Comment by buboard | Hacker News Reader

Comment by buboard | Hacker News Reader

But what did the reviewers say about the paper? (http://arxiv.org/abs/1802.07068)

They used a normal distribution for "Talent", but that's false. There is plenty of evidence that , like everything in life , it's power-law distributed

https://www.forbes.com/sites/joshbersin/2014/02/19/the-myth-...

http://blogs.cornell.edu/info2040/2018/11/16/power-laws-reve...

And the assumtions of the model are rather childish: each random event either doubles or halves your capital (In real life, the multiplier can be much much higher or lower). In fact that's the most flawed point of their tiny "model": by changing the factor 2 and the multiplier distribution they can come up with any result they want.

Also, everyone meets lucky/unlucky events in the same frequency - the presumumption that smart people make paths that provide equal opportunity as the rest or take chances with equal frequency is just false.

They could at least have compared the lifetime wealth fluctuations with real data for a sense of how wrong thir model is. This is a model of a very simple casino game, not real life.

The paper has no discussion, no justification for its parameters, no critical presentation of its limitations , no sensitivity analysis.

Good conversation piece, everyone has an opinion on the subject, but the experiment is rubbish.

Both of your links show a power-law distributions for the returns on talent, not the underlying distribution of raw talent in the population. We already know this. Wealth begets wealth- a successful book author is more likely to sell another successful book than a new author. Likewise with artist paintings and salesmen.

But actual talent, like human height or average running speed, is probably normally distributed.

> But actual talent, like human height or average running speed, is probably normally distributed.

Talent would be normally distributed if it was e.g. the sum of your partial talents, as a result of Central limit. There are different kinds of talent though, and gifted people are talented in one thing, but may even suck badly in others. If there are 5 talents, a person barely above average, with 0.6 in everything has total talent 3 and Mozart who has 1.0 only in music has total talent 1. That's not how talent works though: extremely gifted people are almost unstoppable. It would take extremely adverse circumstances for Mozart of Maria Callas or Michelangelo to not be successful. So, i don't think Talent works at all like a bell curve. It's futile to make parallels with IQ (which IS a sum of variable representing partial skills).

That's why i think those studies find those power laws (and frankly, there are not many ways to measure talent directly). Also: Note that they didn't measure the wealth of the artist, but the price of their works since the time they lived , as a long-term testament to their talent.

> Good conversation piece, everyone has an opinion on the subject, but the experiment is rubbish.

Agreed. The paper also seems to be promoting a view of wealth inequality that would support modern socialist policies.

The paper seems to be trying to scientifically support redistribution of welfare, without actually doing or saying anything, or posing the question directly.

Someone is warping the numbers to fit their agenda. Since it's a popular (and polarizing) agenda, of course everyone will have an opinion to chime in with.

Reply on news.ycombinator.com