4 ms·

I'm sympathetic to the above viewpoint, but the argument presented in the article sounds really dumb. > "But think about it, the authors suggest. If smarts and

by whack 8y ago

I'm sympathetic to the above viewpoint, but the argument presented in the article sounds really dumb. > "But think about it, the authors suggest. If smarts and talents and even effort are so normally distributed and wealth is so abnormally distributed, what’s missing to explain the disparity? “We suggest that such an ingredient is just randomness,” write the authors." Just because talent is normally distributed, but outcomes are abnormally distributed, doesn't suggest in any way that this is caused by randomness. A winner-take-all game-format would perfectly describe the above phenomenon. This can be trivially seen when the same logic is applied to sporting competitions. "If Tennis ability is normally distributed across the population, how can we explain a small handful of players like Federer and Williams winning an abnormally high number of championships? Clearly it must be because they're more lucky than talented."

Avshalom 8y ago
Do you have any evidence that tennis skill is normally distributed?
- im3w1l 8y ago
  The question whether it's normal distributed doesn't even make sense. Tennis skill and intelligence are both Ordinal Scales. IQ is normal but that is just because we take everyones intelligences and assign them to a value from a normal distribution. http://www.mymarketresearchmethods.com/types-of-data-nominal-ordinal-interval-ratio/ http://www.mymarketresearchmethods.com/types-of-data-nominal...
  - tpeo 8y ago
    What do you mean by an ordinal scale, and what does that imply? I'm completely blind on this usage of the expression.
    - paulddraper 8y ago
      It means that values are ordered, but not really numbers. E.g. what does 2x better tennis skill mean? An under that idea, the distribution curve doesn't really mean anything.
      - tpeo 8y ago
        It would mean what this particular variable has been operationalized [0] to mean. Once upon a time things such as "heat" were just words, and as words they were used in a variety of different implicit senses in which at the time people didn't distinguish -- for whatever reason -- but that people thought that nonetheless were informative about the thing in question. Which is to say that words were used as heuristics, partly as reasonable descriptions, partly as expressions of gut feelings. With the association of heat with thermal expansion and the development of better apparatuses (e.g. the ordinary mercury column thermometers) as well as the diffusion of this concept among laymen though stuff such as weather prediction, it became the most salient form of "heat", and thus the standard use of the word when unqualified came to mean just that: what a thermometer reads. But other forms of heat i.e. the personal feelings of heat associated with climatic comfort, or the action of capsaicin or even wounds -- and this is the origin of the word "inflammation" -- came to be used only when qualified. These are all said to be "subjective heat". Now, I can only claim that all that I've said is merely opinion, since I haven't quite yet worked out to properly cite just who might have said this prior to me and for what reasons -- and which I currently wont, because it's scattered over stuff and I don't know even how to begin -- so this is all provided as is. About ordinality, as opposed to cardinality, I can only say that I didn't see the connection at the beginning, but now that I've thought a little while about it, it does seem like a philosophically uncomfortable idea. But I'm not going to be the one to say why, at least not now. [0]: https://en.wikipedia.org/wiki/Operationalization https://en.wikipedia.org/wiki/Operationalization
- barry-cotter 8y ago
  It would be very strange if tennis skill was normally distributed. More than likely it’s a power law. Reaction time, physical strength and speed are normally distributed. The product of them will be approximately exponential. That’s almost certainly a better approximation of tennis skill. Ditto for likelihood to get rich. Intelligence, conscientiousness and luck for those. Luck is obviously an enormous factor in accumulating wealth. You’re a great deal more likely to get wealthy in the US than in Botswana and where you’re born and who you’re born to are luck.
- whack 8y ago
  You would see the same abnormally-distributed-outcomes in any winner-takes-all game where skill is normally distributed, and the higher-skilled-player always wins. The fact that skill is normally distributed but outcomes aren't, doesn't imply in any way that it's due to randomness. In fact, the less random the game is, the more abnormal its outcome-distribution would be - the best player would win every single championship despite being just slightly better than the others.
paulddraper 8y ago
Exactly. That wealth and intelligence distributions are different is simply evidence that wealth is not a simple linear function of intelligence. Wealth might have multivariate interaction: luck, race, parentage, etc. Or have no other variable interactions with intelligence, but the intelligence relationship is non-linear. Shoddy conclusion.
dnautics 8y ago
> A winner-take-all game-format would perfectly describe the above phenomenon. That's not quite true, a winner-take-all with a normally distributed reward should probably also result in a normal distribution of wealth, due to the central limit theorem. If my memory serves me correctly - i am not a professional statistician - it's been long known (but also equally long forgotten) that markets (and for sure many natural risk distributions, like earthquakes), aka, rewards and penalties, are levy-distributed, which has undefined variance, resulting in a pareto distribution.

by whack 8y ago

Avshalom 8y ago

Do you have any evidence that tennis skill is normally distributed?

im3w1l 8y ago
The question whether it's normal distributed doesn't even make sense. Tennis skill and intelligence are both Ordinal Scales. IQ is normal but that is just because we take everyones intelligences and assign them to a value from a normal distribution. http://www.mymarketresearchmethods.com/types-of-data-nominal-ordinal-interval-ratio/ http://www.mymarketresearchmethods.com/types-of-data-nominal...
- tpeo 8y ago
  What do you mean by an ordinal scale, and what does that imply? I'm completely blind on this usage of the expression.
  - paulddraper 8y ago
    It means that values are ordered, but not really numbers. E.g. what does 2x better tennis skill mean? An under that idea, the distribution curve doesn't really mean anything.
    - tpeo 8y ago
      It would mean what this particular variable has been operationalized [0] to mean. Once upon a time things such as "heat" were just words, and as words they were used in a variety of different implicit senses in which at the time people didn't distinguish -- for whatever reason -- but that people thought that nonetheless were informative about the thing in question. Which is to say that words were used as heuristics, partly as reasonable descriptions, partly as expressions of gut feelings. With the association of heat with thermal expansion and the development of better apparatuses (e.g. the ordinary mercury column thermometers) as well as the diffusion of this concept among laymen though stuff such as weather prediction, it became the most salient form of "heat", and thus the standard use of the word when unqualified came to mean just that: what a thermometer reads. But other forms of heat i.e. the personal feelings of heat associated with climatic comfort, or the action of capsaicin or even wounds -- and this is the origin of the word "inflammation" -- came to be used only when qualified. These are all said to be "subjective heat". Now, I can only claim that all that I've said is merely opinion, since I haven't quite yet worked out to properly cite just who might have said this prior to me and for what reasons -- and which I currently wont, because it's scattered over stuff and I don't know even how to begin -- so this is all provided as is. About ordinality, as opposed to cardinality, I can only say that I didn't see the connection at the beginning, but now that I've thought a little while about it, it does seem like a philosophically uncomfortable idea. But I'm not going to be the one to say why, at least not now. [0]: https://en.wikipedia.org/wiki/Operationalization https://en.wikipedia.org/wiki/Operationalization
barry-cotter 8y ago
It would be very strange if tennis skill was normally distributed. More than likely it’s a power law. Reaction time, physical strength and speed are normally distributed. The product of them will be approximately exponential. That’s almost certainly a better approximation of tennis skill. Ditto for likelihood to get rich. Intelligence, conscientiousness and luck for those. Luck is obviously an enormous factor in accumulating wealth. You’re a great deal more likely to get wealthy in the US than in Botswana and where you’re born and who you’re born to are luck.
whack 8y ago
You would see the same abnormally-distributed-outcomes in any winner-takes-all game where skill is normally distributed, and the higher-skilled-player always wins. The fact that skill is normally distributed but outcomes aren't, doesn't imply in any way that it's due to randomness. In fact, the less random the game is, the more abnormal its outcome-distribution would be - the best player would win every single championship despite being just slightly better than the others.

paulddraper 8y ago

Exactly. That wealth and intelligence distributions are different is simply evidence that wealth is not a simple linear function of intelligence. Wealth might have multivariate interaction: luck, race, parentage, etc. Or have no other variable interactions with intelligence, but the intelligence relationship is non-linear. Shoddy conclusion.

dnautics 8y ago

> A winner-take-all game-format would perfectly describe the above phenomenon. That's not quite true, a winner-take-all with a normally distributed reward should probably also result in a normal distribution of wealth, due to the central limit theorem. If my memory serves me correctly - i am not a professional statistician - it's been long known (but also equally long forgotten) that markets (and for sure many natural risk distributions, like earthquakes), aka, rewards and penalties, are levy-distributed, which has undefined variance, resulting in a pareto distribution.