Beyond obliquity

John Kay's new book Obliquity: Why our goals are best achieved indirectly (whose launch at the IPA was hosted last night by Rory Sutherland) follows in the tradition of a number of recent books about decision-making and rationality. Indeed, he mentions Predictably Irrational and Blink in the first ten pages. It goes beyond them in a couple of interesting ways, but leaves some questions - perhaps deliberately - unanswered.

One of my favourite books on business is also by John Kay: The Hare and the Tortoise: An Informal Guide to Business Strategy. I'd recommend it to anyone involved in guiding the strategy of their firm. You need only read a few pages to appreciate the wisdom and insight of the author, and so I have been looking forward to reading this new book.

Kay's definition of obliquity can be summarised as:
  • Aim for diverse goals other than your direct objective. The companies that make the most money are not those whose primary goal is to make money.
  • Accept the limits of knowledge and using a strategy of gradual adaptation
  • Use more judgment, less quantitative analysis
This book is more nuanced than some of its peers because of his willingness to accept that there are situations where the analytic, direct approach is appropriate. He provides a checklist outlining situations where direct approaches are viable, and others where they're not. But the deeper answer must be to find the right way to combine both.

Indeed he touches on this too:
"...it is too easy to jump to the conclusion that poetry and science do not mix, that poetry is oblique but science direct...The world that seems inevitably oblique - the world of poetry - is not without rules or criteria."
There are times in the book when Kay drifts away from this position. One of the concepts he sets himself firmly against is the idea of single quantitative measures. There can be no single profit figure, or human development index, or educational measure, which we can optimise to achieve the outcomes we want.

This can be construed as a direct challenge to the discipline of economics - which is the study of what we do when things are scarce. The immediate problem with Kay's view is that we have to make choices all the time. Because there is scarcity in the world, we cannot allocate our money or our time to everything. How can we decide between two incommensurate options? In order to choose one, there must be some process of comparing and weighting going on in our heads. Ultimately, whichever one we pick is given a measure of 1 - and the other choice 0.

But that's too simplistic an objection. There is indeed no single measure that can be optimised at all times. Think of Arrow's Impossibility Theorem, which says (roughly) that there can be no voting process which accurately ranks individual preferences into a society-wide choice. This technically applies to a whole population - and gives a clue to why companies succeed when they have multiple objectives that different people can commit to at different times - but it's also easy to see the analogy with the decisions of an individual who has multiple objectives.

Thus, when we make a decision, we are temporarily ignoring some of our objectives - and not, as rational theorists would have it, integrating them all into a single maximal utility measure. But this does not mean we have no mechanism for making the decision.

The trick is to understand how individual, small-scale decisions are made, which is a process that can be analysed. Not perfectly, of course, not with a simplistic utility measure, and perhaps with some randomness involved - but it is not a black box.
"We don't reach decisions about how to behave, what should go into a poem, what to teach or how to run a company as a result of performing some direct process that begins with abstract speculation about these large and general questions. We reach these decisions through an oblique process of negotiation, adaptation and compromise."
Indeed. But this oblique process does have some rules and some recognisable, repeatable, analysable components.


Ironically, Kay disapprovingly quotes Herb Simon on artificial intelligence - but Simon was the first to coin the phrase "bounded rationality", recognising that people cannot take everything into account when they make decisions.


Kay's introduction suggests that, instead of complaining that people are irrational, we should change our understanding of rationality. I couldn't agree more. But there is a meaningful way to talk about rationality, which reflects both how we really make decisions and also understands why we get good outcomes from doing so. It simply requires scaling down our ambitions and looking closer at the fine grains of behaviour.

Physicists have learned to apply the laws of motion of a pendulum only to long pendulums slowly oscillating at a narrow angle. When they retain this modesty, their predictions are virtually perfect. But if you try to predict how a ball on a chain will move when violently jerked or pulled out to a 90% angle, you will get it wrong.

Economists are meant to know this too. Microeconomists focus on marginal changes because they are tiny movements in a continuous variable, so that certain assumptions which held before the change will still hold throughout and after it. But if they extrapolate from a marginal change to assume that all changes are marginal, they fail.

So don't throw away your rules; just know their limits. Sometimes, with the right mathematics, you can derive large-scale behaviour from small-scale decision rules. The two are rarely the same, but they are still related.

A table in chapter 13 ends with this comparison:

TopicThe directThe oblique
Process rationalityGood decisions are the product of a structured and careful process of calculation.Good decisions are the outcome of good judgment.

In fact, good decisions are the outcome of good judgment, which in turn is the product of a process. We don't know what that process is while we're doing it, but what goes on inside that judging mind is capable of being understood. Kay doesn't explore that question - it's not that kind of book - but I hope he'd agree that it's a valid question to ask.

In the final chapter, Kay dismisses the concept of decision science. In its business-school definition: a science of how we should make decisions - he's right to do so. But in its meaning in the psychology department - a science of how we do make decisions - it's valid and indeed immensely powerful.

And this is where we come back to Rory Sutherland, who introduced last night's talk in the middle of the IPA's course on behavioural economics. Rory knows his audience and exactly how to perform for them. So I'm not sure whether we disagree on the value of scientific, testable cognitive models, or if he just chooses not to focus on them. [Rory's comments on obliquity are here]

The task for behavioural economics now is to go beyond enumerating the ways in which we depart from economic rationality, and to integrate the low-level discoveries of cognitive decision theory into models that can say something about why our behaviour and our judgment are the way they are.

Just like obliquity:
"Obliquity doesn't mean that we should stop thinking about objectives, fail to examine options or omit to seek information and understand as best we can the complex systems that we deal with. Far from it: we should start and continue. The alternative to a 'rational' process of defining objectives, evaluating options, modelling consequences is an approach that is oblique, but truly based on reason and evidence."

Comments

Popular posts from this blog

Is bad news for the Treasury good for the private sector?

What is the difference between cognitive economics and behavioural finance?

Dead rats and dopamine - a new publication