Opinionista Jacques Rousseau 6 March 2013

Do you know what’s good for you?

Choice architecture (or “nudges”, as in the 2008 book by that name) is an example of what has been called “libertarian paternalism”, in that it is an attempt to get people to do what’s best for them. But if we know that people aren’t always the best judges of their own interests, is there perhaps room for even stronger interventions?

The sorts of people who complain about a nanny state are often the same sorts of people who know what they want, and have at least a rough idea of how to get it. By contrast, being denied a choice is less notable if it occurs in a context in which you don’t make many choices in any case.

Put another way – politically liberal folk who complain about state intrusion on their choices can be accused of an undue focus on “middle class problems”. When you have choices, it’s annoying to have them restricted. Unfortunately, this can manifest in both positive and negative ways, because for every liberal who wants to minimise state intrusion on private choice, there’s a hippie who doesn’t think they should vaccinate their kids.

The overlap here is with regard to our belief that we are being best placed to make decisions for ourselves and our families, and also sometimes our conviction that our model is the appropriate one for states to adopt.

Because I know what’s best for myself (or so I claim), I should be allowed to do it. And, if there are others out there who don’t know what’s best for themselves, they will over time – even perhaps generations – discover what they want and how to get it. The state’s role is to not get in the way of that self-actualisation.

Some take these arguments further than others. Some libertarians might argue that even prescriptions for medication are an undue restriction on my free choices. If I have consulted Doctor Google, and take responsibility for my choices, why may I not purchase medication without paying a third party R350 for a permission slip to do so?

I’ll leave the libertarian arguments to Ivo. For my part, I’m happy to identify as a liberal, but even that more moderate position is becoming increasingly difficult to justify in light of its idealistic underpinnings. I can recall having these debates in tutorial rooms in the early 90s, where we wondered whether John Stuart Mill’s harm principle could be justified with reference to typical humans, instead of the very atypical sort of human represented by Mill.

Today, behavioural economics motivates for a far more pessimistic attitude towards self-awareness and rational choice for even those middle classes – never mind those for whom simply having choices is a luxury.

For those of you who don’t know it, the harm principle is summarised in this passage from On Liberty: “the only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others. His own good, either physical or mental, is not a sufficient warrant. He cannot rightfully be compelled to do or forbear because it will be better for him to do so, because it will make him happier, because, in the opinion of others, to do so would be wise, or even right.”

For Mill, this made sense because we know our own desires and needs better than anyone else does. If others – like the state – were to estimate what those needs might be, they would have to do so by considering the average person’s interests. And of course, none of us think of ourselves as average (even though, as a matter of logic, most of us would have to be). So, to cut a very long and very interesting story short, we should be left alone to make our own mistakes, except in cases where we might cause harm to others.

This is good and well if some of us occasionally smoke ourselves to death or have motorbike accidents without wearing helmets. If reasonable precautions against harms to others are taken while smoking, your only interest in someone else’s smoking is the potential increased costs of your own medical treatment. Similarly with the wearing of a helmet, in that your only interest should be whether accidents without helmet cost more – and how those costs are covered – than accidents with helmets.

But what if we aren’t rational choosers? Or rather – seeing as we already know that we’re not – what if our irrationality is so profound that we typically make sub-optimal choices, or at least make sub-optimal choices reliably enough that something could be done about it?

The reason that this isn’t a traitorous question for a liberal to ask is because when we think of our liberty, there is perhaps a danger of thinking about being impeded in the pursuit of a particular choice, rather than thinking about how we maximise our liberty on aggregate, throughout the course of our lives.

As I mentioned in a previous column on the Western Cape’s “Get Tested” lottery incentive for HIV/Aids testing, we are all prone to hyperbolic discounting (in short, underestimating the value of later rewards in favour of sooner ones), and interventions which involve telling people – albeit subtly – what’s good for them can have very positive results, as for example in the J-PAL immunisation intervention in rural India.

Imagine that the liberty of your current self is impeded through some government agency making it difficult for you to do something. Or rather don’t imagine, but remember the last time you needed to get a medical prescription. One way of perceiving these events is as violations of your current liberty. Another interpretation is however also possible, in which your future self might be rather grateful that your choices were restricted, seeing as she now gets to enjoy the liberties made possible (in an extreme version of the thought experiment) through still being alive.

There’s no doubt in my mind that a fully competent person should be free to make self-harming choices. The question, though, is whether we are as competent as we think, for the reasons I’ve hinted at above (more fully explored in this book review by Cass Sunstein. Or more important, perhaps, the question of whether we are competent enough, regardless of how competent we think we are.

If we are not competent enough, the focus moves to what we should do about it. One option is to allow for social engineering through natural selection, whereby we make our mistakes and live with the consequences of those mistakes. But even though liberals and libertarians haven’t historically been too concerned with political correctness, embracing this view might be a challenge in that it’s likely to be the poor and the uneducated that suffer most, simply through not having the luxury of the choices many of us take for granted.

And if we don’t go that route, consistency problems soon arise, in that there’s a small step between nudges, or “choice architecture”, and banning certain choices entirely. The Conly book, reviewed in the link above, argues for a strictly utilitarian calculation of which choices should be permitted and which not, with a strong bias towards freedom.

The mechanics of and legislation underpinning those calculations is clearly a source for concern, in that we might justifiably be afraid of a state encroaching ever further on our freedom. At the same time, though, as Sunstein points out: “when people are imposing serious risks on themselves, it is not enough to celebrate freedom of choice and ignore the consequences.” DM

Gallery
0