Quoted from a conversation between Tyler Cowen and Jeffrey Sachs
Context matters especially in complex systems
The complex phenomena do not lend themselves to slogans, to easy answers, or to single solutions.
Theyâre more like engineering problems, agronomy problems, or health problems. You have to understand the context, the technologies you have at hand, the choice sets that you really have, in order to be constructive in this.
There may be room for a few pure theorists in a field like this. In my life Iâve met a few brilliant geniuses, very few, who could sit in an office and think great thoughts and contribute to the world. Itâs a very small number, by the way. Then Iâve met lots of people who generalize, which doesnât move me because I donât find it helpful. I find it distracting, confusing, misguided, or misplaced.
For most of us mortals, I think the deep engagement in real problems is crucial. I wouldnât want to train doctors without the medical students walking the wards with their mentors. I donât like training economists without them grappling with real problems in real places and learning the complexity of the interacting physical, technological, political, economic, natural systems.
Economists should act more like biologists
If Watson and Crick had written their 1953 paper saying, âAssume n base pairs.â They can match by [n à (n â 1)] / 2 combinations. It wouldnât be a very good model of DNA.
They actually said there are four base pairs, and there are two natural matchings. It happens to be a double helix.
Weâre going to study the detail out of that for the next 40 years. Yeah, itâs arbitrary. There could be other DNA, but weâre going to study this one.
Economists donât do that, because we have a harder job, in some sense, which is that weâre not studying a stable environment.
Weâre studying a changing environment. Whatever we study in depth will be out of date. Weâre looking at a moving target.
To compensate for that by never getting into detail has been our approach, but weâre always behind the curve, then.
We never have good answers when theyâre needed. Thatâs what I would like us to study.
We have so much statistical machinery to ask the question, âWhat can you learn from this dataset?â Thatâs the wrong question because the dataset is always a tiny, tiny fraction of what you can know about the problem that youâre studying.
If you want to know about the problem, get out there and learn about it. Donât think that youâre going to find it in your dataset.