Quantification – Part 1 – An Introduction
Who was the most important person in the 20th century? Gavrilo Princip? Norman Borlaug? JFK? Einstein? Mao? Ken Lay? Margaret Thatcher? Nick Leeson? Bill Gates? Maurice Hilleman?
On the one hand it’s a silly question. On the other our different approaches to answering it tell us a lot about how we think about issues of great importance. I may argue that Maurice Hilleman is the most important because he is responsible for vaccines that saved more lives than anyone else. You may respond that JFK’s “we choose to go to the moon” speech is one of the defining symbols of our era. One argument is about a quantitative difference, one is a qualitative difference.
The aim of 80,000 Hours is to help people to choose careers where they can do the most good for their chosen cause. How we go about finding out what the best career is also says a lot about how we think about the things that matter. Do you want to be a great statesman with power and prestige, or do you want to be a researcher finding out ways to cure terrible diseases? And more importantly, what is the reasoning process that you use to answer that question?
How would you find out who was the best sprinter ever? Simple, go to the record books, find who ran 100m the fastest. How would you find out who is the richest person ever? Again, find out who had the most money. In these cases you can answer the question unambiguously, you can even rank people in an exact order and say how close the runners-up came.
How would you find out who the best artist ever was? Does this question even have an answer in principle? Surely not. For every person who says that Warhol is the greatest artist ever there is one who says he’s a talentless hack. Other than personal opinion there isn’t an argument that could persuade someone they were wrong about this.
How about “What career is the most effective?” What kind of question is that? Can you list careers in an exact order from best to worst? Or is it fundamentally a matter of taste? Critics of the 80,000 Hours approach argue that, like with art, it isn’t possible to simply put a numerical value on how much good a career does. And this is especially true when predicting how much good a career will do.
What could that mean?
It could be that different careers are simply incomparable, like great art. This may be a defensible position in the abstract, but at the end of the day you must make a choice. You may claim that as a matter of principle you cannot choose between being a cancer researcher and being a trade union leader. But few people have enough time to be both. The question is how we make such a choice.
Others object that while you may be able to choose between things, in practice reducing things to numerical terms loses a lot of important information. I call Maurice Hilleman the most important person because he saved most lives but comparing vaccination with something like the Apollo project misses a huge amount of information about the significance of the two events. There is a strong point to be made here. But nonetheless, if you have a one-use-only time machine and you use it to stop Hilleman there will be more people dead than if you use it to stop JFK. That fact ought to be important, even if it does miss out a huge amount of other information.
One serious problem is when the reduction to a numerical value is difficult or impossible in practice. You cannot know how many lives will be saved by the vaccines you invent. To know that you’d have to know which vaccines would be successful. To the annoyance of inventors everywhere there is no system for finding this out in advance. In general we cannot predict with any great certainty the effect of a lot of very important things. How many lives would be saved by funding the Syrian rebels now? How much does the price of oil have to change before there is a revolution in Saudi Arabia? Nobody can know this with any great accuracy. And yet, so long as we are honest about how uncertain we are estimating gives the best information we can have and any alternative way of thinking about things can only do better by rare fluke.
Some of these criticisms don’t quite hold. But there are some real dangers to quantitative approaches, these will be discussed in the next post. It is important to take these criticisms very seriously. Ultimately, as I will argue in my last post, we will reliably do better in a very wide range of situations by analysing things quantitatively.
You might also be interested in: