquiry, such as science) or facts (such as historical events).

As Hannah Arendt noted in Truth and Politics (New Yorker 25/2/1967), objective truths have quite a different epistemic character than opinions, and that makes their evaluation quite different. Unlike opinions, objective truths are depersonalised 'views from nowhere'. Not my truth, your truth, or even 'our truth', but The Truth. They are not amenable to democratic debate or discussion since whether or not people find them agreeable is quite irrelevant.

To put it in a nutshell, it is right that in a democracy the people can debate and vote to decide whether murderers should be executed, but it is misguided to think the democratic process can also decide the factual issue of whether a particular person (OJ Simpson say) is a murderer. Democracy can determine whether to teach Intelligent Design in schools, but not whether it is true.

The democratic market for ideas

If one tries to consider what a free  - democratic - marketplace for ideas in keeping with Mill's beliefs would look like one would be hard-pressed to find a better candidate than the world wide web.

Traditional media for the dissemination of ideas - like books, newspapers, or even university seminar-rooms - have space constraints and so their content must be curated by professional editors before being submitted to the public. That curation for quality and likely popularity naturally reflects the orthodoxy of the mainstream ('the tyranny of the majority' that Mill worried about) and filters out a lot of 'bad ideas' not fit for print.

On the internet, in contrast, the space constraint disappears (though we still have an attention scarcity constraint!) and hence professional curation recedes in importance. 'The people' are now free to express their opinions directly to the world (if, in some countries, only anonymously). The people can also assess each others' ideas 'democratically'. We can search and browse for what people have written on the subject we're interested in, and bring our favourite ideas to prominence by sharing and linking to them. In other words, we vote up the opinions we find most agreeable, but we do not exclude the opinions we dislike from being found and considered by others.

The web eminently satisfies the requirement for discovery - generating and disseminating heterodox claims about objective truth. It also performs an assessment of those ideas in a democratic way, i.e. weighting them according to their popular appeal. Nevertheless the web offers an object lesson of the flaws in democracy's claim to be a truth machine. Democracy, in and of itself, provides no mechanism for the evaluation of objective truth.

Rational truths

Rational truths are those established by chains of human reasoning that can in principle be replicated by others. Like Euclid's geometry. Science is an archetypal form of rational truth seeking since its authority depends on such replication: that one will always get the same result in the same experiment because the result doesn't depend on who the scientists are, but on the independently existing world. That means for example that if all the climate scientists in the world were wiped out by a freak meteor at a conference, climate science would quickly reappear and say basically the same things again (as more or less happened when the Catholic Church tried to suppress heliocentricism).

Note, however, that rational truth does not work by persuasion, as opinions do in a democracy. Instead it rudely asserts that this is how things are whether you like it or not. The standing invitation to replicate the results of an experiment is not an invitation to have your own opinion about whether they are true.

Now consider how rational truths appear in the democratic 'free market for ideas' on the internet. One can find all sorts of heterodox claims that purport to have a scientific basis. So we can find claims that vaccinations cause autism; organic food is healthier; abortions cause cancer; the world's climate isn't actually changing (or if it is, it isn't our doing; or if it is, it won't be so bad); etc. These claims are often supported by some plausible sciencey sounding arguments and mechanisms that seem a bit like what you learned in high-school.

So how is Mill's truth machine supposed to work here? Supposedly the presence of such heterodox claims benefits us because it gives us the chance to 'exchange error for truth' (or at least improve the partial truth we hold). But what actually results when people with different opinions on the science of climate change debate it in public is an increase in confusion, not enlightenment. The problem stems from an underlying confusion between the generally adequate liberal presumption of equal intellectual capacities (hence, freedom of ethical, religious, and political choice) and the faulty assumption that therefore everyone has the appropriate intellectual capabilities to assess particular truth claims.

To actually assess the truth of the matter - to learn from the debate in the Millian sense - presupposes that we have developed the capabilities to assess the truth of specific scientific claims in the first place. And that seems to require that we be scientists, and in fact specialists on the topics concerned, rather than ordinary citizens.

Take global warming. Suppose someone comes along and claims that there is a systematic problem in the interpolation models (that transform raw data from various sources such as tree-rings, satellite readings, and ice-cores into a standardised climate history that can serve as the input for global climate models). How are you as a non-scientist supposed to assess the scientific truth and significance of that claim? Or of the mainstream scientific consensus? If you do have an opinion on the matter, in what sense do you think it counts as knowledge? If you don't believe what the climate scientists say, what makes you think you know better? If you do believe them, do you consider that your belief in climate change is as well justified by an understanding of the relevant evidence as theirs?

Mill also argued that the exercise of refuting even false claims provides the valuable benefit of improving our own understanding of the truth, by having to think it through for ourselves. Thus, in refuting false claims about climate science, you would presumably come to better understand the scientific credibility of proper climate science.

Is this plausible? Climate science goes back over a century and integrates the refined expertise of dozens of distinct scientific disciplines and sub-disciplines, from meteorology to physics to biochemistry to computer modelling to statistics, as well as specialised techniques such as for analysing pre-historic temperature records from ice-cores or preserved pine-cones. Just how much of this are we required to master in order to claim to be 'thinking for ourselves' in rejecting heterodoxy? Bear in mind that the proponents of these ideas may dedicate their lives to coming up with detailed and complex arguments and defences for their positions, and a truly open minded engagement with them would be extraordinarily demanding. And don't we have other things to do with our lives, including other heterodox claims to refute?

Just treating claims about the truth as contributions to the democratic market for ideas in the first place distorts their character and assessment. It suggests that we should treat such claims as opinions, and engage in constructive mutually respectful debate about them, as if they were of the same kind as other people's opinions about immigration reform or the Republican presidential nomination. But unlike those subjects, the claims of climate science do not depend on how agreeable they seem to us. Thinking that is so allows an invasion of truth status by political status.

The first consequence is that the people believe the truth status of climate science is a matter of opinion; that they have a right to their own opinion on it; and that they have a democratic right to have their opinion counted equally with anyone else's. But counting votes, though it may respect the equal dignity of every citizen, is irrelevant to the truth of scientific claims. Even if everyone based their vote on thorough internet research, a majority view that global warming is true has no more epistemic credibility than a majority view that it is false.

Second, eliding matters of truth with matters of opinion politicises the assessment: the properly political aspect of evaluating the meaning and implications of truths is carried over into the assessment of the truths themselves. If the global warming thesis is true, uncomfortable political deliberations about its implications for our way of life would follow. It seems to some that one can win that debate by denying the truth of the starting point (e.g.).

Third, since most people are manifestly unable to evaluate the science, we reach for non-science related grounds for judgement that we do feel proficient at. In the classical analysis of rhetoric Aristotle divided efforts at persuading an audience into three main components: logosethos; and pathos (respectively, argument; the speaker's personal credibility; and emotional appeal). Since assessing the logos of climate science debates is mostly beyond us, we are considerably swayed in our assessments of it by the other two components.

So it really matters how trustworthy we think the speakers are. We may not understand what they're saying, but if we think they're not being honest, we don't have to. Hence the recent furore over the 'Climategate emails', and the general scrutiny of funding for both climate change proponents and deniers. We are also swayed by the emotional framing of the issue. In terms of saving a beautiful but fragile world for our grandchildren for example. Or in terms of defending individual liberty from the depredations of Big Government and protecting the American way of life. Framing has a directly political character because it resonates with strongly held opinions about the good life and the good society about which we already know how to disagree.

Finally, in a democracy minority opinions deserve equal respect (reflecting Mill's concerns about majoritarian tyranny). Democratic debates are limited to deciding what we should do as a society, not what individual members of a society should think. Thus, unlike in actual scientific debate, losing a political argument does not mean accepting that you are wrong, only that you haven't (yet) managed to persuade enough people to your view.


The second kind of objective truth are facts, those stubborn bits of the objective world that also refuse democracy's authority. Unfortunately these are somewhat weaker even than rational truths, since they can't be replicated. They are one time concurrences of situational factors - like Barack Obama being born in Hawaii - not timeless laws we can access at our will. Our knowledge of them depends on the frailties of witnesses and faded documents, which are easily suppressed and anyway drift apart over time. So although on the one hand the truth of facts exceeds democracy's capacities of assessment, on the other hand the fragile character of their evidence makes knowledge of them all too vulnerable to political fiat (democratic or not).

The problems I noted above apply here too, and I won't belabour the point by repeating them. However an example will illustrate the significance of the issue. Consider a fact like the holocaust that bends history all out of shape and leaves its mark everywhere. Sure we may disagree about whether that traffic light was red or green, or what the Iran-Contra affair was all about so many years later. But a fact like the holocaust is surely on a different level?

Again, look at the internet. Explore the wonderful diversity of opinions, the revisionist histories, conspiracy theories, and photographs 'they don't want you to see'. This can involve elaborate analysis of such obscure details as whether the concrete remains of 'alleged' gas chambers contained sufficient rebar to have fulfilled such a purpose. And then ask yourself how to decide truth from falsity, if you didn't already know, thanks to your upbringing under the hegemony of mainstream orthodoxy (which might well have been different if you had been brought up in Iran or Russia). It isn't possible.

Our ability as individuals to get to true facts merely by considering different arguments is distinctly limited. If we only know of one account of the holocaust - what we were taught in school - we are likely to accept it. But whether it is true or false is a matter of luck rather than our intellectual capacities. Now it is reasonable to suppose that if we were exposed to a diversity of claims about the holocaust then our opinions on the subject would become more clearly our own, and our own responsibility. They would be the product of our own intellectual capacities and character instead of simply reflecting which society we happened to be born into. But so what? Holding sincere opinions about whether the holocaust happened is all very well and Millian, but it has no necessary relation to their truth. As Harry Frankfurt notes in his philosophical essay On Bullshit, sincerity is concerned with being true to oneself, not to the nature of the world: from the perspective of truth seeking, sincerity is bullshit.

Knowing this, we can have no faith that the popularity of certain factual claims among people as ordinary as ourselves is any guide to their truth. Democracy is no more equipped to evaluate facts than rational truths. We can all, of course, hold opinions about the civilisational significance of the holocaust and its status as a justification for the state of Israel, and debate them with others in democratic ways. Yet, when it comes to the facts, neither the sincerity with which individuals believe that 'the holocaust' is a myth nor the popularity of such beliefs can make them epistemically respectable. 90% of the population denying the holocaust is irrelevant to its truth status. And vice versa.

Reconciling truth and democracy

Truth and democracy are in tension, but nevertheless truth and democracy do belong together. Good public deliberation, whether about foreign policy towards Iran or reforming education, requires a foundation of trustworthy true knowledge upon which we the people can construct sensible opinions of our own. Democracies thus require basic commonly held knowledge (such as about the history of slavery or the role of the supreme court) and on-demand access to trustworthy specific knowledge about how the world works (for example, climate science). This stands in contrast to the binary view of knowledge held by totalitarian regimes, as either propaganda or secrets of state.

It is for this reason that successful democracy requires setting up and protecting independent and non-democratic spaces and institutions - specialised epistemic communities with the authority to investigate truth. As Arendt noted, these trusted institutions include universities and law-courts with their explicitly non-political, non-democratic ethos. These are the real truth machines that are supposed to burrow after the truth wherever it may take them, and then report their findings back to the rest of us, who get to decide what to make of it.

In other words, on matters of truth we have to take the specialists' evaluation on trust. Fortunately, since we are talking about truths here and not opinions it doesn't matter that we do not come to such beliefs by our own personal experience and reflection. Therefore we can make use of another characteristic of efficient markets besides freedom: specialisation.

Specialisation is a context-specific role, not a kind of person. Scientists, mathematicians, historians, investigative journalists, jurists, statisticians, and so forth have political opinions like other people (and may also be found expressing embarrassing beliefs in conspiracy theories, and making uninformed judgements on subjects they think they understand, but don't). Nevertheless when they are working in their day jobs, within properly functioning epistemic communities, the judgements they come to deserve to be taken seriously in a way that general discussions in society do not. When the system works, we are able to deal with whacky claims about Obama's birthplace or genetically modified killer tomatoes by simply checking whether or not they came from respectable mainstream institutions (i.e. ethos lines up with logos).

The central problem for these truth machines is that they will always be in tension with the democratic values and politics of the society that set them up. Institutions like universities and courts are deeply anti-democratic and anti-individualistic. And that is how they are supposed to be. They try by various institutionalised organisational structures, values, and discipline-specific methodologies to assess ideas on the basis of their objective truth without regard to how agreeable they are to few or many people. They represent both a massive contradiction to and a necessary foundation for our shared liberal commitments about the pre-eminence of individual judgement, respect for the opinions of all, and a co-operatively determined political rule. Not surprisingly they are in fact extremely vulnerable to political pressure because "the truth" has no intrinsic power to triumph in a democracy. Quite the reverse. Setting them up and maintaining them requires a self-binding political commitment from society, a collective agreement to place them outside the sphere of political contestation.

All of this means that democratic polities are in the uncomfortable position of voluntarily giving up their authority to decide what truth is, of setting up and actively supporting somewhat unaccountable truth machines that then proceed to tell us all sorts of things about ourselves and the world that we would rather not believe.

Summing up

Mill argued that there are positive externalities to other people's freedom of expression: the chance for us to improve our own beliefs and to better understand them. I have argued that this cannot be assumed for objective truths in the way that it may be for opinions about ethics or politics. Proponents of pluralism are wrong when they argue for 'letting a thousand flowers bloom' - that the presence of vaccine-autism or holocaust-denial claims in public discourse enhances a society's quality of debate and hence understanding on these topics. Democratic debate is exactly the wrong way to treat claims about rational truths and facts. At the least, there is no good reason to encourage such pluralism.

In fact there may be negative externalities if false claims are publicised and individuals are unable to evaluate their truth. There may be direct harm, as when children die of measles because their parents read in newspapers that there really is a debate among doctors about whether MMR vaccines cause autism. There may be indirect harm if people come to believe that because they can't evaluate truth claims themselves the truth itself must be a matter of opinion (e.g. the politicisation of climate science).

I believe that the case for pluralism is actually an ethical argument masquerading as an epistemic one. It promotes two important liberal ethical commitments: people should have the right to express themselves and to think for themselves. But once we start with ethical argument there is no need to restrict ourselves to these essential components of individual moral autonomy. One should also pay some attention to the harm that pluralism about the truth can do to both vulnerable individuals and society as a whole.

This brings us to what we should do about such 'illegitimate' uses of free speech. Most interventions focus on the supply side of the market for truth claims through institution building and censorship.

First, governments seek to improve the political conversation by building up the independence, credibility and effectiveness of the real truth machines. For example, if the national bureau of statistics acquires a reputation for independence and accuracy, it can provide a standard reference point for political debate and more politicised sources of socio-economic statistics (like think tanks or Fox News) will be sidelined.

Second, and much more controversially, democracies sometimes curtail the supply of falsity into the market when the risk of significant harm to individual welfare or to the truth is judged high enough to justify insulting the rational autonomy of their citizens. For example, we may privilege the judgements of relevant scientific experts over public opinion about the medical effects or safety of certain products (e.g. by banning pharmaceutical companies from making claims for their products that the FDA disagrees with). Sometimes it may be that a truth itself is considered politically significant enough to protect from disputation by banning its denial (as is the fact of the holocaust in 15 European democracies). But of course any such censorship must be a democratic political decision, to be made in the usual way on the basis of opinion - our values, principles, and collective judgements - and not by the unaccountable bureaucracies that decide what the truth is.

This essay suggests a different and much less controversial approach focused on the demand-side of the market for truth claims. It is set in train simply by reminding people of the central argument of this essay: that democracy is not a truth machine because the truth is not a matter of opinion and the popularity of truth claims is no guide to whether we should believe them. In general, the truth is something one looks up, not something one has to decide on for oneself. If this argument works, it may create a savvier public that distinguishes between opinions and matters of truth, and is less flattered by the appeals to our reason that people flogging heterodox truth claims are so keen to make. If making such arguments no longer pays so well, we will hear them less often. Public deliberation will then focus less on debating what is true or false and more on what we should make of inconvenient truths, and will be much the better.

Prize details

This essay was published in Think, 2013, Volume 12: pp 75-88