2 ms·

Just a mathematical question : can we systematically distinguish randomly distributed luck from randomly distributed competence ‎?

by Rexxar 11y ago

EliRivers 11y ago
If someone is competent, we can expect them to continue showing competence into the future. If it's just luck, we can expect their future performance to revert.
- Rexxar 11y ago
  I'm not sure if it's enough when you study a group, during a finite time, if you don't know a priori the relative impact of competence compared to luck. Example 1 : fictive world where everybody has 50% chance of success every year (success = 1 point). After 4 years the distribution looks like : 0pt -> 1/16 1pt -> 4/16 2pt -> 6/16 3pt -> 4/16 4pt -> 1/16 Example 2 : fictive world with 6 competence levels : group 1: always lose, population 1/16 group 2: win 1/4, population 4/16 group 3: win 1/2, population 6/16 group 4: win 3/4, population 4/16 group 5: always win, population 1/16 If the final distribution is not Gaussian, I think we can distinguish luck vs competence but if the observed curve is Gaussian I'm not sure we can distinguish luck vs competence.
  - EliRivers 11y ago
    If we repeat for another four years, resetting the scores to zero, the relative rankings of people in the first group have swapped around, but the relative rankings of the second group has not (or at least, has changed to a much smaller degree). We could run this experiment over and over (two periods of four years), showing me only the relative rankings of the participants in each paired period (i.e. I get to see the results of the two random trials together, and the results of the two competence-based trials together), and I'd have a very good record of identifying the luck based competition from the competence based competition. The real world is a mix of luck and competence, but if I see the same names coming out top in each periodic batch into the future, I can safely assume that they are not operating on luck. Given enough people there will always be someone who tosses heads twenty times in a row, but the majority of luck-based achievers will perform averagely in the future, while the competence based achievers will perform according to their competence.

Just a mathematical question : can we systematically distinguish randomly distributed luck from randomly distributed competence ‎?

by Rexxar 11y ago

Just a mathematical question : can we systematically distinguish randomly distributed luck from randomly distributed competence ‎?

EliRivers 11y ago

If someone is competent, we can expect them to continue showing competence into the future. If it's just luck, we can expect their future performance to revert.

Rexxar 11y ago
I'm not sure if it's enough when you study a group, during a finite time, if you don't know a priori the relative impact of competence compared to luck. Example 1 : fictive world where everybody has 50% chance of success every year (success = 1 point). After 4 years the distribution looks like : 0pt -> 1/16 1pt -> 4/16 2pt -> 6/16 3pt -> 4/16 4pt -> 1/16 Example 2 : fictive world with 6 competence levels : group 1: always lose, population 1/16 group 2: win 1/4, population 4/16 group 3: win 1/2, population 6/16 group 4: win 3/4, population 4/16 group 5: always win, population 1/16 If the final distribution is not Gaussian, I think we can distinguish luck vs competence but if the observed curve is Gaussian I'm not sure we can distinguish luck vs competence.
- EliRivers 11y ago
  If we repeat for another four years, resetting the scores to zero, the relative rankings of people in the first group have swapped around, but the relative rankings of the second group has not (or at least, has changed to a much smaller degree). We could run this experiment over and over (two periods of four years), showing me only the relative rankings of the participants in each paired period (i.e. I get to see the results of the two random trials together, and the results of the two competence-based trials together), and I'd have a very good record of identifying the luck based competition from the competence based competition. The real world is a mix of luck and competence, but if I see the same names coming out top in each periodic batch into the future, I can safely assume that they are not operating on luck. Given enough people there will always be someone who tosses heads twenty times in a row, but the majority of luck-based achievers will perform averagely in the future, while the competence based achievers will perform according to their competence.