I’ve been thinking a lot about this comment from my friend Anvesh:
In 2005, when I took the all India engineering entrance, I was ranked an incredible 51 out of 500,000 students who took the exam….I started to think greedily that if had gotten a few more right, would I have made it to the top 10 (!). I eventually met another guy whose rank was 40 or so and I found out that his score was almost 20 more than me. To put that in context, 20 marks would make a difference of at least 50,000 ranks if you were close to median. But at my rank, it would change only 10 ranks. I also later found out that the difference in scores between ranks 1 and 2 again was more than 20. Rank 2 guy might think he was so close, but in reality even if he had a particularly great day, it is likely that he still would have been the same rank. Very few people realise this. It’s like looking at the sky and thinking that the clouds must be touching the moon.
I think that last sentence is rather poetic and beautiful. But mostly I like this idea because it is counterintuitive, not something I have considered explicitly. But now I am seeing it everywhere.
I’m in a tennis phase1, and was just reading a David Foster Wallace essay about his observations at a low-level professional tournament he attended. Foster Wallace was a regionally competitive youth tennis player; he was in the top ~thousand 16 year olds in the country and he continued to play competitively into adulthood. He writes:
The idea that there can be wholly distinct levels to competitive tennis — levels so distinct that what’s being played is in essence a whole different game — might seem to you weird and hyperbolic. I have played probably just enough tennis to understand that it’s true.
I confess that I arrived in Montreal with some dim unconscious expectation that these professionals — at least the obscure ones, the nonstars — wouldn’t be all that much better than I… I have watched the competition between 64 fairly low-ranked world-class players for the eight qualifying slots in the Canadian Open field with a mixture of awe and sad surprise…I do not play and never have played the same game as these low-ranked pros.
I, in turn, will confess that I have had similar thoughts when watching MLS players play soccer (it’s not even a top league! How much better than me can they be?) or watching professional poker on TV.
It is not easy for my brain to comprehend that there are people who could be much, much better at something that I take pride in being good at. But of course that is the case, and as Anvesh articulated it becomes much more visible at the extreme tail of the distribution, where the differences between each successive rank become larger.
In college, Jeff Bezos began as a physics major and was a very good student, even by Princeton’s standards; he’s clearly got a 1 in at least 10,000 kind of mind. He recalls in this video the moment when he realized that he would never be a great theoretical physicist, because one of his classmates was much, much better at it than he was:
What we’re talking about is clear if you look at a distribution curve. The center of the distribution is very dense; most people fall around there, so if you improve even a little bit, you’ll pass many people. The tails of a distribution are sparser; they are less dense, so it becomes harder to change rank or “pass” people.
Think about running a 5k: it’s relatively easy to pass 5 people if you are an average runner in the middle of the pack (it probably just takes a few paces). It takes a much bigger effort to pass 5 people if you’re near the front where there are far fewer runners and they’re more spread out.
This phenomenon gets more extreme when the distributions are not like the bell curve above but are skewed like the ones I described in the article that Anvesh commented on.
For example, look at the top 20 companies by revenue in the fortune 500. Walmart at number 1 has revenues $250B larger than Apple at number 3! But if you go just a little further down the list, the differences are much smaller: 17-20 all have revenues between $171B and $177B.
I’ve been trying to think through the implications of this.
On is that “layering” different skills on top of one another until you can become world-class at something can be a way to adapt to not being world-class at any one skill. Bezos, for example, was incredibly good at physics, but not world-class. But then he became an options trader and learned to understand business incredibly well, and he is incredibly hardworking and disciplined. So he combined these skills to start a company and he obviously became a world-class CEO.
Another implication that Anvesh brought up to me is that effectively teaching or learning from someone who is on a different level from you is very difficult; it’s hard to put yourself in the frame of mind of someone who has drastically different abilities (and perhaps that is why the very best athletes do not make the very best coaches).
Still another is that it is hard for us to comprehend the difference in scale at the very top, and so it’s hard to determine policies to deal with the very top. It’s easy to think of the 1,000 richest people in the US as “all super rich” and to want to tax them all the same way, for example. But there are huge differences in wealth and power within those thousand, and those people should be thought of quite differently from one another.
Finally, I think it’s particularly difficult to imagine artificial intelligence that’s much more intelligent than even the smartest humans, and clearly that’ll be terribly difficult for us to regulate or control if and when it exists.
If you’re in New York and want to play, let me know. I am not very good.
I am reminded of a cartoon from decades ago. A man is alone behind a big desk, talking on a phone: "Mommy, I'm lonely".
The communication gap between a CEO and even the next highest in the hierarchy may be hard to traverse.
You mentioned me 4 times. I am flattered :)