5 ms·

I applied for an Apple Card. What they offered was a sexist insult

5 points by sonabinu 7y ago

zepto 7y ago
Did DHH have 20 times her wealth prior to their marriage? It seems likely given that he is a guy who collects Zondas. GS have stated that they don’t know your marital status or gender when they make the decision. There may well be sexism somewhere in how this works, but she doesn’t seem to be making a good argument for where.
- eesmith 7y ago
  They have been married "a long time", tweets Hansson. How long do you think is long enough? She's a millionaire too. And with a higher credit score. How do you know that she doesn't also collect expensive items? Because it sounds like you are fishing for a reason. Her credit limit got raised, so apparently GS now thinks the algorithm messed up. You write: "GS have stated that they don’t know your marital status or gender when they make the decision". That's great - so, can they explain how they came to their assessment? And can these models be published or otherwise made available for review in case there are mistake or systemic errors? Why can't customer service do anything better than say "wait 6 months and see what happens?" What information does GS use, and does that information include an implicit gender bias? For example, even if gender isn't known, it can often be inferred from someone's first name. Finally, Hansson also tweeted: > It gets even worse. Even when she pays off her ridiculously low limit in full, the card won’t approve any spending until the next billing period. How does that make any sense for a millionaire with a high credit score? What would you consider a good enough argument that you wouldn't have written your comment? Because it seems impossible for either Hansson to present an argument that can satisfy you, at least, not without lawsuit and publication of materials found through discovery.
  - zepto 7y ago
    Their marriage is irrelevant because as GS have stated, they don’t have that information so they don’t take it into account. They claim they don’t know the person’s gender. If they are inferring that from a first name or from other information they have then their statement is a lie. By all means claim that GS are lying. I’m curious to see what the result of the investigation is. As for paying the card off in full and still not being able to spend more, indeed that doesn’t make sense for a millionaire with a high credit score. Clearly the system is not doing a good job of assessing what credit to offer her. However the fact that she is being treated differently from her husband, who is not just a millionaire - but someone who is estimated to have >$100m, is not evidence of sexism. Sexism would be a systematic bias. She hasn’t provided any evidence whatsoever of that, although she has demonstrated that the system is bad at recognizing creditworthiness. So yes - the bar for demonstrating sexism is indeed much higher, and yes, materials about how the algorithm works so implicit biases can be evaluated would necessarily be part of that. I hope the New York Financial services investigation does get access to these materials. My guess is that if there is a sexist bias it will take some tracking down, and that it will be a result of industry wide practices. I think this should be reviewed across the board. But I stand by my original comment that she hasn’t shown that there is sexism at work in her own case.
    - eesmith 7y ago
      If it's irrelevant than your comment "prior to their marriage" is also irrelevant, yes? Because that's what I was responding to. They don't claim to not know the person's gender. They write "Our credit decisions are based on a customer's creditworthiness and not on factors like gender, race, age, sexual orientation or any other bias prohibited by law," That is, they might know the person's gender, but not include that in the assessment. Why must their statement be a lie if there is a gender bias? There are plenty of reports of risk models which have an illegally discriminatory bias even when not designed that way. For one example https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing https://www.propublica.org/article/machine-bias-risk-assessm... . In that case, 'Brennan said it is difficult to construct a score that doesn’t include items that can be correlated with race — such as poverty, joblessness and social marginalization. “If those are omitted from your risk assessment, accuracy goes down,” he said.' As I pointed out, there are correlations between names and gender. Almost certainly there are other correlations. A model might have trained on those factors to produce a result with an illegal gender bias, even if that wasn't deliberate. When you write "the bar for demonstrating sexism is indeed much higher", do you mean the bar for legal action, or the bar by which someone can make a public accusation? If I am black, and walk into a restaurant and refused service, then my white spouse walks in 5 minutes later and gets service, do you expect that I must remain quiet about it until I've demonstrated in a court of law that actual illegal discriminatory practice occurred? And if I go public before that's done, will you say that I haven't given you enough evidence? If yes, then in both cases, I think you are using the wrong threshold for when it's relevant and useful to go public, because doing so is part of one way to drive the resolution process. Or, if I have a child who gets a rare illness due to some heavy metal poisoning, and a local company is dumping those heavy metals into the water supply, do I need to wait until there's conclusive evidence, as for a court of law, before I go public about it? In all three cases, sure, I can be wrong. (Courts can be wrong too.) But ... so what?
      - zepto 7y ago
        Yes - it doesn’t matter whether it was prior to their marriage or not. It seems entirely possible that DHH has 20x her net worth. Perhaps that’s involved in their decision making, no? See: https://twitter.com/gsbanksupport/status/1194022629419704320?s=20 https://twitter.com/gsbanksupport/status/1194022629419704320... “In fact we do not know your gender or marital status during the Apple Card application process”. So your comments about them knowing her gender are either false, or they are lying. The analogy of your examples is invalidated by GS’s statement that they didn’t know her gender or marital status. And yes - I agree that there may be some implicit bias in a scoring algorithm somewhere along the line, or some other systemic influence. As I have said, I’d like to see the investigation look into this possibility. But “I got lower credit than my spouse”, simply is not evidence of sexism, especially if the spouse is 20x richer.

5 points by sonabinu 7y ago

Did DHH have 20 times her wealth prior to their marriage? It seems likely given that he is a guy who collects Zondas. GS have stated that they don’t know your marital status or gender when they make the decision. There may well be sexism somewhere in how this works, but she doesn’t seem to be making a good argument for where.

eesmith 7y ago
They have been married "a long time", tweets Hansson. How long do you think is long enough? She's a millionaire too. And with a higher credit score. How do you know that she doesn't also collect expensive items? Because it sounds like you are fishing for a reason. Her credit limit got raised, so apparently GS now thinks the algorithm messed up. You write: "GS have stated that they don’t know your marital status or gender when they make the decision". That's great - so, can they explain how they came to their assessment? And can these models be published or otherwise made available for review in case there are mistake or systemic errors? Why can't customer service do anything better than say "wait 6 months and see what happens?" What information does GS use, and does that information include an implicit gender bias? For example, even if gender isn't known, it can often be inferred from someone's first name. Finally, Hansson also tweeted: > It gets even worse. Even when she pays off her ridiculously low limit in full, the card won’t approve any spending until the next billing period. How does that make any sense for a millionaire with a high credit score? What would you consider a good enough argument that you wouldn't have written your comment? Because it seems impossible for either Hansson to present an argument that can satisfy you, at least, not without lawsuit and publication of materials found through discovery.
- zepto 7y ago
  Their marriage is irrelevant because as GS have stated, they don’t have that information so they don’t take it into account. They claim they don’t know the person’s gender. If they are inferring that from a first name or from other information they have then their statement is a lie. By all means claim that GS are lying. I’m curious to see what the result of the investigation is. As for paying the card off in full and still not being able to spend more, indeed that doesn’t make sense for a millionaire with a high credit score. Clearly the system is not doing a good job of assessing what credit to offer her. However the fact that she is being treated differently from her husband, who is not just a millionaire - but someone who is estimated to have >$100m, is not evidence of sexism. Sexism would be a systematic bias. She hasn’t provided any evidence whatsoever of that, although she has demonstrated that the system is bad at recognizing creditworthiness. So yes - the bar for demonstrating sexism is indeed much higher, and yes, materials about how the algorithm works so implicit biases can be evaluated would necessarily be part of that. I hope the New York Financial services investigation does get access to these materials. My guess is that if there is a sexist bias it will take some tracking down, and that it will be a result of industry wide practices. I think this should be reviewed across the board. But I stand by my original comment that she hasn’t shown that there is sexism at work in her own case.
  - eesmith 7y ago
    If it's irrelevant than your comment "prior to their marriage" is also irrelevant, yes? Because that's what I was responding to. They don't claim to not know the person's gender. They write "Our credit decisions are based on a customer's creditworthiness and not on factors like gender, race, age, sexual orientation or any other bias prohibited by law," That is, they might know the person's gender, but not include that in the assessment. Why must their statement be a lie if there is a gender bias? There are plenty of reports of risk models which have an illegally discriminatory bias even when not designed that way. For one example https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing https://www.propublica.org/article/machine-bias-risk-assessm... . In that case, 'Brennan said it is difficult to construct a score that doesn’t include items that can be correlated with race — such as poverty, joblessness and social marginalization. “If those are omitted from your risk assessment, accuracy goes down,” he said.' As I pointed out, there are correlations between names and gender. Almost certainly there are other correlations. A model might have trained on those factors to produce a result with an illegal gender bias, even if that wasn't deliberate. When you write "the bar for demonstrating sexism is indeed much higher", do you mean the bar for legal action, or the bar by which someone can make a public accusation? If I am black, and walk into a restaurant and refused service, then my white spouse walks in 5 minutes later and gets service, do you expect that I must remain quiet about it until I've demonstrated in a court of law that actual illegal discriminatory practice occurred? And if I go public before that's done, will you say that I haven't given you enough evidence? If yes, then in both cases, I think you are using the wrong threshold for when it's relevant and useful to go public, because doing so is part of one way to drive the resolution process. Or, if I have a child who gets a rare illness due to some heavy metal poisoning, and a local company is dumping those heavy metals into the water supply, do I need to wait until there's conclusive evidence, as for a court of law, before I go public about it? In all three cases, sure, I can be wrong. (Courts can be wrong too.) But ... so what?
    - zepto 7y ago
      Yes - it doesn’t matter whether it was prior to their marriage or not. It seems entirely possible that DHH has 20x her net worth. Perhaps that’s involved in their decision making, no? See: https://twitter.com/gsbanksupport/status/1194022629419704320?s=20 https://twitter.com/gsbanksupport/status/1194022629419704320... “In fact we do not know your gender or marital status during the Apple Card application process”. So your comments about them knowing her gender are either false, or they are lying. The analogy of your examples is invalidated by GS’s statement that they didn’t know her gender or marital status. And yes - I agree that there may be some implicit bias in a scoring algorithm somewhere along the line, or some other systemic influence. As I have said, I’d like to see the investigation look into this possibility. But “I got lower credit than my spouse”, simply is not evidence of sexism, especially if the spouse is 20x richer.