Things to Hang on Your Mental Mug Tree

Things to Hang on Your Mental Mug Tree

Rory Sutherland [7.10.17]

I don't think there's any huge amount of intelligence required to look at the world through different lenses. The difficulty lies in that you have to abandon four or five assumptions about the world simultaneously. That's what probably makes it difficult.

RORY SUTHERLAND is Vice-Chairman, Ogilvy London; Columnist, The SpectatorRory Sutherland's Edge Bio page


On a list of ten things that matter in terms of how you look at the world, satisficing would be number one, along with the whole business of decision making under uncertainty. This opens up the idea that something which may seem at first glance to be completely irrational may in fact be a useful and highly effective non-catastrophic heuristic.

At around the same time as Herbert Simon was doing his work, David Ogilvy was talking to someone who ran the Chicago office of Ogilvy called Joel Raphaelson [Joel, who is still alive and as quick-minded as a 25-year-old, is the son of Sampson Raphaelson, scriptwriter for The Jazz Singer and Heaven Can Wait]. David agreed with Joel’s idea that people bought brands to avoid badness rather than to maximize perfection. They bought brand A over brand B not because they thought it was better, but because they were more certain it was good.

The idea is that when you make decisions in an uncertain setting, you have to care about not only the expected outcome, but also the possible variance. We'll pay a premium not only for "better," but for "less likely to be terrible." That seems to be an important thing to understand when analyzing decision making.                                 

Human decision making is also pretty path-dependent. In one case in my life I've been able to profit from this. I live in a house, which in the UK is something called Grade 1 listed. It's by the great eighteenth century Robert Adam, and the grounds are by Capability Brown. I'm in a four bedroom flat on the roof of a house built for the doctor of George III in about 1785. For a time it was the home of Napoleon III.

I didn't pay anything extra for the architecture or the genius of the landscaping because nobody places that very high on their list of priorities when they buy a house. I asked our next-door neighbor, an economist, how much he thought we paid for this property relative to a property in an identical location of the same size by an indifferent architect. In other words, what's the premium we pay for the Robert Adam-ness of it? He said somewhere between zero and about 5 percent.

That's quite interesting if you think about it. When we buy a property, the order in which we look at things places location as the highest priority. Next. In the UK, it might be the number of bedrooms it has (in the US, it might be the floor area, by square footage). We might then look at the size of the garden, a few other features, and whether it has a pool. But architecture generally comes pretty low down the list. We only look at architectural aesthetics when we've got down to a final selection of four or five.

Whereas a painting by a great artist might cost literally 10,000 times more than an equivalent-sized painting by an indifferent artist, a building by a fantastic architect costs about 1-3 percent more than a building by an average or even indifferent architect.

If we bought paintings the way we bought property, we'd say, "I want a painting that's exactly this size; I want these three colors to predominate; I want it to be in exactly these proportions." In the fifth iteration or so, we'd get down to who the artist was. Most of the time, we wouldn't end up with a Picasso. It seems to me that architecture is an incredibly cheap way of buying art, compared to art.

If you're interested and want to take advantage of this, go to a website in the US called, which shows Frank Lloyd Wright properties for sale. In most cases, they seem to be no more expensive than the neighboring property. Similarly, there was a Gropius flat that came for sale in the UK. It was expensive because it was in Notting Hill, but no more expensive than a totally average building next door. What intrigues me about human decision making is that there seems to be a path-dependence involved—to which we are completely blind.

Number two? Costly signaling theory is also pretty useful. I suppose the peacock's tail is the standard view. What seems undoubtedly true is that humans, like peahens, attach significance to a piece of communication in some way proportionally to the cost of generating or transmitting it. Instinctively, you will open the FedEx package on your desk before you open the letter. One of the problems of email is there's no cost attached to sending it, so we don't have a useful heuristic for deciding what to read first. That strikes me as interesting.

Cost might mean anything from financial expense to use of a scarce resource—anything that requires something in scarce supply, whether it's money, talent, effort, or time. The more that is invested in the communication, essentially, the more weight that communication carries. That makes perfect sense because, in some ways, that means that your sincerity is hard or costly to fake. That could also apply to creativity—the effort required to communicate humorously, which you have invested up front, connotes something about the sender’s intent.

Let me give you an example. You receive two wedding invitations on the same day, one of which comes in an expensive envelope with gilt edges and embossing, and the other, which contains exactly the same information, is in the shape of an email. You're probably going to go to the first wedding. (The second one, well, there's a dangerous implication there might be a cash bar.) Now, what happens if you haven't got much money, but you want to do an impactful wedding invitation? Then you make it really creative.

If you can't put actual money into the paper or the printing. What you can put in is something different, which is imagination. You can make it a fantastically imaginative or beautiful [or humorous] wedding invitation.

But there has to be some investment of a scarce, costly thing. It's the difference between saying your horse is going to win and visibly betting on your horse. In the latter case, you have skin in the game; in the former case, you don't. That's important to understand. To some extent, effective communication will always require some degree of inefficiency because if it's perfectly inefficient, it then becomes meaningless.

There's a rather lovely company in the UK that pays people who are housebound—whether for medical reasons, or are caregivers—to handwrite envelopes and letters. You could regard this as a very silly thing to do, but in costly signaling theory terms, it makes perfect sense. The open rate of these letters, and the response they generate, is an order of magnitude higher than for laser-printed letters.

Another thing worth bearing in mind is countersignaling, which, unlike signaling, seems to be uniquely human. There aren't cases of peacocks who demonstrate their extraordinary genetic quality by having really shitty tails. What seems to happen with humans is you have multiple parallel status currencies, and quite often you will signal your position on status by adopting none of the status currencies of the class immediately below your own, or by essentially demonstrating zero effort in standard status currencies. An unwashed bass guitarist in a cool rock band, for example, can get away with poor levels of hygiene, which signals: "I'm so sexy by dint of my bass guitar playing skills that I can get away with not making an effort in any of these conventional areas." Sometimes it's done as a positional thing, and sometimes it's done as a pure demonstration of handicap.

Relevance theory [from Dan Sperber and Deirdre Wilson] might be another thing that's interesting. In other words, replacing the “conduit” idea of communication with this idea that we communicate the minimum necessary for the recipient to recreate the message within their own head using context as a very large part of the information. Those interesting new theories of communication, which don't always sit with the Claude Shannon theories, are worth exploring. A very simple manifestation would be jokes which, like IKEA furniture, demand some self-assembly on the part of the recipient.

Quite a lot of advertising works heavily on the basis that 10 percent of it is information, and the rest of it is inference. I suppose that's the way to put it. We will understand the communication very differently depending on context.

The general idea of advertising is that it manipulates people into liking things more than they should. Now, I'll dispute that. I completely agree with various psychologists who point out that marketers undoubtedly exploit attentional bias. The only defense I'll make is that we will exhibit attentional bias anyway if left to our own devices. Changing what people pay attention to, and what people think of as important is quite useful in terms of any form of innovation. Most significant innovations effectively operate by ignoring the established rules of a category and creating some new rules somewhere else. This often requires a shift in the consumer’s attention from the strength of the old thing to the strength of the new thing that replaces it.

There was a period where the whole gig about mobile phones was about miniaturization, how small they were. That reached levels of absurdity, literally, the reductio ad absurdum, where phones were so small you could hardly retrieve them from your pocket. Eventually, people stopped paying attention to smallness, and marketers encouraged them to pay attention to usability and large screen size—you now pay more for a bigger phone. We stopped paying attention to one thing, and we started using a different heuristic. Marketing, in some ways, makes markets more dynamic by preventing people being fixated for too long on any one particular dimension of comparison.

An interesting example which suggests there's a large amount of inference and context that goes into our contextual, metacognitive interpretation of advertising, is that in Eastern Europe under communism, if you advertised a product, demand often went down.

Why was this? If advertising works as we conventionally think it does, without contextual translation, how on earth can that happen?

But, under communism, anything that was worthwhile or desirable was generally in short supply. Consumers inferred that the only possible reason that the government might be promoting something was that they'd accidentally managed to produce something of such unremitting crappiness that people weren't willing to queue for it. Advertising in that context told you what not to buy.

The conventional view of advertising is that when you say something is good, perceived value rises. In fact, you could argue that quite a lot of advertising is a reliable “costly signal” because you wouldn't do it if you had a bad product.

It's an expensive activity, which disproportionately pays back when it’s for something that becomes widely and repeatedly popular over time, but leads to disproportionate losses when you are selling something where customer satisfaction and hence repeat purchase are going to be low.

In other words, it's a sunk-cost demonstration of faith in what you're selling, which is sincere precisely because there's a cost attached.

We need to be alert to this because there's a move in advertising to make it more efficient by making it all digital. Arguably, when you make it more efficient and less costly, you make it less convincing.

We shouldn’t forget that traditional mass advertising has the virtue of taking place in a mass medium. A promise you make to multiple people simultaneously is a bigger bet than making a promise to one individual at a time. When we get married, we state our vows in a crowded church full of 200 people rather than going door to door reading out our vows to people one at a time. In a mass setting, only one person has got to call your bluff and you've lost. It shows instinctively more confidence to make a promise in a public space than it does to make it one person at a time.

All I am saying is that we have to be careful, when we obsess about the newly efficient digital modes of advertising, that we aren't throwing away the baby with the bathwater.

If you take programmatic digital banners, displays, you might be reaching exactly the same audience as a TV ad does. But if you're creating a lower level of conviction in that same audience, what you're doing may appear to be efficient, but it's less effective. Costly signaling is one of those mental tools you can't do without.

Okay, mug number five. Nassim Taleb is very interesting on the Lindy effect. He believes that if you have a worthwhile finding in social science or psychology, you will find mention of it going back to the ancient world. You might argue that Aesop's "The Fox and the Grapes" is an early instance of avoiding regret. (Aesop was an early behavioral economist, in some ways.)

I had an interesting conversation about this with HMRC (Her Majesty’s Revenue & Customs), the people who do taxation in the UK [and the same wonderful folks who gave you the Boston Tea Party]. My contention is that when you give people information—or a choice—that contains a bit of bad and a bit of good, or even quite a lot of bad and a little bit of good, the natural human reaction is to amplify the good and minimize the bad.

Sometimes the opposite of sour grapes is called sweet lemons. Sweet lemons is when we say, "Yes, I did spend three years in prison, but if I hadn't had that time in prison, I never would've met those interesting people and had such a formative experience." We take things that perhaps a more objective person would think of as unremittingly awful, and we construct a little bit of a positive out of it.

I explained to the tax people that the way tax is done, where there is absolutely no tradeoff, it provides us with no means of performing that mental trick. Now, try to imagine that if by paying more income tax, you got a small amount of a token benefit. For example, let's say that when you pay more income tax at age thirty, it might increase your inheritance tax threshold—the amount you could pass on to people—or your base level tax rate in retirement would go up a little bit—just some small upside. Would we hate paying tax a lot less?

Let me tell you a little bit about providing people with the opportunity to construct a more positive story. In ancient Rome and Greece, they did this. If you paid wealth tax, you got your name on a column, honoring your contribution. There was now some reputational upside.

In fact, it worked so well that people came out of the woodwork and said, "You haven't asked me to pay this wealth tax, and I'm much wealthier than him. I should be on the column, too." This is an example of a status benefit, but there are other benefits that you can write into a narrative involving trade-off.

The most extraordinary thing happened to me, which suddenly made this Aesop point very well. Aesop is what, 7th century BC? It's a comfortable two-and-a-half thousand years old as an insight, so it's stood the test of time.

Okay, so I'm landing in an Easyjet plane. Normally, when you get a bus to the terminal, you're really pissed off. You say, "Dammit, I'm being shortchanged here. I was expecting an air bridge, and you're going to drive me to the terminal in a ruddy bus." Generally, everybody's kind of spitting and angry.

But, in this case, the pilot said something extraordinary.

He says something I've never heard before. Instead of saying, "I'm terribly sorry, we haven't been able to get an air bridge, so you'll be bused to the terminal. If you just wait while the bus draws up on the port side of the aircraft—" No. Instead, he says, "I've got some bad news and I've got some good news. The bad news is we haven't been able to get an air bridge because there's a plane blocking our gate. The good news is that the bus will take you right next to passport control, so you won't have far to walk with your bags."

Suddenly, I realized that actually that's always true. When you get a bus, there's an upside to the bus, which is you don't have a long walk through a shopping center with your carry-on luggage in order to get to passport control and then the baggage carousel. But because no one had brought our attention to it, we had no opportunity to derive the positive.

Here's a little social science experiment everybody can do: The next time there's a bus and they simply announce, "I'm afraid we'll have to bus you to the terminal building," say quite loudly, "I'm pleased there's a bus because the bus drives you all the way to passport control so you don't have to walk there with your bags." You will have synthesized a dose of happiness among everybody within earshot without changing objective reality at all simply by changing the way that people look at something

I thought that pilot was a genius because there always was an upside to the bus, but because bus replacement services on trains are always widely hated, we never even assumed there was one to find. The second it was pointed out to us, our attitude to having a bus is completely changed.

This comes down to Robert Cialdini’s argument in his book Pre-Suasion, which is about the fact that we automatically assume that what we pay attention to is what's important. Another valuable concept is Daniel Kahneman’s dictum that “nothing is as important as we think it is while we're thinking about it.”

There's a value to marketing. This is a very strange, and some people might say massively self-interested defense of marketing, but it's perfectly possible to create value not by changing the material world at all, but simply by changing the way we direct our attention or the way attention is directed.

In environmental terms, you can say this has rather a lot of potential. If you can change people's focus, attention, and their status currencies so they derive more pleasure from what already exists, rather than from what has to be created to sate their demands, you can essentially increase wealth without increasing consumption. That seems to me a rather important finding. One of the dangerous assumptions of economics, among many, is that because they assume that we are already blessed with perfect information, perfect trust, and they assume that we are already making optimal use of the money we have, the only way in which you can increase wellbeing—according to economic assumptions—is essentially by burning more things.

I'd argue that intangible value, once you learn to respect it, shows that you can make people very content with less. Think about it. A Prius is a reasonably high status car. It's an interesting car, the Prius, in the sense that . . . you probably have to be a bit of a liberal to get away with this; I don't think this works in Texas . . . but a megastar can turn up at the Oscars in a Prius and no one would think there was anything remotely weird about it.

If you generally define wealth not in purely financial terms but in the number of choices people can freely and enjoyably make, well, by essentially enabling people to have a modest car without any stigma being attached to it, that's a pretty useful thing you've just done in terms of wealth creation.

I don't think there's any huge amount of intelligence required to look at the world through different lenses. The difficulty lies in that you have to abandon four or five assumptions about the world simultaneously. That's what probably makes it difficult.

~ ~ ~ ~

One essential problem is exactly the same in the private sector, the public sector, or indeed, the charitable sector, which is that most decisions are made in a risk-averse manner. In most jobs, there's an asymmetric reward mechanism. If you do something well, you get a pat on the back. If anything goes disastrously wrong, you lose your job.

Very simply, in individual decision making, we are very strongly motivated by fear of regret. In collective decision making, in institutional decision making—in business or in government—the thing that drives us is fear of blame. Cleaving to a narrow, simplified rational model is usually a good way to protect yourself against blame.

I have a simple mantra here, which is that it's far easier to be fired for being illogical than it is for being unimaginative. If you make a decision that proves to be disastrous but it is based on solid, logical foundations, you keep your job. If you make a decision that may be a better decision on average but involves a degree of counterintuitive thinking, second-order thinking, or indeed, a degree of subjectivity, well, if it goes well, you won't get much more credit than you would using the logical approach. If it goes wrong, you're now out of a job, you're vilified.

Gerd Gigerenzer calls this defensive decision making. That is a useful thing to understand. In medicine it probably leads to massive over intervention. (It's much easier to get sued for not doing something than it is for doing something.) This defensiveness creates extraordinary distortions of decision making.

What happens is that, unconsciously, we're not thinking “what's the best decision here?” What our instincts tell us to do is to choose whatever course in which the worst-case scenario is least catastrophic for us personally. That's one of the reasons why business meetings proliferate to such an extent, which is that it's a way of burying decision making in the collective, rather than forcing any one individual to take responsibility.

One thing which emerges from this is habit and groupthink. When people choose very conventionally, the consequences of failure are less catastrophic. The downside risk of a bad decision is proportional to how weird it is.

 This is probably why there are four big accounting firms. If you appoint one of the big four and something goes wrong, people will blame, say, Price Waterhouse. If you appoint a small boutique accounting firm (which might be faster, cheaper, and better in most objective measures) if anything goes wrong, they'll now blame you because you've taken a nonstandard decision.

That very strong drive to conformity and risk avoidance probably explains the necessity of entrepreneurs in a business sense, the kind of people who aren't constrained by that institutional thinking.

But it's also something to be alert to in understanding how people decide. Government and business will both tend to play it too safe. They are all operating under certain norms or assumptions that are safely traditional, yet some of which are probably wrong. Neoclassical economics creates one such template, which risk averse people can always cling to because no one will ever get fired for assuming that economics is true. Make decisions assuming complete economic rationality in all actors and, even when it’s a terrible decision, your job is safe.

Let's imagine you want to compete with Coca-Cola. You're thinking that for about 150 years it has been the most popular cold, nonalcoholic drink—other than water—anywhere in the world. We want some of that action. You go in and you say, "We need to produce a drink that tastes nicer than Coke, costs less than Coke, and comes in a big bottle so people will get great value for money." No one in that meeting is going to say, "That sounds pretty wack." Everyone is going to nod along.

So you research the drink, and it does indeed research better than Coke. You've reduced the price. All of those things seem wonderful. You will never lose your job for doing that, even when the tasty, cheap, voluminous drink is a total failure in the market.

The only problem with this is the most successful attempt to compete with Coke in the last fifty years is Red Bull. This costs a fortune, comes in a tiny can, and it tastes disgusting. Patently, the psychological processes at work here are not the same psychological processes that would be assumed by a first-year economist.

(I suspect there's something going on with the placebo effect there, which is that in order for us to think that something has psychotropic or medicinal powers, it has to taste a bit weird. Health food basically tastes a bit shit, doesn't it? Wheatgrass is like licking the underside of your lawnmower. We tend to infer there's a trade-off in these things—that if it's going to have psychotropic or medicinal effects, there's going to be a downside somewhere. We believe in Red Bull’s energy-boosting powers in a way that we wouldn't believe from a nicer tasting drink.

There are some interesting opportunities for entrepreneurs in taking a product and adding a negative. Green tea is a possible example; we believe it’s wholesome because it tastes weird.

~ ~ ~ ~

You probably know the work that Shlomo Benartzi has done on the Save More Tomorrow Pension, which is effectively a case where you take two pensions that should be economically identical, if we were indeed homo economicus, and yet the rate of uptake on Shlomo's pension, although it “costs” exactly the same, is two or three times greater. The amount saved is also greater.

This new, evolved-brain-friendly pension is ingeniously designed it so that, rather than paying a fixed amount, a proportion of all your pay rises go into a pension. That way you never get poorer through having a pension, you just get richer slower. Now, to economists, those two states are identical. In our real-world human brains, however, for fairly obvious evolutionary reasons, we feel losses more acutely than we enjoy gains.

One of the great mistakes that the British pension policy has made is that it hasn't set a low enough limit on how much we can save. That seems completely counterintuitive—the idea that people would save more if there was a maximum amount that you could save.

But, in reality, we are rivalrous creatures that probably compare our savings to each other, and the limit would set an implicit norm. If we saved well below our limit, we would feel we are missing out. And a limit would set an implicit target. We might not save up to the limit, we might say, "Well, I better do at least half," whereas having no limit provides no anchor point.

I have some evidence to suggest that I might be right on this theory. There is a British savings product, which is called the ISA. It used to be called the PEP. That does have a maximum (it is going up to £20,000 this year). What was remarkable is that a very large number of people saved in ISA, and what they saved would tend to be either the whole amount or a fraction of the maximum amount allowed. But economics won't get you there because it's only a very partial explanation of human motivation. Economics is the study of human motivation but with all the interesting variables set to zero, other than greed or prudence. Everything else is set to zero. 

~ ~ ~ ~

Taleb's recent work on the minority rule is fascinating. This interests me as a marketer. Whereas we tend to think that markets are the result of aggregated preferences, actually, a small minority of people who veto something can have a surprisingly large effect on a marketplace if the people who are non-vetoers are happy to go along with a universally acceptable alternative. 

So, if you have a school in the UK where 5 percent of the pupils are Muslim, the whole kitchen will go halal. Why? Because non-Muslims don't mind eating halal food. Muslims will only eat halal food. As a result, the convenient thing is to have one kitchen that can serve everybody, so everybody eats halal. (Slightly unfair to Sikhs, incidentally, who aren't supposed to eat halal food, but they don't seem to make much fuss about it.)

Similarly, if you have a party that is more exclusively male, you can almost be certain that the default drink is beer. However, if you have a mixed-gender party, something new enters the equation. About 18 to 20 percent, maybe as many as 30 percent of women won't drink beer under any circumstances. Men, however will, without much encouragement, drink pretty much anything.

Therefore, if you have a group that is 50-50 female-male, and let's say 20 percent of the entire group won't drink beer, everybody has wine. It's the type O negative, the universal donor. Pizza is probably very popular as a foodstuff because nobody really hates it.

Understanding that complex systems like consumer markets sometimes have slightly counterintuitive rules to them is a useful thing.

With regard to government, I don't think we've yet satisfactorily made the case for why nudging is preferable to, say, legislation or using economic incentives. We need to make this case again and again. Let me give you a very simple example here. Let's say you use the price mechanism to get people to use their gas and electricity late at night rather than during the day.

You have several tariffs on a smart meter, and you charge people a lot more for daytime electricity use than you do for electricity use in the evening. Therefore, people are incentivized to put their tumble dryers, their washing machines, and their dishwashers on in the evening rather than the middle of the day. This generally reduces peaks in demand in the network. That then reduces the number of dirty power stations that have to be switched on at times of peak demand. Generally, that's a victory in terms of carbon emissions.

But I would argue that the problem with using the price mechanism or legislation is that they are universal blunt instruments. There are an enormous number of cases where you don’t need to change everyone’s behavior—you just need enough people to change to make the system more effective. You don’t need everyone to become nocturnal launderers—just enough to take away the worst of the daytime peaks.

Setting doctors' appointments would be another case where nudging could be optimal. One, it's unacceptable to use the price mechanism in the UK because it's an item of religious faith in the UK that we don't pay to visit the doctor. But it would be preferable if retired people or people who didn't work went to the doctor in the middle of the day so people in jobs could visit the doctor at either end of the day before they went to work or after they got home. You could achieve this easily with gentle nudging by appointment bookers.

Now, the thing about using the price mechanism is that in some cases, it's unacceptable. It's also unfair to certain groups. If you take the group of people I was talking about, who you're encouraging to use their dishwashers late in the evening, it's a nice idea, but it's also unfair to people who work nights. The great thing about persuasion, versus pricing or legislation, is that if you have a good reason not to play ball, you’re free to go your own way.

If you take persuasion rather than compulsion, for example, with organ donation, there will always be 5 percent of people who for some reason or other object to what you want them to do. They may have perfectly good reasons. In the case of organ donation, it may be religious. Certain religions believe you have to keep all your organs intact after death. I don't necessarily think you're right to believe that, but I respect your belief. If you nudge people, everybody has an out. If you fine people, you end up punishing a load of people who may have perfectly good individual reasons for objecting to your intended direction of behavioral shift.

Essentially, you should try persuasion first, bribery second, and legislation third. What tends to happen in government, because it's their natural mode of approach, is they tend to use legislation first before they've even tried persuasion. The point is that legislation is a blunter instrument than persuasion. One of the beautiful attributes of persuasion is people can always give a good reason and choose not to be persuaded. No reputational or other costs are incurred so long as your reason for opting out is generally acceptable.

I imagine it was Charlie Munger who lists about ten or fifteen things out of a list of forty or fifty that it really helps to hang on your mental mug tree. You might not want to use them all the time. Quite a lot of the time, just using standard rational approaches may be absolutely fine, but it's useful to have a mental mug tree of the complex and counterintuitive.

When you get stuck, the likelihood is that you're looking at something in the wrong way, so just grabbing one of these mental mugs might be the "get out of jail" card that moves your thinking on.

I'll finish with one more, which seems to me something I've noticed, which may help you increase diversity in your workforce without resorting to quotas or other compulsion.

It seems obvious that people instinctively increase the variance of things as they buy more of them. When people do small weekly shops, they go to a greater variety of shops than they do when they do one enormous £150 shop every ten days.

If you shop every three days, you'll go to a lot of different shops, some upmarket, some downmarket. You might go to Walmart one day and Whole Foods the next. If you do one enormous shop, you'll go very close to the middle. You'll go somewhere that's absolutely midmarket.

Similarly, when everybody had one car, most families had a saloon car at the time, a fairly standard car. People with two cars almost never have one saloon car, let alone two. In fact, they tend to have two cars that are wildly different—one very large one and a smaller one. Variety increases with frequency, as it were. That seems like a relatively obvious point.

 If I asked two people each to go and spend half a million dollars on a house, you'd probably have two quite middle-of-the-road houses. If I asked you to spend $1 million on two houses, the two houses would not be remotely similar. One would be a flat in the center of town, the other one would be somewhere miles out by the beach. All obvious enough, right?

Now, what seems interesting to me is that no one has applied this thinking to recruitment. If you want greater diversity of recruitment, hire people ten at a time. When you hire people for one job, one at a time, you're very risk-averse, and you go for someone close to the expected norm. If you're hiring ten graduate recruits, you're looking for breadth. This happens without any quotas, without any conscious affirmative action. It seems to me a natural property of the human brain that the way we decide changes in variety simply by dint of choosing a lot simultaneously rather than choosing one person at a time. That seems to me an enormously important thing.

This is borne out by the fact that the partners in accounting firms show much less diversity than the graduate recruits do. You appoint partners one at a time, and you hire graduates en masse. The fact that our instinctive mode of choice changes depending on the way we choose, this seems to me just adding to the literature on choice architecture. When you make a choice, it's always worth asking, "How would I make this choice if the choices were presented differently, if they were presented in a different order?" 

Let's talk about motor insurance. How do you get people to tell the truth? One of the things that's been reliably found is that if you get people to sign at the beginning of the process rather than at the end, they're more honest. This seems revolutionarily weird. In the context of a form, we always sign the form at the end. Yet when we have a court trial, you swear the oath before you start giving evidence, not after you've finished.

What's very interesting is this domain dependence, where things which are bloody obvious in one domain, like where you get people to swear an oath in the context of a court, become completely lost when we shift domain to the area of an insurance document.

And, on those forms, aren't we kind of encouraging people to lie? What happens if you redesign an insurance application so you give people the chance to tell the truth again at the end? I don't want to get into trouble here, but I have in the past been mildly dishonest on insurance forms for my car. I've claimed that the car was garaged when, in reality, it was probably only garaged one night in twenty. When I leave the car out, it's a fairly long way from the main road, so it’s not significantly at more risk than if it were garaged. So I’m not lying much. But the reason I veer on the side of dishonesty is for all I know, it might cost me £100 more a year to tell the truth. Even £200. The cost of insurance is not transparent. I don't know how much it costs to tell the truth, so I don't know whether the cost of telling the truth is worth it for the feeling of being an honest person. 

If I get to the end of my insurance form and it quotes me, say, £340 a year, and then it gives me an extra last-minute option to be honest by saying: "If you don't want to garage your car every night, you can park it on the drive for an extra £50 a year," now it's a finite quotable amount. I'll tell the truth and pay £50 pounds just for the good feeling that results. It's when I don't know what the cost of honesty is that I'm disproportionately encouraged to be economical with the truth.