3 ms·

Is the FTC accusing of online ad equivalent of redlining? in part, these sorts of ads are determined by a designed bid/match thresh-hold - so why is it not Ome

Is the FTC accusing of online ad equivalent of redlining? in part, these sorts of ads are determined by a designed bid/match thresh-hold - so why is it not Omega Psi Phi accepting better matches on the Yield side? (also remember, if they deny the ad placement, they deny themselves ad revenue) Also this is going to get a fundamental oddball question - being part of a protected group could be correlated but not causative for the targeting on that site, but not other sites. For example, for the lower quality credit cards, it could be tied to credit scores, and and there could be a higher percentage of people showing up in these protected groups on these sites (but not in other sites that these protected groups go to that don't have have this correlation of low credit scores, which is the actual targeting criteria.) Targeting, weirdness.

mdasen 12y ago
You've brought up an interesting point: correlation vs causation. One of the issues in discrimination contexts today is disparate impact. For example, we know that a bank couldn't say, "we don't want to lend money to black people." Race is a protected class and that's discrimination against a protected class. What if the bank said, "we don't want to lend money to people in cities"? Now we're into disparate impact land. Racial minorities are often more in cities and this policy could have an adverse impact against certain groups along racial lines. The kind of questions that are then often asked are things like, "is there a business purpose?" and "is there no tighter indicator that the bank could use for their business purpose that wouldn't have the same disparate impact?" Ultimately, disparate impact is about not using indicators that have a racial bias when better indicators exist. The bank might say, "we use city residence because people in cities usually have a history of missed debt payments." Regulators could retort, "then why not just use an individuals history of made and missed debt payments?" Disparate impact isn't about preventing the use of all indicators that correlate with a protected class. However, companies often use indicators that are of dubious predictive power (things where one can't show a significant correlation with the bad outcome the company wants to avoid, but show a strong correlation with a protected class). Since the article is about analytics, one can imagine building a bayesian model and a few test subjects to put through it. We could think about features that are probably predictive like missed payments. We could also have features like height which should be nonsense. For the sake of this silly model, let's say that one racial group is shorter and is a greater credit risk. You create your "height < 6ft" feature and your "has > 2 missed payments" bernoulli variables and start training your model based on your training set. You find that "has > 2 missed payments" is very predictive, but you also find that "height < 6ft" is also predictive. But there's a certain, "um, we know that the feature is correlated with a protected class and there's no good, logical reason to use it unless you're trying to create a proxy for that protected class so you can discriminate" that might go through one's head.
- yummyfajitas 12y ago
  Disparate impact isn't about preventing the use of all indicators that correlate with a protected class. However, companies often use indicators that are of dubious predictive power (things where one can't show a significant correlation with the bad outcome the company wants to avoid, but show a strong correlation with a protected class). This is done in a LOT of machine learning settings - the vast majority of variables used in classification/regression scenarios fall into this category. Most machine learning consists of turning a large number of nearly insignificant correlations into a single big correlation. Consider category A, which has features x[i] located in the ball of radius 100 around the point [0,0,...,0] in N dimensional space. Category B has features x[i] located in the ball of radius 100 around the point [1,1,...,1]. Any individual feature is useless - if it's from category A, it'll be located in [-100,100], whereas if it's from category B it'll be located in [-99,101]. On the other hand, the classifier sum(x) works perfectly for N > 100. That's because in N dimensions, the distance between [0,0,...,0] and [1,1,...,1] is sqrt(N) (proof: use the pythagorean theorem). That's an oversimplified example, but I hope it illustrates the point. It's a bit scary that the FTC wants to start regulating feature engineering. Will they also take a position on Bayesian classifiers vs geometric/vector space methods?
- shanacarp 12y ago
  That's part of a larger set of questions about protected classes and targeting. We don't know until we see the full campaign and targeting criteria. RTB makes this even harder, as well as Yield management issues. This is but a dance. If we don't know all the criteria of who is asking for what at a given time, no dice on if the FTC is right

mdasen 12y ago

You've brought up an interesting point: correlation vs causation. One of the issues in discrimination contexts today is disparate impact. For example, we know that a bank couldn't say, "we don't want to lend money to black people." Race is a protected class and that's discrimination against a protected class. What if the bank said, "we don't want to lend money to people in cities"? Now we're into disparate impact land. Racial minorities are often more in cities and this policy could have an adverse impact against certain groups along racial lines. The kind of questions that are then often asked are things like, "is there a business purpose?" and "is there no tighter indicator that the bank could use for their business purpose that wouldn't have the same disparate impact?" Ultimately, disparate impact is about not using indicators that have a racial bias when better indicators exist. The bank might say, "we use city residence because people in cities usually have a history of missed debt payments." Regulators could retort, "then why not just use an individuals history of made and missed debt payments?" Disparate impact isn't about preventing the use of all indicators that correlate with a protected class. However, companies often use indicators that are of dubious predictive power (things where one can't show a significant correlation with the bad outcome the company wants to avoid, but show a strong correlation with a protected class). Since the article is about analytics, one can imagine building a bayesian model and a few test subjects to put through it. We could think about features that are probably predictive like missed payments. We could also have features like height which should be nonsense. For the sake of this silly model, let's say that one racial group is shorter and is a greater credit risk. You create your "height < 6ft" feature and your "has > 2 missed payments" bernoulli variables and start training your model based on your training set. You find that "has > 2 missed payments" is very predictive, but you also find that "height < 6ft" is also predictive. But there's a certain, "um, we know that the feature is correlated with a protected class and there's no good, logical reason to use it unless you're trying to create a proxy for that protected class so you can discriminate" that might go through one's head.

yummyfajitas 12y ago
Disparate impact isn't about preventing the use of all indicators that correlate with a protected class. However, companies often use indicators that are of dubious predictive power (things where one can't show a significant correlation with the bad outcome the company wants to avoid, but show a strong correlation with a protected class). This is done in a LOT of machine learning settings - the vast majority of variables used in classification/regression scenarios fall into this category. Most machine learning consists of turning a large number of nearly insignificant correlations into a single big correlation. Consider category A, which has features x[i] located in the ball of radius 100 around the point [0,0,...,0] in N dimensional space. Category B has features x[i] located in the ball of radius 100 around the point [1,1,...,1]. Any individual feature is useless - if it's from category A, it'll be located in [-100,100], whereas if it's from category B it'll be located in [-99,101]. On the other hand, the classifier sum(x) works perfectly for N > 100. That's because in N dimensions, the distance between [0,0,...,0] and [1,1,...,1] is sqrt(N) (proof: use the pythagorean theorem). That's an oversimplified example, but I hope it illustrates the point. It's a bit scary that the FTC wants to start regulating feature engineering. Will they also take a position on Bayesian classifiers vs geometric/vector space methods?
shanacarp 12y ago
That's part of a larger set of questions about protected classes and targeting. We don't know until we see the full campaign and targeting criteria. RTB makes this even harder, as well as Yield management issues. This is but a dance. If we don't know all the criteria of who is asking for what at a given time, no dice on if the FTC is right