Using the term over-simplified was my attempt at generosity. As presently stated, your claim is entirely wrong. Intelligence is the single best predictor of job performance for all but the most narrowly-focused manual tasks, see for example Ree & Earles, Current Directions in Psychological Science vol. 1, No. 3 (Jun., 1992), pp. 86-89.
The strong claim you made in your original comment was entirely false, and I get the impression you were just speculating wildly about something you don’t actually know much about.
After intelligence, Conscientiousness is probably the single best predictor of job success since it predicts even after controlling for IQ, education level, etc. (Cribbing from my usual footnote, the best starting point is the meta-analysis http://people.tamu.edu/~mbarrick/Pubs/1991_Barrick_Mount.pdf )
It’s important to note that employers are not seeking to maximize employee performance. They’re seeking to maximize the difference between the value provided by the employee and the wage provided to the employee.
As doubly pointed out, the system is unlikely to be sane. In an insane system, you cannot predict that most employers will even know that intelligence is the best predictor of performance, let alone that they will effectively apply the best available method to select candidates by this criterion.
The fact is, from personal observation (which I admit is anecdotal evidence from a tiny, biased sample size), employers generally do not care to effectively figure this out. All employers I’ve encountered have had an attitude of wanting everything to “just work” (through the magic of being awesome, presumably) and land them the best employees because they will it to be so. If this would expand to the population in a proportional manner, it would mean that the vast majority of “employers” are either simply acting irrationally for this situation (AKA not only is the system insane, but nearly all its players are, too) or do not assign sufficient utility to obtaining better employees for it to be worth the perceived cost of finding them.
I believe this was the main point being made. It’s not being argued that intelligence makes you a better actual performer, what is being argued is that employers do not effectively pick the most intelligent candidates, or worse, that they are not even remotely aware of what they should select for, and that they believe it is relatively worthless for them to attempt to find out more on this subject than they already know.
Most employers want a track of record of doing job X successfully when hiring people to do job X. If job X requires intelligence, then they will be indirectly selecting intelligent people … whilst filtering out “smart but doesn’t get things done” people. Seems sane to me.
Yes, of course. These particular traits you have deigned to consider for your worthy evaluation do seem, to me as well, perfectly sane.
I think you forgot to activate your Real World Logic coprocessor before replying, and I’m being sarcastic and offensive in this response.
In more serious words, these particular selected characteristics do not comprise the entirety of “the system” aforementioned. I’ve said that the system is /unlikely/ to be sane, as I do not have complete information on the entire logic and processes in it. I also think we’re working off of different definitions of “sane”—here, IIRC, I was using a technical version that could be better expressed as “close to perfectly rational, in the same way perfect logicians can be in theoretical formal logic puzzles”.
That leads to a much-noted chicken-and-egg problem… but that aside, for all but the most menial and interchangeable X, employers don’t generally have access to data about how well and how long prospective hires have done X. They have access to candidates’ word for how well they’ve done more or less imperfectly related work, and usually to recommendations from their former employers and coworkers—but the former is unreliable, and the latter demonstrates only that the candidate isn’t a complete schlub.
I haven’t read the paper in the ancestor, but it seems reasonable to me that IQ would often end up being a better predictor of performance, given these constraints.
No. But it is evidence for the other thing being better, when the constraints under question don’t apply to that other thing.
Of course, while we’re talking evidence, we shouldn’t neglect the fact that the traditional interview/resume method has reached fixation and doesn’t look to be in immediate danger of being displaced. But “current practice” doesn’t necessarily imply “optimal” or even “best known”, especially when psychometric methods are legally problematic.
The fact is, from personal observation (which I admit is anecdotal evidence from a tiny, biased sample size), employers generally do not care to effectively figure this out.
They don’t have to, they just have to observe what other successful employers are doing and copy that, the ones who copy the correct features will themselves be more successful, a.k.a., memetic evolution works.
Doubly pointed out! Doubly, no evidence needed, just pointing it out.
You have not cited evidence that it is insane. We now have you citing Grognor, who claims Wedrifid thinks it is insane. So the font of pure epistemic truth is a throwaway line by noted anonymous internet commenter wedrifid. I’ll be sure to let all the academics who have been scientifically studying employee selection for decades know about your proclamation.
Better shut down the journals, ladies and gentlemen. The internet has proclaimed you insane. They figured it out from the comfort of their homes, using the power of pure reason and vague memories of reading Dilbert cartoons.
This comment probably got downvoted for unnecessary ranting and sarcasm, even though the point it makes is a valid one: a published study has more credibility than a speculation on a forum.
The point it makes is valid. It is, however, irrelevant and strawman to the point that I made to which it apparently attempts to respond: The science might be right, but the implemented system is potentially-insane, as shown by anecdotal evidence, because it seems to us that the agents of the system do not apply the science and knowledge in question whether it’s correct or not.
In fact, the agents appear to completely ignore the system’s potential rules and instead rely on the Universal Theory of Magic. Whether the study is correct or not, whether it is informative or not, whether it is biased or not, whether it is useful or not… is all completely ignored by the agents of the system, by my observations.
Even “anecdotal” evidence is sufficient for a posterior to become P(There Cannot Ever Be A Case Where X | Anecdote of X) < 1.
If you read carefully, you’ll notice that I wrote potentially-insane. Not “It Must Be Insane Because [Insert Anecdote X]”.
If I’m wrong, please point to me where I’ve misinterpreted / misread Bayes’ Theorem on this. I’d really like to get rid of this irrational manner of thought ASAP.
For the sake of Nitpick, I’ll first argue that I neither ever read Dilbert cartoons nor ever used such a thing as “pure reason” to my knowledge nor wrote any of this from within or benefiting-of-the-comfort-of my own home. However, since I’m just pointing this out, you should, if you persist in this course of argumentation, completely discard what I just said and assume that I did somehow.
The reason I did not present factual evidence, for my part, is that I considered it unnecessary on the prior that it be unlikely that someone who has read Eliezer’s Core sequences (and reflected while doing so) would disagree, if only upon the notion that any prior in favor of “a lot of smart people have thought of this before us and yet we’re still using it so it must be right” has already been shown in said sequences to be biased.
Notice that you’ve also completely ignored my main point and built a massive, chain-woven strawman painted black standing in the middle of the highway. The primary argument of my comment is that it is not the science which is entirely wrong, but the way the elements of the system fail to even acknowledge that there is something better they could be doing to select employees. Namely, employers being stupid. I back this up very weakly with anecdotal, statistically-insignificant and underpowered “evidence”. Is there any more convenient a world you would wish for?
And here I was hoping that someone would rebuild my argument in stronger form before giving me reason to reconsider by showing that stronger argument wrong.
Right now, I have weak evidence (apparent lack of rational decision-making regarding employee selection on the part of employers) that the system is insane, yet strong evidence that it is at least not entirely sane in all situations. Conversely, there is no evidence suggesting to me that the system is “Sane”, and every other variable that I suspect is correlated to this system’s sanity shows indirect evidence towards insanity (examples in politics and religion come to mind most immediately, followed by various forms of warfare, systemic abuse and wilful neglect).
What’s more, the system being sane is, in my opinion, only trivially relevant if it remains inefficient and sub-optimal due to lack of awareness of key variables that are, in hindsight, absolutely crucial and would be the first thing I go for. Naturally, the cost of learning this missing data is unknown at present, its deviation range being too large for me to even make a good educated guess.
Using the term over-simplified was my attempt at generosity. As presently stated, your claim is entirely wrong. Intelligence is the single best predictor of job performance for all but the most narrowly-focused manual tasks, see for example Ree & Earles, Current Directions in Psychological Science vol. 1, No. 3 (Jun., 1992), pp. 86-89.
The strong claim you made in your original comment was entirely false, and I get the impression you were just speculating wildly about something you don’t actually know much about.
After intelligence, Conscientiousness is probably the single best predictor of job success since it predicts even after controlling for IQ, education level, etc. (Cribbing from my usual footnote, the best starting point is the meta-analysis http://people.tamu.edu/~mbarrick/Pubs/1991_Barrick_Mount.pdf )
It’s important to note that employers are not seeking to maximize employee performance. They’re seeking to maximize the difference between the value provided by the employee and the wage provided to the employee.
As doubly pointed out, the system is unlikely to be sane. In an insane system, you cannot predict that most employers will even know that intelligence is the best predictor of performance, let alone that they will effectively apply the best available method to select candidates by this criterion.
The fact is, from personal observation (which I admit is anecdotal evidence from a tiny, biased sample size), employers generally do not care to effectively figure this out. All employers I’ve encountered have had an attitude of wanting everything to “just work” (through the magic of being awesome, presumably) and land them the best employees because they will it to be so. If this would expand to the population in a proportional manner, it would mean that the vast majority of “employers” are either simply acting irrationally for this situation (AKA not only is the system insane, but nearly all its players are, too) or do not assign sufficient utility to obtaining better employees for it to be worth the perceived cost of finding them.
I believe this was the main point being made. It’s not being argued that intelligence makes you a better actual performer, what is being argued is that employers do not effectively pick the most intelligent candidates, or worse, that they are not even remotely aware of what they should select for, and that they believe it is relatively worthless for them to attempt to find out more on this subject than they already know.
Most employers want a track of record of doing job X successfully when hiring people to do job X. If job X requires intelligence, then they will be indirectly selecting intelligent people … whilst filtering out “smart but doesn’t get things done” people. Seems sane to me.
Yes, of course. These particular traits you have deigned to consider for your worthy evaluation do seem, to me as well, perfectly sane.
I think you forgot to activate your Real World Logic coprocessor before replying, and I’m being sarcastic and offensive in this response.
In more serious words, these particular selected characteristics do not comprise the entirety of “the system” aforementioned. I’ve said that the system is /unlikely/ to be sane, as I do not have complete information on the entire logic and processes in it. I also think we’re working off of different definitions of “sane”—here, IIRC, I was using a technical version that could be better expressed as “close to perfectly rational, in the same way perfect logicians can be in theoretical formal logic puzzles”.
Insane is not an obvious synonym for imperfect.
Opinions vary on the role of intelligence in the first place
That leads to a much-noted chicken-and-egg problem… but that aside, for all but the most menial and interchangeable X, employers don’t generally have access to data about how well and how long prospective hires have done X. They have access to candidates’ word for how well they’ve done more or less imperfectly related work, and usually to recommendations from their former employers and coworkers—but the former is unreliable, and the latter demonstrates only that the candidate isn’t a complete schlub.
I haven’t read the paper in the ancestor, but it seems reasonable to me that IQ would often end up being a better predictor of performance, given these constraints.
One thing being imperfect doesn’t make another thing better.
No. But it is evidence for the other thing being better, when the constraints under question don’t apply to that other thing.
Of course, while we’re talking evidence, we shouldn’t neglect the fact that the traditional interview/resume method has reached fixation and doesn’t look to be in immediate danger of being displaced. But “current practice” doesn’t necessarily imply “optimal” or even “best known”, especially when psychometric methods are legally problematic.
They don’t have to, they just have to observe what other successful employers are doing and copy that, the ones who copy the correct features will themselves be more successful, a.k.a., memetic evolution works.
Doubly pointed out! Doubly, no evidence needed, just pointing it out.
You have not cited evidence that it is insane. We now have you citing Grognor, who claims Wedrifid thinks it is insane. So the font of pure epistemic truth is a throwaway line by noted anonymous internet commenter wedrifid. I’ll be sure to let all the academics who have been scientifically studying employee selection for decades know about your proclamation.
Better shut down the journals, ladies and gentlemen. The internet has proclaimed you insane. They figured it out from the comfort of their homes, using the power of pure reason and vague memories of reading Dilbert cartoons.
This comment probably got downvoted for unnecessary ranting and sarcasm, even though the point it makes is a valid one: a published study has more credibility than a speculation on a forum.
The point it makes is valid. It is, however, irrelevant and strawman to the point that I made to which it apparently attempts to respond: The science might be right, but the implemented system is potentially-insane, as shown by anecdotal evidence, because it seems to us that the agents of the system do not apply the science and knowledge in question whether it’s correct or not.
In fact, the agents appear to completely ignore the system’s potential rules and instead rely on the Universal Theory of Magic. Whether the study is correct or not, whether it is informative or not, whether it is biased or not, whether it is useful or not… is all completely ignored by the agents of the system, by my observations.
...
Even “anecdotal” evidence is sufficient for a posterior to become P(There Cannot Ever Be A Case Where X | Anecdote of X) < 1.
If you read carefully, you’ll notice that I wrote potentially-insane. Not “It Must Be Insane Because [Insert Anecdote X]”.
If I’m wrong, please point to me where I’ve misinterpreted / misread Bayes’ Theorem on this. I’d really like to get rid of this irrational manner of thought ASAP.
For the sake of Nitpick, I’ll first argue that I neither ever read Dilbert cartoons nor ever used such a thing as “pure reason” to my knowledge nor wrote any of this from within or benefiting-of-the-comfort-of my own home. However, since I’m just pointing this out, you should, if you persist in this course of argumentation, completely discard what I just said and assume that I did somehow.
The reason I did not present factual evidence, for my part, is that I considered it unnecessary on the prior that it be unlikely that someone who has read Eliezer’s Core sequences (and reflected while doing so) would disagree, if only upon the notion that any prior in favor of “a lot of smart people have thought of this before us and yet we’re still using it so it must be right” has already been shown in said sequences to be biased.
Notice that you’ve also completely ignored my main point and built a massive, chain-woven strawman painted black standing in the middle of the highway. The primary argument of my comment is that it is not the science which is entirely wrong, but the way the elements of the system fail to even acknowledge that there is something better they could be doing to select employees. Namely, employers being stupid. I back this up very weakly with anecdotal, statistically-insignificant and underpowered “evidence”. Is there any more convenient a world you would wish for?
And here I was hoping that someone would rebuild my argument in stronger form before giving me reason to reconsider by showing that stronger argument wrong.
Right now, I have weak evidence (apparent lack of rational decision-making regarding employee selection on the part of employers) that the system is insane, yet strong evidence that it is at least not entirely sane in all situations. Conversely, there is no evidence suggesting to me that the system is “Sane”, and every other variable that I suspect is correlated to this system’s sanity shows indirect evidence towards insanity (examples in politics and religion come to mind most immediately, followed by various forms of warfare, systemic abuse and wilful neglect).
What’s more, the system being sane is, in my opinion, only trivially relevant if it remains inefficient and sub-optimal due to lack of awareness of key variables that are, in hindsight, absolutely crucial and would be the first thing I go for. Naturally, the cost of learning this missing data is unknown at present, its deviation range being too large for me to even make a good educated guess.