In the spirit of politics isn’t about policy, Shelby Steele: The original goal of affirmative action was to achieve two redemptions simultaneously. As society gave a preference to its former victims in employment and education, it hoped to redeem both those victims and itself. When America — the world’s oldest and most unequivocal democracy — finally acknowledged in the 1960s its heartless betrayal of democracy where blacks were concerned, the loss of moral authority was profound. …
I supported race-based affirmative action until I saw it operate in reality. The ideal of "leveling the playing field" sounds great, but as with most things, the ideal is far from the reality. All that I've seen affirmative action do is bring exceptionally incompetent, lazy, unqualified or minimally qualified individuals into organizations and promote them up the chain of command past truly hard working, intelligent, driven people. The result is devastating. Staff become demoralized and morale plummets. Productivity goes down, both because the non-recipients of affirmative action loose incentive and because the recipients themselves don't have incentive to excel. Also, lets face it, the average black iq is about 85. My experience working with large numbers of blacks bears this statistic out. So when we have to meet a quota, we are forced to draw from a pool of applicants with relatively lower cognitive ability than other other groups. In my experience,the main beneficiaries of race-based aa learn and function as a much slower pace than other groups. This means that it takes more resources, time, and ultimately more staff to get work done.What I want to know is how long will affirmative action be "necessary" for blacks. At what point can will they be required to perform equally and pull their own weight? 50 years from now will they still require this type of coddling in reparation for 100 year old wrongs done to their ancestors?
Tolerance of Affirmative Action is based on white guilt.
A fairer, faster and more effective system to level the playing field is to remove all race/gender/ethnic barriers and to make the barriers illegal.
If you want us all to live together, unite us, dont split us through AA.
If the only reason why previously disadvantaged individuals didnt succeed was due to oppression, remove all oppression and let them flourish naturally.
AA, is apartheid in reverse.
Nature is a perfect example of a working system. You dont introduce a new animal to kill a dominating species. It makes things worse. Instead, nature will balance it out by making the dominating species battle for resources and eventually die down.
If we're talking genetics, let's bring haplogroups and haplotypes into the discussion. No need for baskets like White ,Black, East Asian -even if they may be fairly accurate substitutes. Especially with "white" and "black" I'm feeling gamed into team socialization. Like I'm a pawn (okay I suppose we're at least knights -the pawns watch Fox News and BET) defending defending the white king or going after the black one.
Speaking of obvious things to study, I'd like to see some good game theory/evo psych/behavioral economic study identification and nomenclature that weirdly break out of the normal pageantry. Stuff that Steve Sailer, Anthony Cumia, Professor Gates, and Patrice O'Neal would blurt out at each other in an argument if they were all highly literate in current social science.
Africans are quite diverse, but are West Africans (much) more diverse than Europeans? (African Americans surely are, because of the white admixture.)
Black immigrants to the US are largely the descendants of slaves, just slaves in the Caribbean.
Black Americans average slightly shorter than whites, but have slightly larger standard deviation. Blacks in America and Japanese in Japan have lower standard deviation on IQ tests than whites. Height isn't really a bell curve at the extremes. IQ is a bell curve in the middle by definition, but it looks pretty reasonable. I'd be a lot more nervous extrapolating it, though.
Might well be. On the other hand, slavery may have exerted extreme selective pressure on American blacks. As long as we are speculating about selective pressure, we should consider all sources. It is consistent with the common observation that recent black immigrants do better than American blacks descended from slaves (though, obviously, there are alternative explanations, if it's even true).
Yes, but the further out you go on the tail on a bell-curve-like distribution, the greater the discrepancy in group representation produced by an apparently small difference in mean or even variance.
It occurred to me long after this was said that all evidence points to greater genetic diversity among, say, Africans than among European descendants. Significant fraction of would-be European diversity would be expected to be "chopped off" by their historical exposure to more epidemics, and developing light skin, both of which select completely tangential to intelligence. Thus if one has to guess, I think the best bet is that African descendants have greater variance. (It probably helps too that African American's have mixed ancestry.)
As Larry Summers pointed out, the effect of different variances overwhelms different means at the tails, which one might take to suggest that there should be a higher fraction of African descendants at the upper tail of ability.
I don't have a good model for understanding affirmative action in the US, or even race. But I haven't been impressed with any of the comprehensive theories I've been presented with yet. I think one of the best starting points with race is to acknowledge that it is extraordinarily robust as an ideology, a pageant, a mythology, an identity capture technology. It's an extraordinary social weapon for Goffman-smart people to use against Asperger-smart people. It's one of the few macrosocial frames that can compete effectively with the major religions as shaping mass behavior and attracting mass attention. And I think it may be the hardest to study from a "neutral" position -it seems to me that every observer becomes "raced" and gets subordinated to the grand pageantry. They all either become a Steve Sailor, or become skewerable by a Steve Sailor for their gamings and insere claims at objective participation in a social epistemology of racial identity as a natural phenomenon.
There has been and is on-going really great scientific and empirical inquiry into Christianity. When it comes to race, even our best, most pure empiricists seem to me to become immediately tainted. Perhaps we can at least start by acknowledging that it's one motherfucker of a bias-inducing macrosocial phenomenon.
DOJ, that implementation makes sense. But you have not addressed my second and final point, which is that "in principle it can be reasonable to assign a very very low P to some hypothesis". If the number of possibilities is high enough, then the average P assigned to one of those possibilities must be low enough so that the sum of probabilities does not exceed 1. And it is not that hard to define 10^100 possibilities. Uncertainties about our ability to reason, or about our fundamental assumptions, do not negate this point.
One response is that in the specific case at hand we weren't talking about 10^100 possibilities, but just two. But I could reply that if you randomly pick one possibility X out of these 10^100 possibilities you can define a set of two possibilities, X or not-X, and still in the average case P(X) is no larger than 10^-100.
Another response is to say that we ought to "privilege" possibilities that are simple to describe. The description of a typical element of a set of 10^100 possibilities is long (100 numerals in base 10), but a description such as "the sun will rise tomorrow" is short, so we ought to assign a somewhat higher probability to it. (This is reminiscent of Occam's razor.) My problem with this response is that this is a statement about the possibility itself and its expression in our language. It does not have any connection I can see with our own fallibility as rational thinkers. And it was our fallibility that was the basis of the argument that we ought never to assign a probability as low as 10^-100 to something like "the sun will rise tomorrow".
Constant, I think ZMD's argument would be that there's at least a 10^{-8} chance or so that you mess up in defining the mutually exclusive possibilities and don't realize it. Which seems right.
But you're right about it being mostly meta/"outside the model". It means you should listen to evidence that the mathematical model you're using is wrong (rather than assigning literal 100% probability to it being accurate), but you immediately run into the problem of, how the hell do you treat the possibility that Bayesian inference itself is unsound? Well, you can say all your numbers are conditional probabilities on that not being the case, and in most situations treat those conditional probabilities as if they didn't have a condition attached, but when a pathological situation comes up you can still deal with it.
My 10^{-100} bound is a conditional probability, and there are more conditions than just the validity of Bayesian inference. Some of those conditions, such as the validity of basic genetic and evolutionary theory, and the quality of the information I have access to, can be legitimately questioned at more like the 10^{-7} to even 10^{-3} level, so ZMD was right to object to taking my number literally. But without those conditions, I can't make any inferences at all. So, as a decision-making agent, I take those conditions for granted, even though as a philosopher I don't.
To my knowledge the largest and most far reaching affirmative action program to ever have taken place was the program in India to address the issue of the "untouchables" -- a group that had experienced centuries of broad based, institutionalized exclusion.
Set-asides of 50 to 90% for public sector jobs, housing and education were made. The result of this was that "the rising tide lifted all boats". Instead of loosing out to the recipients of set asides, middle class and lower middle class populations were the beneficiaries of a new class of CUSTOMERS! Those untouchables that were provided education and opportunity joined the mainstream economy.
As far as living in a equanimous merit based society goes... Colin Powell -- whose record of performance excellence over a period of decades speaks for itself -- was promoted to General in 1977 only after Secretary of the Army Clifford Alexander threatened that there would be no promotions to general for anyone if he were given yet another promotions lists without a single minority listed, with the excuse that no military officers of color met the competitive criteria.
I am surprised that neither of these instances are mentioned given the vociferous and beefy response to this posting.
ZMD - I think if you look at a typical teaching example, say a coin toss scenario of some sort, of how to do Bayesian inference, you'll likely see that iteration can easily take you to the extreme you are arguing against. Granted, we are imperfect, but the process if inference is what it is. You propose being uncertain about our rationality, but I don't see quite how to implement this in, say, a coin toss scenario, except as a "meta" process that acts on the strict Bayesian inference. You talk about uncertainty about our rationality, but that is "meta" rational in a way, and in particular if we are reasoning by Bayesian inference, this uncertainty must, to be consistent, manifest as uncertainty about the validity of Bayesian inference. Or some such thing.
Also it is easy to define a set of 10^100 possibilities, and the average probability assigned to each of these cannot exceed 10^-100, on pain of adding up to P>1. This is true even though we are imperfect. Thus in principle it can be reasonable to assign a very very low P to some hypothesis, despite our imperfection.
Constant---honestly, I was using an informal sanity check, but it seems legitimate: a human who knows that humans are bad at social science can only put so much confidence in her conclusions, because "Humans are bad at social science and have a tendency to place unjustified confidence on their favored conclusions" is part of your prior information. To overcome this burden, it would seem like you would need some specific evidence that you have superhuman rationality---although what form could that evidence possibly take? I don't know.
10^(-100) is really really really ridiculously small.
Granted, that seems right, but I do have a question. It doesn't seem at all difficult to get to that number starting from a much less ridiculously small number, by iteration of Bayesian reasoning. Can you mathematically characterize the factors in probabilistic reasoning that would prevent you from arriving at such a ridiculously small number? I'm hoping it's something other than a kind of non-mathematical sanity check that you apply to the numbers after your calculations have reached a really really ridiculously small number.
How often claims like that turned out to be false? If it happened even once in history of humanity (and we know it did, many times), then you should consider that you might be overconfident.
The way I see the evidence, I'm over 90% (but not 10^-100) confident that most of IQ differences between populations are of environmental origin. There could be some leftover genetic difference, and it might even be significant in a way, but environmental differences right now are huge, and changes to environment are proven to cause massive changes to IQ - Flynn effect alone caused about 10 point difference over span of 30 years without any genetic change, and environmental change which caused it wasn't even that big.
Perhaps after everything is equal between populations (or getting asymptotically there due to diminishing returns on environmental factors) some small difference will persist as you say. Or perhaps it won't. I would be extremely surprised if it wasn't much much smaller than the difference today.
If you read carefully, I'm not actually supporting a conception of justice per se, only pointing out that without a moral dimension, arguments about public policies are meaningless. So you are quite free to say with Huemer that morality (and by extension any politics based on morality) is nonsensical, and (as an aside) I might be inclined to agree in a large, philosophical sense. But then, as you make clear in your own response, that's not a statement about politics on its own terms; it's a statement about your philosophical view of morality.
So to reiterate once again the main thrust of my response to Vichy's dismissive posting: if at bottom you believe that a reasonable and meaningful argument about a political issue can't be had, why are you joining one here? To say "you don't deserve a living" is itself, for all its detached coyness, not an amoral statement.
I supported race-based affirmative action until I saw it operate in reality. The ideal of "leveling the playing field" sounds great, but as with most things, the ideal is far from the reality. All that I've seen affirmative action do is bring exceptionally incompetent, lazy, unqualified or minimally qualified individuals into organizations and promote them up the chain of command past truly hard working, intelligent, driven people. The result is devastating. Staff become demoralized and morale plummets. Productivity goes down, both because the non-recipients of affirmative action loose incentive and because the recipients themselves don't have incentive to excel. Also, lets face it, the average black iq is about 85. My experience working with large numbers of blacks bears this statistic out. So when we have to meet a quota, we are forced to draw from a pool of applicants with relatively lower cognitive ability than other other groups. In my experience,the main beneficiaries of race-based aa learn and function as a much slower pace than other groups. This means that it takes more resources, time, and ultimately more staff to get work done.What I want to know is how long will affirmative action be "necessary" for blacks. At what point can will they be required to perform equally and pull their own weight? 50 years from now will they still require this type of coddling in reparation for 100 year old wrongs done to their ancestors?
Tolerance of Affirmative Action is based on white guilt.
A fairer, faster and more effective system to level the playing field is to remove all race/gender/ethnic barriers and to make the barriers illegal.
If you want us all to live together, unite us, dont split us through AA.
If the only reason why previously disadvantaged individuals didnt succeed was due to oppression, remove all oppression and let them flourish naturally.
AA, is apartheid in reverse.
Nature is a perfect example of a working system. You dont introduce a new animal to kill a dominating species. It makes things worse. Instead, nature will balance it out by making the dominating species battle for resources and eventually die down.
If we're talking genetics, let's bring haplogroups and haplotypes into the discussion. No need for baskets like White ,Black, East Asian -even if they may be fairly accurate substitutes. Especially with "white" and "black" I'm feeling gamed into team socialization. Like I'm a pawn (okay I suppose we're at least knights -the pawns watch Fox News and BET) defending defending the white king or going after the black one.
Speaking of obvious things to study, I'd like to see some good game theory/evo psych/behavioral economic study identification and nomenclature that weirdly break out of the normal pageantry. Stuff that Steve Sailer, Anthony Cumia, Professor Gates, and Patrice O'Neal would blurt out at each other in an argument if they were all highly literate in current social science.
Africans are quite diverse, but are West Africans (much) more diverse than Europeans? (African Americans surely are, because of the white admixture.)
Black immigrants to the US are largely the descendants of slaves, just slaves in the Caribbean.
Black Americans average slightly shorter than whites, but have slightly larger standard deviation. Blacks in America and Japanese in Japan have lower standard deviation on IQ tests than whites. Height isn't really a bell curve at the extremes. IQ is a bell curve in the middle by definition, but it looks pretty reasonable. I'd be a lot more nervous extrapolating it, though.
Might well be. On the other hand, slavery may have exerted extreme selective pressure on American blacks. As long as we are speculating about selective pressure, we should consider all sources. It is consistent with the common observation that recent black immigrants do better than American blacks descended from slaves (though, obviously, there are alternative explanations, if it's even true).
Yes, but the further out you go on the tail on a bell-curve-like distribution, the greater the discrepancy in group representation produced by an apparently small difference in mean or even variance.
It occurred to me long after this was said that all evidence points to greater genetic diversity among, say, Africans than among European descendants. Significant fraction of would-be European diversity would be expected to be "chopped off" by their historical exposure to more epidemics, and developing light skin, both of which select completely tangential to intelligence. Thus if one has to guess, I think the best bet is that African descendants have greater variance. (It probably helps too that African American's have mixed ancestry.)
As Larry Summers pointed out, the effect of different variances overwhelms different means at the tails, which one might take to suggest that there should be a higher fraction of African descendants at the upper tail of ability.
There has been and is on-going really great scientific and empirical inquiry into Christianity
Could you give examples?
I don't have a good model for understanding affirmative action in the US, or even race. But I haven't been impressed with any of the comprehensive theories I've been presented with yet. I think one of the best starting points with race is to acknowledge that it is extraordinarily robust as an ideology, a pageant, a mythology, an identity capture technology. It's an extraordinary social weapon for Goffman-smart people to use against Asperger-smart people. It's one of the few macrosocial frames that can compete effectively with the major religions as shaping mass behavior and attracting mass attention. And I think it may be the hardest to study from a "neutral" position -it seems to me that every observer becomes "raced" and gets subordinated to the grand pageantry. They all either become a Steve Sailor, or become skewerable by a Steve Sailor for their gamings and insere claims at objective participation in a social epistemology of racial identity as a natural phenomenon.
There has been and is on-going really great scientific and empirical inquiry into Christianity. When it comes to race, even our best, most pure empiricists seem to me to become immediately tainted. Perhaps we can at least start by acknowledging that it's one motherfucker of a bias-inducing macrosocial phenomenon.
DOJ, that implementation makes sense. But you have not addressed my second and final point, which is that "in principle it can be reasonable to assign a very very low P to some hypothesis". If the number of possibilities is high enough, then the average P assigned to one of those possibilities must be low enough so that the sum of probabilities does not exceed 1. And it is not that hard to define 10^100 possibilities. Uncertainties about our ability to reason, or about our fundamental assumptions, do not negate this point.
One response is that in the specific case at hand we weren't talking about 10^100 possibilities, but just two. But I could reply that if you randomly pick one possibility X out of these 10^100 possibilities you can define a set of two possibilities, X or not-X, and still in the average case P(X) is no larger than 10^-100.
Another response is to say that we ought to "privilege" possibilities that are simple to describe. The description of a typical element of a set of 10^100 possibilities is long (100 numerals in base 10), but a description such as "the sun will rise tomorrow" is short, so we ought to assign a somewhat higher probability to it. (This is reminiscent of Occam's razor.) My problem with this response is that this is a statement about the possibility itself and its expression in our language. It does not have any connection I can see with our own fallibility as rational thinkers. And it was our fallibility that was the basis of the argument that we ought never to assign a probability as low as 10^-100 to something like "the sun will rise tomorrow".
Constant, I think ZMD's argument would be that there's at least a 10^{-8} chance or so that you mess up in defining the mutually exclusive possibilities and don't realize it. Which seems right.
But you're right about it being mostly meta/"outside the model". It means you should listen to evidence that the mathematical model you're using is wrong (rather than assigning literal 100% probability to it being accurate), but you immediately run into the problem of, how the hell do you treat the possibility that Bayesian inference itself is unsound? Well, you can say all your numbers are conditional probabilities on that not being the case, and in most situations treat those conditional probabilities as if they didn't have a condition attached, but when a pathological situation comes up you can still deal with it.
My 10^{-100} bound is a conditional probability, and there are more conditions than just the validity of Bayesian inference. Some of those conditions, such as the validity of basic genetic and evolutionary theory, and the quality of the information I have access to, can be legitimately questioned at more like the 10^{-7} to even 10^{-3} level, so ZMD was right to object to taking my number literally. But without those conditions, I can't make any inferences at all. So, as a decision-making agent, I take those conditions for granted, even though as a philosopher I don't.
To my knowledge the largest and most far reaching affirmative action program to ever have taken place was the program in India to address the issue of the "untouchables" -- a group that had experienced centuries of broad based, institutionalized exclusion.
Set-asides of 50 to 90% for public sector jobs, housing and education were made. The result of this was that "the rising tide lifted all boats". Instead of loosing out to the recipients of set asides, middle class and lower middle class populations were the beneficiaries of a new class of CUSTOMERS! Those untouchables that were provided education and opportunity joined the mainstream economy.
As far as living in a equanimous merit based society goes... Colin Powell -- whose record of performance excellence over a period of decades speaks for itself -- was promoted to General in 1977 only after Secretary of the Army Clifford Alexander threatened that there would be no promotions to general for anyone if he were given yet another promotions lists without a single minority listed, with the excuse that no military officers of color met the competitive criteria.
I am surprised that neither of these instances are mentioned given the vociferous and beefy response to this posting.
ZMD - I think if you look at a typical teaching example, say a coin toss scenario of some sort, of how to do Bayesian inference, you'll likely see that iteration can easily take you to the extreme you are arguing against. Granted, we are imperfect, but the process if inference is what it is. You propose being uncertain about our rationality, but I don't see quite how to implement this in, say, a coin toss scenario, except as a "meta" process that acts on the strict Bayesian inference. You talk about uncertainty about our rationality, but that is "meta" rational in a way, and in particular if we are reasoning by Bayesian inference, this uncertainty must, to be consistent, manifest as uncertainty about the validity of Bayesian inference. Or some such thing.
Also it is easy to define a set of 10^100 possibilities, and the average probability assigned to each of these cannot exceed 10^-100, on pain of adding up to P>1. This is true even though we are imperfect. Thus in principle it can be reasonable to assign a very very low P to some hypothesis, despite our imperfection.
Constant---honestly, I was using an informal sanity check, but it seems legitimate: a human who knows that humans are bad at social science can only put so much confidence in her conclusions, because "Humans are bad at social science and have a tendency to place unjustified confidence on their favored conclusions" is part of your prior information. To overcome this burden, it would seem like you would need some specific evidence that you have superhuman rationality---although what form could that evidence possibly take? I don't know.
10^(-100) is really really really ridiculously small.
Granted, that seems right, but I do have a question. It doesn't seem at all difficult to get to that number starting from a much less ridiculously small number, by iteration of Bayesian reasoning. Can you mathematically characterize the factors in probabilistic reasoning that would prevent you from arriving at such a ridiculously small number? I'm hoping it's something other than a kind of non-mathematical sanity check that you apply to the numbers after your calculations have reached a really really ridiculously small number.
"10^{-100}"
How often claims like that turned out to be false? If it happened even once in history of humanity (and we know it did, many times), then you should consider that you might be overconfident.
The way I see the evidence, I'm over 90% (but not 10^-100) confident that most of IQ differences between populations are of environmental origin. There could be some leftover genetic difference, and it might even be significant in a way, but environmental differences right now are huge, and changes to environment are proven to cause massive changes to IQ - Flynn effect alone caused about 10 point difference over span of 30 years without any genetic change, and environmental change which caused it wasn't even that big.
Perhaps after everything is equal between populations (or getting asymptotically there due to diminishing returns on environmental factors) some small difference will persist as you say. Or perhaps it won't. I would be extremely surprised if it wasn't much much smaller than the difference today.
If you read carefully, I'm not actually supporting a conception of justice per se, only pointing out that without a moral dimension, arguments about public policies are meaningless. So you are quite free to say with Huemer that morality (and by extension any politics based on morality) is nonsensical, and (as an aside) I might be inclined to agree in a large, philosophical sense. But then, as you make clear in your own response, that's not a statement about politics on its own terms; it's a statement about your philosophical view of morality.
So to reiterate once again the main thrust of my response to Vichy's dismissive posting: if at bottom you believe that a reasonable and meaningful argument about a political issue can't be had, why are you joining one here? To say "you don't deserve a living" is itself, for all its detached coyness, not an amoral statement.