Think back to any big decision that had you stumped. Maybe it was choosing a new car or a new apartment. Maybe it was a life-changing decision, like choosing a major at university, or a new job, or even a partner (in business or in life). Or maybe it was a decision that you would be held responsible for, like hiring a new coworker or a contractor.

Those decisions are not easy, and nobody can tell you what the best option is. They are not math problems or business investments. Rather, they are personal, subjective, and complex. Does that mean that “there are no wrong choices”?

Quite the contrary. There are many wrong ways to make big decisions. But before we can understand the right way, let us have a look at why it is so easy to fall for bad advice.

Important decisions, like choosing an apartment or a career, are multidimensional, meaning that they have many attributes you want to think of. If two apartments are identical in size, location, and everything else that matters to you, but one is cheaper, the decision is a no-brainer. Regrettably, this is not how choices present themselves in real life. Life is full of tradeoffs. The nicer apartment is more expensive. The faster car guzzles more gas. For each of those attributes (speed, gas consumption, apartment size, etc.), options are easy to compare. You know what is larger, faster, or even nicer. But since there are many attributes and tradeoffs are hard, it is not easy to decide. You need a way to balance the size of the apartment with its price and the length of the commute. You need a way to decide: a “decision method.” No problem! Many such methods are just a web search away.

So, do you have to choose among decision methods now? Isn't that even harder? Actually, no. Almost all "decision methods" that we could find intuitively reasonable are wrong. Not subjectively wrong, but truly wrong. Let us see why.

Example: Choosing an apartment

Suppose you want to rent a new apartment in your area, and, to keep things simple, you just care about the size (the larger the better), the monthly rent (the cheaper the better), the length of the commute to work (the faster the better), and whether the building has an elevator (“yes” better than “no”). You get a (short) list of empty apartments from a realtor. One apartment will be the largest, one will be the cheapest, and one will have the fastest commute, but, naturally, no single apartment will be best in all dimensions. You figure, “Hey, two or three out of four would not be so bad,” and a method is born: you will pick the apartment which is the best along as many dimensions as possible. That is, for each apartment you simply count how many of the four dimensions it is the best (size, monthly rent, commute time, elevator), and then pick the one which is best more often. Sounds reasonable? Try it yourself on this list:

• Apartment A: 900 sqft, $3,000, 40min, elevator.
• Apartment B: 800 sqft, $2,500, 1h, no elevator.
• Apartment C: 700 sqft, $2,000, 1.5h, elevator.

Apartment A is the largest and has the fastest commute, and also has an elevator, so it is best (for some dimension) three times. Apartment B is never the best. Apartment C is the cheapest and it has an elevator, so it is best twice. Three is better than two (or zero). Perfect! According to your method, you choose Apartment A. Problem solved!

Well, no. Just before you call your realtor, you get a new message. There are two new apartments in the market! Your list now has five apartments:

• Apartment A: 900 sqft, $3,000, 40min, elevator.
• Apartment B: 800 sqft, $2,500, 1h, no elevator.
• Apartment C: 700 sqft, $2,000, 1.5h, elevator.
• Apartment D: 1000 sqft, $5,000, 1h, no elevator.
• Apartment E: 400 sqft, $2,400, 30min, no elevator.

Well, the new apartments do not look good at all: D is awfully expensive and E is tiny. But you have a method, so you apply it again. Surprise! Now Apartment A is only best once instead of three times: it is not the largest, nor the cheapest, nor does it have the fastest commute, but it does still has an elevator. As before, Apartment B is never best. Apartment C is still the cheapest and has an elevator, so it is best twice. Apartment D is the largest, and apartment E has the fastest commute, but that’s it: each of those is best just once. So now you choose Apartment C!

Wait. What?!

To recap: Your method told you to choose Apartment A, even though you could also have Apartment C. Then, two other apartments showed up, and it is clear that you will not choose them, so they should not matter. But suddenly, simply because these new, irrelevant apartments are there, the method tells you to choose Apartment C instead of A. Which means two things. First, your decisions make little sense: in academic jargon, they are “inconsistent.” Second, your decisions are not really yours, are they? A clever realtor will be able to manipulate your decision by giving you some additional options, depending on whether they earn a bigger commission from Apartment A or C. If they want you to choose A, they will not show you D and E. If they want you to choose C, they will.

Most decisions methods are inconsistent

The bad news is that many intuitive-looking methods suffer from the same problem. There is a branch of the decision sciences called “decision theory” which, among other things, studies decision methods. And there are some powerful mathematical results which say that practically all methods will make you inconsistent and manipulable in one way or the other. The intuition is that, if a method makes your choices depend on how you compare one alternative to a bunch of others (for example, which apartment is the largest), then the option you end up choosing will depend on which other options are offered, even if those options themselves do not matter in the end. For example, any method that asks you to think about your future regret (how much will you regret not having done something else), or that asks you how you feel when discarding certain alternatives compared to others, will have the same problems as the one we discussed for our apartment example.

What to do, then? Is there a method that makes you consistent and prevents people from maybe manipulating you? The good news is that yes, there is. Or, rather, a class of them. But more on that in the next post...