This has actually proven to be the case; there are a number of studies on assaults committed against taxi drivers, and all (I've seen) come to the conclusion that males and African-Americans are much more likely to be violent.
I am not apologizing for anyone. It should be rather unsurprising that men are more violent to taxi-drivers than women, as men are incarcerated at a rate between 10-30 times higher than women.
Can you provide any reputable data that proves this to be accurate? Please share the link or provide evidence, otherwise you are exasperating the problem that the article outlines.
I have stopped responding when people simply ask for citations. Saying "citation needed" is a low-value comment, and I have better things to do than look up studies which I've read, and you will probably dismiss out-of-hand. If you have good evidence which contradicts my comment, please post it, so that myself and others can learn.
I definitely understand the inclination to avoid doing this sort of citation-finding work, but I think the approach you've chosen is a worse one than you think. You say comments that only ask for citation are low value, and that may be true, but I think comments that are of the form, "the studies I have seen say (insert very debatable thing)", without providing any support, are even lower value. By this logic, you can say anything you want, while implying that it has been studied by experts, when in fact you just made it up. I wish this were a straw-man hypothetical, but this sort of discourse is actually so common now that whenever I see someone vaguely cite "something I have seen in studies", I assume it actually means "something I heard someone say they saw in studies", and that it is totally fabricated.
It's also pretty rich to have the position that you won't put in the work to back up your statements, but that those who call you out on that should be willing to do the work to disprove you. Just replying to you with "I have seen studies that say the opposite" is equally as (un)convincing as what you've said.
I wrote a comment which described the state of my knowledge, and would be fine if another commenter said they disagreed for any reason (anecdotal or evidenciary). If the respondent had said "I looked, but could not find studies which correspond with what you are saying", I would have made more effort.
The problem behavior which I have seen cropping up on Hacker News is people asking for citations as a quick and easy way of invalidating other people's views, while justifying downvotes (which have become far too common). I have also observed that many will ask for studies, then either ignore them for viewpoint or ad hominiem reasons, or attempt to invalidate the study by nit-picking. If the original commenter lied or pulled numbers out of thin air, it is fair to call them out (in my view), but laziness on the part of the respondent is no justification for snark.
Apparently, some of us consider what you said to be an "extraordinary claim" requiring evidence. You are essentially saying:
"Prove 'the state of my knowledge' is wrong."
The burden of proof lies with the original claim maker, not the respondent. This is not a bar room conversation from the 90s, we are literally on the internet; citing one's sources isn't too much to ask.
What is your standard for evidence? Without knowing that, I cannot hope to give 'conclusive evidence'.
Do you have a white-list, black-list, or other standard for sources? What type of evidence would satisfy your standards, and what are the minimum sample size, maximum p-value, and robustness with respect to alternate causes? Do you require some amount of replication, peer review, or publication?
Frequentist statistics is racist, don't you know? In today's progressive society, it is forbidden to make informed decisions based on statistics. Racial profiling is the devil - if the clerk said the robber was black, you better spend 50% of the time searching for a white guy, or else you're really deplorable.
They always need to consider crime rates in these studies. One was posted about Amazon Prime a while back showing they weren't delivering to a number of mostly-black areas. This they said was evidence of racial discrimination. They didn't mention that those areas, which I recognized, were a bunch of "hoods" very high in violent crime and theft. Most truck drivers will resist going into such places not to mention liability or cost conscious companies. So, the truth about them avoiding dangerous areas got distorted into them just being racist.
Need to always account for that. In this case, they're talking Seattle and Boston. I don't know how much crime is involved in these areas or what amount of the study happened in ghetto areas. If these were minimal, then it would be evidence of racism. If not, then crime angle has to be examined.
I'm reminded of an article I read in a magazine or newspaper a long time ago. Briefly, a school district gave IQ tests to all of the children in a particular grade. There was a bimodal distribution, with the black children scoring significantly lower than the white children.
Most officials in the district concluded that the test must be culturally biased or flawed in some other way, and were going to ignore the results.
One teacher, though, decided to operate under the assumptions that (1) the test was not flawed and so was showing a real difference between the children, and (2) there was no inherent racial difference in the students, and so started looking for other explanations for the lower performance of the black kids.
He found that all the black children lived in older housing, built in the in the early '70s or earlier. All the white children lived in newer housing, built after 1980. He knew that we had stopped using lead paint in housing in 1978, and wondered if the black kids could be suffering from lead poisoning.
They and their houses were tested, and sure enough, the black kids were living in high lead environments. One of the effects of lead poising is a reduction in cognitive abilities.
The houses were cleaned of lead, and the children treated, and within a couple of years the gap on IQ tests went away.
If that teacher thought like you did he would have just dismissed the question of whether or not the black kids' performance was due to a real cognitive difference as offensive, unacceptable and ridiculous, and those kids would have went through much or all of the rest of their childhood and teen years suffering from cognitive impairment and the accompanying limitations on their futures.
The question was offensive. The teacher asked and very carefully answered a question about demography, using IQ tests not as an assessment of racial superiority but as an epidemiological indicator. Here, the the question is "maybe black people are just problem passengers". It's a pipe bomb thrown into the conversation, not an attempt to get closer to the truth.
I can see how it could be read that way. I gave a lot of weight to the words "disproportionately likely" in his phrasing ("might be disproportionately likely to cause problems"), and assumed he was suggesting things like the correlations between race and crime or between race and tipping (both things that might be of interest to a taxi driver).
The race and crime correlation comes about because poor people are more likely then middle class or wealthy people to commit the kind of crimes that a taxi driver would worry about, and wealth distribution is not independent of race in the United States, with black people disproportionately represented among the poor.
I have no idea where the race and tipping correlation comes from. It's well known in waiter circles, with both white and black servers complaining about poor tips from black people. There was an interesting literature review and discussion of this topic published by the Cornell University School of Hotel Administration: http://scholarship.sha.cornell.edu/cgi/viewcontent.cgi?artic...
These and similar correlations raise some interesting questions. Humans naturally generalize from experience and make predictions based on correlations. Since race is visible it gets involved in those correlations. Other factors, like economic status, that are probably closer to the actual cause of whatever behavior one is observing and generalizing from are not directly visible so we don't use them as much in our generalizations.
So one question that comes to mind is can we make those lower level factors more visible? What if an Uber driver got essentially a background check on their potential passenger, giving income level, education level, arrest record, and so on? Would drivers be less likely to generalize based on race if they could instead generalize on those other factors? Then if a driver had 300 white passengers and 100 black passengers in a month and 5 whites and 10 blacks caused problems, instead of generalizing that blacks are problem passengers the driver might notice that all 15 of the problem passengers were involved in gang activity and the generalization becomes "don't pick up people involved in gang activity".
I realize that giving the aforementioned information to drivers would be a big loss of privacy for passengers. Maybe there is an inherent conflict between privacy and reducing race-based generalizations, much like the conflict between privacy and security.
I bet a good writer could make an interesting science fiction story around that idea and its implications.
That matches my intuitive opinion about it. And it only gets worse with the sharing economy. As soon as you choose your peer, it introduced human judgement.
One interesting thing would be to measure which race is the most racist, so we can sensibilize the right population.
Another interesting thing would be to have objective metrics to judge whether a transaction went well (whether the passenger was at the predefined location, on time, and whether the car was used with care).
I'll say it: Sometimes racism is based on a correct evaluation of the risk. So what can we do to diminish the risk?
The abstract states "[p]eer transportation companies such as Uber and Lyft present the opportunity to rectify long-standing discrimination or worsen it", but the study appears (from my quick glance) not to have directly compared Uber, Lyft, and taxis with the same methodology. The (scant) evidence regarding likely racial discrimination by taxis seems to indicate that Lyft and Uber are much less discriminatory.
Interesting, but the standard deviations of the metrics are as large or larger than the mean in every case. Also, there is a lack of trend between independent replicates.
So, very noisy data to say the least. I think their taxi figure in the appendix highlights the problem of discrimination much more clearly than their analysis of this dataset.
Consider a world where you are assigned a random phrase (i.e. "Correct Horse Battery Staple") when you request a ride. The driver is never shown a name. When they are shown the set of people they could pick up, they are not shown a photo or passphrase until they accept the ride. As soon as they accept your ride, they are shown your passphrase and your photo (the former to ensure that they are picking up the right passenger, and the latter to ensure that a previously vetted passenger isn't cheating and providing their account to someone else / buying a ride for someone else).
Wouldn't this go a pretty long way towards removing obvious sources of discrimination?
Not letting the drivers see a location and trying to force them to drive into ghettos where they might get carjacked will probably not go over too well, especially with female drivers.
I think this is absolutely true; and you'd want to show drivers locations - I can't think of any way to obscure this, and it does leave a source for discrimination ("don't drive into the dangerous neighborhoods").
I'm not sure I have any good solution to that particular problem, but surely perfect is the enemy of good, and the human anonymization approach is reasonable low hanging fruit?
Isn't the answer to the latter known? Working class citizens are more likely to be victimized in the presence of those neighborhoods/people/demographics.
Hence why arguments against racism are never on the grounds that it's irrational, but rather that it's low-status/unpopular etc.
33 comments
[ 0.16 ms ] story [ 86.2 ms ] threadIt's also pretty rich to have the position that you won't put in the work to back up your statements, but that those who call you out on that should be willing to do the work to disprove you. Just replying to you with "I have seen studies that say the opposite" is equally as (un)convincing as what you've said.
The problem behavior which I have seen cropping up on Hacker News is people asking for citations as a quick and easy way of invalidating other people's views, while justifying downvotes (which have become far too common). I have also observed that many will ask for studies, then either ignore them for viewpoint or ad hominiem reasons, or attempt to invalidate the study by nit-picking. If the original commenter lied or pulled numbers out of thin air, it is fair to call them out (in my view), but laziness on the part of the respondent is no justification for snark.
"Prove 'the state of my knowledge' is wrong."
The burden of proof lies with the original claim maker, not the respondent. This is not a bar room conversation from the 90s, we are literally on the internet; citing one's sources isn't too much to ask.
Do you have a white-list, black-list, or other standard for sources? What type of evidence would satisfy your standards, and what are the minimum sample size, maximum p-value, and robustness with respect to alternate causes? Do you require some amount of replication, peer review, or publication?
Need to always account for that. In this case, they're talking Seattle and Boston. I don't know how much crime is involved in these areas or what amount of the study happened in ghetto areas. If these were minimal, then it would be evidence of racism. If not, then crime angle has to be examined.
Albeit it might be more acceptable to discuss it from a monetary perspective, e.g. "Why are urban poor crime rates so high"
It could very well be the case that race is itself a factor for anti-social behavior.
Most officials in the district concluded that the test must be culturally biased or flawed in some other way, and were going to ignore the results.
One teacher, though, decided to operate under the assumptions that (1) the test was not flawed and so was showing a real difference between the children, and (2) there was no inherent racial difference in the students, and so started looking for other explanations for the lower performance of the black kids.
He found that all the black children lived in older housing, built in the in the early '70s or earlier. All the white children lived in newer housing, built after 1980. He knew that we had stopped using lead paint in housing in 1978, and wondered if the black kids could be suffering from lead poisoning.
They and their houses were tested, and sure enough, the black kids were living in high lead environments. One of the effects of lead poising is a reduction in cognitive abilities.
The houses were cleaned of lead, and the children treated, and within a couple of years the gap on IQ tests went away.
If that teacher thought like you did he would have just dismissed the question of whether or not the black kids' performance was due to a real cognitive difference as offensive, unacceptable and ridiculous, and those kids would have went through much or all of the rest of their childhood and teen years suffering from cognitive impairment and the accompanying limitations on their futures.
The race and crime correlation comes about because poor people are more likely then middle class or wealthy people to commit the kind of crimes that a taxi driver would worry about, and wealth distribution is not independent of race in the United States, with black people disproportionately represented among the poor.
I have no idea where the race and tipping correlation comes from. It's well known in waiter circles, with both white and black servers complaining about poor tips from black people. There was an interesting literature review and discussion of this topic published by the Cornell University School of Hotel Administration: http://scholarship.sha.cornell.edu/cgi/viewcontent.cgi?artic...
These and similar correlations raise some interesting questions. Humans naturally generalize from experience and make predictions based on correlations. Since race is visible it gets involved in those correlations. Other factors, like economic status, that are probably closer to the actual cause of whatever behavior one is observing and generalizing from are not directly visible so we don't use them as much in our generalizations.
So one question that comes to mind is can we make those lower level factors more visible? What if an Uber driver got essentially a background check on their potential passenger, giving income level, education level, arrest record, and so on? Would drivers be less likely to generalize based on race if they could instead generalize on those other factors? Then if a driver had 300 white passengers and 100 black passengers in a month and 5 whites and 10 blacks caused problems, instead of generalizing that blacks are problem passengers the driver might notice that all 15 of the problem passengers were involved in gang activity and the generalization becomes "don't pick up people involved in gang activity".
I realize that giving the aforementioned information to drivers would be a big loss of privacy for passengers. Maybe there is an inherent conflict between privacy and reducing race-based generalizations, much like the conflict between privacy and security.
I bet a good writer could make an interesting science fiction story around that idea and its implications.
One interesting thing would be to measure which race is the most racist, so we can sensibilize the right population.
Another interesting thing would be to have objective metrics to judge whether a transaction went well (whether the passenger was at the predefined location, on time, and whether the car was used with care).
I'll say it: Sometimes racism is based on a correct evaluation of the risk. So what can we do to diminish the risk?
So, very noisy data to say the least. I think their taxi figure in the appendix highlights the problem of discrimination much more clearly than their analysis of this dataset.
Wouldn't this go a pretty long way towards removing obvious sources of discrimination?
I'm not sure I have any good solution to that particular problem, but surely perfect is the enemy of good, and the human anonymization approach is reasonable low hanging fruit?
At some point the question should probably change from
"How can we force Uber drivers to increase risk of losing well-being, wallet, and car in the name of fairness to customers?"
to
"Why are these neighborhoods/people/demographic so scary to working class citizens, and what can we do to fix it?"
Hence why arguments against racism are never on the grounds that it's irrational, but rather that it's low-status/unpopular etc.