Inequality and the Power Law

[Togo]

How so? The paper claims to be measuring performance, not ability. Performance following a Paretian distribution says something about how society is organised, not about how people act. For example, take academic paper publication. If you identify a particularly prolific author, and shoot him, would there be fewer publications next year, or the same? Unless you think that some publications would suddenly fold without their favourite contributor, it seems obvious that the same number would be published. Which means that what the distribution is showing is not that publishable ideas are concentrated in a few people, but rather that the system is geared up to reward the already sucessful with greater recognition.

Your conclusion does not follow.

Lets put it in a different realm where it's clearer. Al-Qaeda hits the All-Star game, a bunch of top players are killed.

Does that magically make the remainder move up in ability? Or does it mean we will see less good play next year?

That demonstrates the point I'm making. Ask yourself this. Will the people who move up magically become as good as the previous lot by the end of the season? If not, then they're showing the same performance indicators (games played) as the previous bunch, but clearly don't have the same ability.

It's not hard to spot the gap here. You have 100 players, and take the top 20 to play in the NFL. The gap between number 20 and number 21 might be large, or it might be small. But the number of NFL games they play will not magically depend on their skills, only their position relative to each other.

The performance indicator being used appears to be an ordinal indicator being used as an interval measure. The number of games you play, number of papers you publish, etc. Does not depend on your skill, but rather your position in the hierarchy. Your position in the hierarchy may depend on relative skill (or it might not), but because the hierarchy not the skill drives the performance indicators being measured, the performance indicators being measures can not possibly tell you anything about the distribution of skills.

 

[doubtingt]

Your conclusion does not follow.

Lets put it in a different realm where it's clearer. Al-Qaeda hits the All-Star game, a bunch of top players are killed.

Does that magically make the remainder move up in ability? Or does it mean we will see less good play next year?

That demonstrates the point I'm making. Ask yourself this. Will the people who move up magically become as good as the previous lot by the end of the season? If not, then they're showing the same performance indicators (games played) as the previous bunch, but clearly don't have the same ability.

It's not hard to spot the gap here. You have 100 players, and take the top 20 to play in the NFL. The gap between number 20 and number 21 might be large, or it might be small. But the number of NFL games they play will not magically depend on their skills, only their position relative to each other.

The performance indicator being used appears to be an ordinal indicator being used as an interval measure. The number of games you play, number of papers you publish, etc. Does not depend on your skill, but rather your position in the hierarchy. Your position in the hierarchy may depend on relative skill (or it might not), but because the hierarchy not the skill drives the performance indicators being measured, the performance indicators being measures can not possibly tell you anything about the distribution of skills.

Your conclusion does not follow.

Lets put it in a different realm where it's clearer. Al-Qaeda hits the All-Star game, a bunch of top players are killed.

Does that magically make the remainder move up in ability? Or does it mean we will see less good play next year?

That demonstrates the point I'm making. Ask yourself this. Will the people who move up magically become as good as the previous lot by the end of the season? If not, then they're showing the same performance indicators (games played) as the previous bunch, but clearly don't have the same ability.

It's not hard to spot the gap here. You have 100 players, and take the top 20 to play in the NFL. The gap between number 20 and number 21 might be large, or it might be small. But the number of NFL games they play will not magically depend on their skills, only their position relative to each other.

The performance indicator being used appears to be an ordinal indicator being used as an interval measure. The number of games you play, number of papers you publish, etc. Does not depend on your skill, but rather your position in the hierarchy. Your position in the hierarchy may depend on relative skill (or it might not), but because the hierarchy not the skill drives the performance indicators being measured, the performance indicators being measures can not possibly tell you anything about the distribution of skills.

Great point. Many of their performance measures are not reliable or valid measures of ability, skill, or effort, but rather measures of opportunities given or rewards given, which are often based upon status and those opportunities and rewards generate more status, making status breed status and past rewards increase future rewards independent of whether current ability and efforts warrants more rewards. This is quite glaringly true in academics for both things like number of publications and grant $. This is true for multiple reasons. Institutional prestige imparts automatic status that makes grant getting and pubs easier, regardless of intellectual merit of the work. In turn, getting one publication or a grant has a direct impact upon your getting a future publication or grants, regardless of the objective quality of your future work. Contrary to myth, blind review does not exist. Most editors and top reviewers know enough about an area to be able to infer the likely authors based upon topic, perspective, writing style, and citations within the document (authors usually refer to their own past work far more than would be expected based on merit and relevance of past work). Reviews and editors give people who’ve already published similar work more benefit of the doubt and thus require lower standards for publication. An added effect of this is that people with publications in the area are more likely to be asked to be a collaborator or consultant on someone else's paper and thus get easy authorship, often without having to make any quality contribution or any contribution at all, other than lending their name to the paper to boost chances of publication. All of this results in a multiplicative effect in which no objective difference in ability or effort can still mean massive differences in publications, and/or relatively small differences in ability or effort wind up resulting in massive differences in publications.
All of this speaks to the problem that many of their “performance” measures are not valid indices of objective quality or quantity of performance, but rather are measures of how different people’s performance is subjectively valued (such as which of 2 researcher papers of equal quality gets published). I’m not saying that peer review is useless, just that the amount of publications you actually get (and relatedly the amount of grant $ you get) has only a weak and non-linear relation to the quality or your work or the amount of work you do. The same is true in athletics and entertainment.
My prior point is separate from this, namely that even for the measures that might be valid measures of differences in quality or quantity of work, using only the people who have been pre-selected for being in the top 5-20% of the population in ability and willingness to do the task will guarantee a non bell curve distribution in performance similar to the one they find. The majority of people in such samples will be just above the minimal cut-off point below which you don’t even get a chance to be in that profession.

 

[Loren Pechtel]

Your conclusion does not follow.

Lets put it in a different realm where it's clearer. Al-Qaeda hits the All-Star game, a bunch of top players are killed.

Does that magically make the remainder move up in ability? Or does it mean we will see less good play next year?

That demonstrates the point I'm making. Ask yourself this. Will the people who move up magically become as good as the previous lot by the end of the season? If not, then they're showing the same performance indicators (games played) as the previous bunch, but clearly don't have the same ability.

It's not hard to spot the gap here. You have 100 players, and take the top 20 to play in the NFL. The gap between number 20 and number 21 might be large, or it might be small. But the number of NFL games they play will not magically depend on their skills, only their position relative to each other.

The performance indicator being used appears to be an ordinal indicator being used as an interval measure. The number of games you play, number of papers you publish, etc. Does not depend on your skill, but rather your position in the hierarchy. Your position in the hierarchy may depend on relative skill (or it might not), but because the hierarchy not the skill drives the performance indicators being measured, the performance indicators being measures can not possibly tell you anything about the distribution of skills.

The number of games played won't change. The quality of the play will change.

Just like getting rid of the top people won't change the number of businesses, just how well they are run.

 

[Togo]

That demonstrates the point I'm making. Ask yourself this. Will the people who move up magically become as good as the previous lot by the end of the season? If not, then they're showing the same performance indicators (games played) as the previous bunch, but clearly don't have the same ability.

It's not hard to spot the gap here. You have 100 players, and take the top 20 to play in the NFL. The gap between number 20 and number 21 might be large, or it might be small. But the number of NFL games they play will not magically depend on their skills, only their position relative to each other.

The performance indicator being used appears to be an ordinal indicator being used as an interval measure. The number of games you play, number of papers you publish, etc. Does not depend on your skill, but rather your position in the hierarchy. Your position in the hierarchy may depend on relative skill (or it might not), but because the hierarchy not the skill drives the performance indicators being measured, the performance indicators being measures can not possibly tell you anything about the distribution of skills.

The number of games played won't change. The quality of the play will change.

Just like getting rid of the top people won't change the number of businesses, just how well they are run.

If they've developed an interval-level measure of objective quality for playing football, writing an academic paper, or running a business, then they're well on the way to a Nobel price. They won't have done, though. They'll be measuring games played, papers published, and other crude proxies.

It's a shame we can't see the paper itself, rather than relying on a journalist's account of it.