This is actually very similar to an intuition I’ve had about this problem. The difference is that I compare it to a different scenario, and regard it as a reason, not to reject the trolley problem, but a reason to justify the optimality of not pushing the innocent bystander onto the tracks.
You compared it to wealth-redistribution attempting to optimize total utility, while I think a better comparison is discrimination law, something that matters close to me. (Edit: sorry for awkward phrasing, keeping it there because it was quoted.) In short, just as milking the cow makes it waste effort avoiding being milked, discrimination law simply makes employers drastically revise hiring practices so that they don’t use the one that counts under the law as a “job posting”.
So they rely more heavily on network hires and indirect signals of employee quality, and, ironically, make it much harder for people outside these networks—usually the “protected” classes—to even get a chance in the first place, no matter how good they could demonstrate their work to be!
People think they can just shove these “surprise” costs onto agents with “too much utility”, ignoring the long-term ways they’ll react. As I said before about the problem, one non-obvious result of shoving the guy onto the track is that drastically shifts around the risk profiles of various activities. Previously, standing on a track was riskier than staying off it. But a policy of switching this risk around makes it harder overall for people to ascertain risks, which makes them spend wasteful effort avoiding risks (since they have to assume they’re understated).
Generally, it really bothers me when people talk in terms of policies that attempt to shift utility from privileged group A to pitiable group B, not realizing that A will still gets its privileges and exclude B—it’ll just become a more complex dance for B to figure out.
I’ve been rereading this comment for the past 10 minutes and I have no idea whether this is an (attempted) arms’-length assessment of discrimination law (I say attempted because of the “matters close to me” acknowledgement) or the bitter result of the author being turned down for a job. At first glance it looks like the latter, but this is exactly the sort of situation I would expect to see someone to make a completely rational analysis and not pay any attention to how it’s going to come across to someone who doesn’t know you’re not just another bigot. (I call it Larry Summers Syndrome.
It’s one thing to talk about “risk profiles” or “incentives” in general terms, but when you actually want to implement something, it becomes a particular incentive, and there is no a priori reason to assume the cost will outweigh the benefit. When you concentrate on the existence of a cost (or benefit) and ignore the magnitude, you start making statements like “[the Bush tax cuts] increased revenue, because of the vibrancy of these tax cuts in the economy”. Similarly, if you try to transfer utility from group A to group B, group A is going to be upset and try to minimize their loss—that doesn’t mean that group A is going to completely negate it, or that group B is going to be worse off.
Now I don’t know what you’re trying to say about me. My views on this don’t count because you think I was turned down for a job and blamed discrimination law for it? Huh?
In any case, I agree that the costs of A’s reaction don’t necessarily negate the benefit. What I criticize is models that view such utility shifting as a one-shot enterprise without complications, which is a sadly common belief. What’s more common—just from my personal experience—is that attempts to nobly shift utility this way result in making life more kafkaesque.
For example, when you ban IQ tests in employment screening, you don’t get employers happily ignoring the information value of IQ tests—rather, they just fob it off to a university, who will gladly give the IQ test and pass on a weaker measure of ability. If you mandate benefits, employers don’t simply continue their hiring practices exactly as before but give workers more utility; rather, they cut back in other ways.
I believe I’m impacted by this mentality, because I would much rather be told, you don’t qualify because of ____ than have to follow some complicated, unspoken signaling dance (i.e. the relative significance of having a network to having ability) that avoids officially-banned screening methods.
Yes, I’ve been turned down for jobs, but a) I’ve been gainfully employed in my field for 5 years, and b) my concern is not with being turned down, but with it being harder to find opportunities that could result in being turned down in the first place.
What would be a non-bigoted way to make the point I just did?
“I have no idea what criteria were used when I’m rejected for a job, and I’m not even seeing the jobs that never get posted because it’s easier to hire someone you know than go through the formal process and jump through all its hoops.” Maybe.
I don’t think your views don’t count—I was hoping that I’d gone to sufficient lengths to point out that while it might have just been bitterness, there was a substantial chance it wasn’t. Maybe I underestimated the LW rationalist:racist ratio… actually, probably by a huge margin. %#$@.
So what would happen if you traded the kafkaesque life for the officially-banned screening methods? Would you rather have twice the number of job opportunities and lose 3⁄4 of them right away because you’re ? Or would you rather that other people get rejected for them, if you don’t personally have many of the ‘undesirable’ attributes?
Finally, let’s go to story mode. A friend of mine applied for a job. They weren’t allowed to ask her about her religion. But she has a name commonly found among members of a particular one. She got the job, and became the only employee (out of a couple dozen) not sharing the faith of the rest of them. So I guess they took a guess at her religion based on her name, and chose using that metric. I have no idea whether this is a success or failure of antidiscrimination laws. Without them, she’d have had no chance. With them, they tried anyways. But at least it was possible, even if she didn’t fit in… and quit a few years later, when her husband got cancer and they blamed it on her not praying.
At risk of exhibiting confirmation bias, your anecdote makes my point for me. With the ban on discrimination, your friend got misleading signals of which employers she would like, and her employer relied on weaker signals of things they can’t ask about, leading to an incorrect inference—and later, a disastrous mismatch.
So, far from eliminating the pernicious effects of bigotry, the law made people waste effort trying to route their efforts around it. Had there been no law, people could openly communicate their preferences in both directions, without having to go through the complication of sending weaker signals because they can’t send the banned ones. And with less “noise”, the cost of bigotted preferences becomes clearer. Explanation
I don’t see why you’re using this as an example of why anti-discrimination laws are good.
So what would happen if you traded the kafkaesque life for the officially-banned screening methods? Would you rather have twice the number of job opportunities and lose 3⁄4 of them right away because you’re ? Or would you rather that other people get rejected for them, if you don’t personally have many of the ‘undesirable’ attributes?
Certainly, you can pick the numbers to reach the conclusion that you want. But averaging over all possibilities, and weighting by likelihood, yes, I would prefer the environment that makes all job applicants not a potential albatross for employers, for the same reason I would prefer not to be exempt from lawsuits—yes, there’s a narrow benefit to laws saying “you can’t discriminate against people named Silas”, and to a personal exemption from lawsuits. But realistically, the way people respond to this will more than eliminate the benefit. (In case it’s not clear—do you see a hazard in associating with a stranger when you’re guaranteed to have no legal recourse against what they do?)
So yes, I would rather have the option to most clearly communicate preferences, than have to dance around them, which results in situations where someone can be turned down for a job because of false beliefs that they can’t even refute. This remains true, even and especially if I’m in a less fortunate end of the applicant pool (which I have been, like everyone else who has been under 14/16/18 at some point in their life).
I’m not using it as an example of why they’re good. I’m offering it as an example because it’s relevant to the topic.
Adding a cost to circumvent the law makes you less likely to do so, though. If you keep hiring people who are decidedly suboptimal because you have to use a lousy approximation of whatever characteristic you want, you might give up on it.
I get that you would rather, given that you’re going to be rejected for your age/skin color/gender/etc, be told why. But if you want to reduce the use of those criteria, then banning it will stop the people who care a small amount (i.e not enough to bother getting around it.)
This is actually very similar to an intuition I’ve had about this problem. The difference is that I compare it to a different scenario, and regard it as a reason, not to reject the trolley problem, but a reason to justify the optimality of not pushing the innocent bystander onto the tracks.
You compared it to wealth-redistribution attempting to optimize total utility, while I think a better comparison is discrimination law, something that matters close to me. (Edit: sorry for awkward phrasing, keeping it there because it was quoted.) In short, just as milking the cow makes it waste effort avoiding being milked, discrimination law simply makes employers drastically revise hiring practices so that they don’t use the one that counts under the law as a “job posting”.
So they rely more heavily on network hires and indirect signals of employee quality, and, ironically, make it much harder for people outside these networks—usually the “protected” classes—to even get a chance in the first place, no matter how good they could demonstrate their work to be!
People think they can just shove these “surprise” costs onto agents with “too much utility”, ignoring the long-term ways they’ll react. As I said before about the problem, one non-obvious result of shoving the guy onto the track is that drastically shifts around the risk profiles of various activities. Previously, standing on a track was riskier than staying off it. But a policy of switching this risk around makes it harder overall for people to ascertain risks, which makes them spend wasteful effort avoiding risks (since they have to assume they’re understated).
Generally, it really bothers me when people talk in terms of policies that attempt to shift utility from privileged group A to pitiable group B, not realizing that A will still gets its privileges and exclude B—it’ll just become a more complex dance for B to figure out.
I’ve been rereading this comment for the past 10 minutes and I have no idea whether this is an (attempted) arms’-length assessment of discrimination law (I say attempted because of the “matters close to me” acknowledgement) or the bitter result of the author being turned down for a job. At first glance it looks like the latter, but this is exactly the sort of situation I would expect to see someone to make a completely rational analysis and not pay any attention to how it’s going to come across to someone who doesn’t know you’re not just another bigot. (I call it Larry Summers Syndrome.
http://en.wikipedia.org/wiki/Lawrence_Summers#Differences_between_the_sexes )
It’s one thing to talk about “risk profiles” or “incentives” in general terms, but when you actually want to implement something, it becomes a particular incentive, and there is no a priori reason to assume the cost will outweigh the benefit. When you concentrate on the existence of a cost (or benefit) and ignore the magnitude, you start making statements like “[the Bush tax cuts] increased revenue, because of the vibrancy of these tax cuts in the economy”. Similarly, if you try to transfer utility from group A to group B, group A is going to be upset and try to minimize their loss—that doesn’t mean that group A is going to completely negate it, or that group B is going to be worse off.
Now I don’t know what you’re trying to say about me. My views on this don’t count because you think I was turned down for a job and blamed discrimination law for it? Huh?
In any case, I agree that the costs of A’s reaction don’t necessarily negate the benefit. What I criticize is models that view such utility shifting as a one-shot enterprise without complications, which is a sadly common belief. What’s more common—just from my personal experience—is that attempts to nobly shift utility this way result in making life more kafkaesque.
For example, when you ban IQ tests in employment screening, you don’t get employers happily ignoring the information value of IQ tests—rather, they just fob it off to a university, who will gladly give the IQ test and pass on a weaker measure of ability. If you mandate benefits, employers don’t simply continue their hiring practices exactly as before but give workers more utility; rather, they cut back in other ways.
I believe I’m impacted by this mentality, because I would much rather be told, you don’t qualify because of ____ than have to follow some complicated, unspoken signaling dance (i.e. the relative significance of having a network to having ability) that avoids officially-banned screening methods.
Yes, I’ve been turned down for jobs, but a) I’ve been gainfully employed in my field for 5 years, and b) my concern is not with being turned down, but with it being harder to find opportunities that could result in being turned down in the first place.
What would be a non-bigoted way to make the point I just did?
“I have no idea what criteria were used when I’m rejected for a job, and I’m not even seeing the jobs that never get posted because it’s easier to hire someone you know than go through the formal process and jump through all its hoops.” Maybe.
I don’t think your views don’t count—I was hoping that I’d gone to sufficient lengths to point out that while it might have just been bitterness, there was a substantial chance it wasn’t. Maybe I underestimated the LW rationalist:racist ratio… actually, probably by a huge margin. %#$@.
So what would happen if you traded the kafkaesque life for the officially-banned screening methods? Would you rather have twice the number of job opportunities and lose 3⁄4 of them right away because you’re ? Or would you rather that other people get rejected for them, if you don’t personally have many of the ‘undesirable’ attributes?
Finally, let’s go to story mode. A friend of mine applied for a job. They weren’t allowed to ask her about her religion. But she has a name commonly found among members of a particular one. She got the job, and became the only employee (out of a couple dozen) not sharing the faith of the rest of them. So I guess they took a guess at her religion based on her name, and chose using that metric. I have no idea whether this is a success or failure of antidiscrimination laws. Without them, she’d have had no chance. With them, they tried anyways. But at least it was possible, even if she didn’t fit in… and quit a few years later, when her husband got cancer and they blamed it on her not praying.
At risk of exhibiting confirmation bias, your anecdote makes my point for me. With the ban on discrimination, your friend got misleading signals of which employers she would like, and her employer relied on weaker signals of things they can’t ask about, leading to an incorrect inference—and later, a disastrous mismatch.
So, far from eliminating the pernicious effects of bigotry, the law made people waste effort trying to route their efforts around it. Had there been no law, people could openly communicate their preferences in both directions, without having to go through the complication of sending weaker signals because they can’t send the banned ones. And with less “noise”, the cost of bigotted preferences becomes clearer. Explanation
I don’t see why you’re using this as an example of why anti-discrimination laws are good.
Certainly, you can pick the numbers to reach the conclusion that you want. But averaging over all possibilities, and weighting by likelihood, yes, I would prefer the environment that makes all job applicants not a potential albatross for employers, for the same reason I would prefer not to be exempt from lawsuits—yes, there’s a narrow benefit to laws saying “you can’t discriminate against people named Silas”, and to a personal exemption from lawsuits. But realistically, the way people respond to this will more than eliminate the benefit. (In case it’s not clear—do you see a hazard in associating with a stranger when you’re guaranteed to have no legal recourse against what they do?)
So yes, I would rather have the option to most clearly communicate preferences, than have to dance around them, which results in situations where someone can be turned down for a job because of false beliefs that they can’t even refute. This remains true, even and especially if I’m in a less fortunate end of the applicant pool (which I have been, like everyone else who has been under 14/16/18 at some point in their life).
(Btw, I’m not the one modding you down.)
I’m not using it as an example of why they’re good. I’m offering it as an example because it’s relevant to the topic.
Adding a cost to circumvent the law makes you less likely to do so, though. If you keep hiring people who are decidedly suboptimal because you have to use a lousy approximation of whatever characteristic you want, you might give up on it.
I get that you would rather, given that you’re going to be rejected for your age/skin color/gender/etc, be told why. But if you want to reduce the use of those criteria, then banning it will stop the people who care a small amount (i.e not enough to bother getting around it.)