Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

time.com

Post reply on HN
Print view View original

91–100 of 111 posts

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#91

Post by dragonwriter »

yummyfajitas wrote: post #88
thaw13579 wrote: post #75

Earlier quoted context omitted.

I appreciate this analysis, but it's important to note that not everyone wants unbiased model output. Historically, redlining was intentional, and it took laws to stop it. A major risk of automating mortgage engines is that redlining could easily be reinstated (in the form of model parameters) in a way that is both hard to detect and easily deniable.

There are two possible reasons for redlining: 1) I hate black people more than I like money, but blacks do pay back their loans, so I'll redline even if it costs me money. 2) I like money and am neutral towards blacks, but blacks are deadbeats so redlining gets me more money. (There are also 2 other cases. "I love/am neutral towards blacks and they pay back their loans", which results in no disparate impact. Also "I…

There is also "I am neutral to blacks and view them as, ceteris paribus, equally likely to pay back loans, but white people on average are not neutral toward them, and particularly are not neutral to living in close proximity to them, and the decrease in property values when black people are allowed in otherwise white neighborhoods makes both their default and the default of white people in the neighborhood more likely, and makes deficiency more likely in the event of default of either."

(Statistical models that properly account track all-factors likelihood of default on a particular loan do not solve this.)

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#92

Post by rlucas »

yummyfajitas wrote: post #71
lmm wrote: post #58

Earlier quoted context omitted.

There's no historical data on the default rates of people who didn't get mortgages, because they didn't get mortgages. I'm willing to believe that theoretical solutions exist. I know from direct personal experience in the big data/lending industry that they are not always applied. If you are claiming that real-world lenders never train their models on human decisions then you are simply wrong. Nice zinger - I hope yo…

Sure, but there is plenty of historical data on people who marginally got mortgages. If those people are solidly profitable, as opposed to marginally profitable, you try a marginally more expansive lending criteria. I.e., if I know f(0) = 10, f(1) = 9, f(2) = 8, but I don't have data on f(3), it's worth running a bit of an experiment to see if f(3) is actually 7. (I happen to know from experience I can't talk about t…

You're right within the boundaries of what you're considering, meaning testing the profitability of marginal borrowers, etc. But you're missing the big picture question about the use of ML in making political decisions (and yes, the question of who gets mortgages is very much a political one).

For example, in the earliest days of the creation of the US national mortgage market, conforming to FHA standards meant following more or less explicit racial guidelines (not to mention car-promoting platting and urban planning guidelines). That amounted to a political decision to reward people who invested (and/or resided) in white suburbia with greater liquidity and hence price appreciation.

Focusing on the marginal profitability of individual loans ignore the bigger issue, which is that choosing the rules often encodes a much larger set of political preferences or goals. Then, the critique of using opaque algos is, for me, less that they hide the use of some verboten input variable like race or sex, but that by focusing on the algos we ignore the bigger question.

(it's a subclass of what I call the scalar fallacy, where any real world outcome being reduced to a scalar always collapses a multidimensional space into a single dimension, and that projection contains a huge set of silent preferences)

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#93

Post by rlucas »

yummyfajitas wrote: post #71
lmm wrote: post #58

Earlier quoted context omitted.

There's no historical data on the default rates of people who didn't get mortgages, because they didn't get mortgages. I'm willing to believe that theoretical solutions exist. I know from direct personal experience in the big data/lending industry that they are not always applied. If you are claiming that real-world lenders never train their models on human decisions then you are simply wrong. Nice zinger - I hope yo…

Sure, but there is plenty of historical data on people who marginally got mortgages. If those people are solidly profitable, as opposed to marginally profitable, you try a marginally more expansive lending criteria. I.e., if I know f(0) = 10, f(1) = 9, f(2) = 8, but I don't have data on f(3), it's worth running a bit of an experiment to see if f(3) is actually 7. (I happen to know from experience I can't talk about t…

You're right within the boundaries of what you're considering, meaning testing the profitability of marginal borrowers, etc. But you're missing the big picture question about the use of ML in making political decisions (and yes, the question of who gets mortgages is very much a political one).

For example, in the earliest days of the creation of the US national mortgage market, conforming to FHA standards meant following more or less explicit racial guidelines (not to mention car-promoting platting and urban planning guidelines). That amounted to a political decision to reward people who invested (and/or resided) in white suburbia with greater liquidity and hence price appreciation.

Focusing on the marginal profitability of individual loans ignore the bigger issue, which is that choosing the rules often encodes a much larger set of political preferences or goals. Then, the critique of using opaque algos is, for me, less that they hide the use of some verboten input variable like race or sex, but that by focusing on the algos we ignore the bigger question.

(it's a subclass of what I call the scalar fallacy, where any real world outcome being reduced to a scalar always collapses a multidimensional space into a single dimension, and that projection contains a huge set of silent preferences)

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#94

Post by rlucas »

yummyfajitas wrote: post #88
thaw13579 wrote: post #75

Earlier quoted context omitted.

I appreciate this analysis, but it's important to note that not everyone wants unbiased model output. Historically, redlining was intentional, and it took laws to stop it. A major risk of automating mortgage engines is that redlining could easily be reinstated (in the form of model parameters) in a way that is both hard to detect and easily deniable.

There are two possible reasons for redlining: 1) I hate black people more than I like money, but blacks do pay back their loans, so I'll redline even if it costs me money. 2) I like money and am neutral towards blacks, but blacks are deadbeats so redlining gets me more money. (There are also 2 other cases. "I love/am neutral towards blacks and they pay back their loans", which results in no disparate impact. Also "I…

There's also:

You like money and are neutral toward blacks but there exists a reverse Keynesian beauty contest where you accurately predict that others will redline and hence drive down liquidity and prices, thus making lending less attractive. Hence you are "blamelessly" (but predictably) racist.

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#95

Post by rlucas »

yummyfajitas wrote: post #88
thaw13579 wrote: post #75

Earlier quoted context omitted.

I appreciate this analysis, but it's important to note that not everyone wants unbiased model output. Historically, redlining was intentional, and it took laws to stop it. A major risk of automating mortgage engines is that redlining could easily be reinstated (in the form of model parameters) in a way that is both hard to detect and easily deniable.

There are two possible reasons for redlining: 1) I hate black people more than I like money, but blacks do pay back their loans, so I'll redline even if it costs me money. 2) I like money and am neutral towards blacks, but blacks are deadbeats so redlining gets me more money. (There are also 2 other cases. "I love/am neutral towards blacks and they pay back their loans", which results in no disparate impact. Also "I…

There's also:

You like money and are neutral toward blacks but there exists a reverse Keynesian beauty contest where you accurately predict that others will redline and hence drive down liquidity and prices, thus making lending less attractive. Hence you are "blamelessly" (but predictably) racist.

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#96

Post by hood_syntax »

bko wrote: post #59
hood_syntax wrote: post #48

Earlier quoted context omitted.

> So having more cops in a high-crime neighborhood is somehow a bad thing, because the presence of more police actually confirms the fact that there is more of a need of police in this neighborhood? Did you understand what you quoted? The author is saying that citations for petty offences (which practically everyone commits with some non-zero frequency, intentionally or not) have a higher chance of affecting future p…

> the Author isn't saying police are bad in general, rather that the negative effects of police presence are more significant in communities with higher police presence. Yes, but are there any positive effects of police presence? Does this outweight the negative effects? > The author is saying (assuming you think like them) that ignorance of suffering is a bad thing So seeing poorly targeted ads tells you that people…

> Yes, but are there any positive effects of police presence? Does this outweight the negative effects?

Certainly I would agree that police have positive effects as well. Your second question really is the crux of the issue, and I can't answer that.

> So seeing poorly targeted ads tells you that people are suffering? It is the role of corporations to inform you that there are people who have different needs than you?

Suffering was perhaps poor word choice. Obviously it's not the responsibility of some faceless corporation that I be informed about what's going on in the world, but it would be nice if they didn't cut down incidental information the masses might receive so it fits in their little corner of the world.

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#97

Post by yummyfajitas »

dragonwriter wrote: post #91
yummyfajitas wrote: post #88

Earlier quoted context omitted.

There are two possible reasons for redlining: 1) I hate black people more than I like money, but blacks do pay back their loans, so I'll redline even if it costs me money. 2) I like money and am neutral towards blacks, but blacks are deadbeats so redlining gets me more money. (There are also 2 other cases. "I love/am neutral towards blacks and they pay back their loans", which results in no disparate impact. Also "I…

There is also "I am neutral to blacks and view them as, ceteris paribus , equally likely to pay back loans, but white people on average are not neutral toward them, and particularly are not neutral to living in close proximity to them, and the decrease in property values when black people are allowed in otherwise white neighborhoods makes both their default and the default of white people in the neighborhood more lik…

True, a model which predicts reality will in fact predict reality.

I've never seen any evidence that your hypothesis is true, or even a model that would account for it, but I'm hardly an expert in the field. Assuming this is more than speculation/theoretical nitpicking, do you have citations on the effects you describe?

If your theory is more than just idle speculation, this is a great way for non-racists to engage in house price arbitrage insofar as they wish to consume housing. Do you know of any neighborhoods in NYC where I can exploit this arbitrage?

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#98

Post by dsfyu404ed »

ar0 wrote: post #22
s_q_b wrote: post #15

Earlier quoted context omitted.

That's a solid theoretical argument. However, it is contradicted by the empirical evidence. What usually happens is that the police issue fines so frequently that they become considered an occupying force by the populace. In turn, the populace withholds information from the police, and both violent crimes and petty violations rise. Police statistics make quantitative conclusions difficult, since they are so often fud…

I would suggest, however, that this is not an argument for less police but instead an argument for fewer laws that are only infrequently and inconsistently enforced. It is these laws that give police the opportunity to "issue fines so frequently" and to give affected populations (rightfully or not) the impression that they are singled out. (Laws against loitering or the consumption of alcohol in public come into mind…

Right... but those laws only ever get revoked when they hit the rich with them. And for every one you revoke there's ten more wealthy/connected people with sob stories that could have been prevented if X were illegal who are contacting their elected representative.

I agree with you on principal but IMO society on Mars will be having this debate (I hope I'm wrong).

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#99

Post by yummyfajitas »

rlucas wrote: post #92
yummyfajitas wrote: post #71

Earlier quoted context omitted.

Sure, but there is plenty of historical data on people who marginally got mortgages. If those people are solidly profitable, as opposed to marginally profitable, you try a marginally more expansive lending criteria. I.e., if I know f(0) = 10, f(1) = 9, f(2) = 8, but I don't have data on f(3), it's worth running a bit of an experiment to see if f(3) is actually 7. (I happen to know from experience I can't talk about t…

You're right within the boundaries of what you're considering, meaning testing the profitability of marginal borrowers, etc. But you're missing the big picture question about the use of ML in making political decisions (and yes, the question of who gets mortgages is very much a political one). For example, in the earliest days of the creation of the US national mortgage market, conforming to FHA standards meant follo…

I'm aware that political processes surrounding loans were racially biased. That's not evidence that a profit maximizing algorithm will be. That's really just a claim that "this was once politicized, therefore it must remain political forever".

The whole benefit of using algorithms is that you make your goals explicit and non-opaque; you just look at the objective function. Opacity of the models isn't really a big deal - machine learning is fundamentally the mathematics of studying opaque models. It's the objective function that should be transparent.

If the objective function says "minimize defaults", that's your explicit goal. If the objective function says "maximize # of blacks getting loans", that's what you'll do. But it will do one or the other, unless you explicitly set your objective function to a x "minimize defaults" + b x "maximize blacks".

The tricky bit is that by doing this, we are explicitly and verifiably lowering the bar. Just witness how much flak people get for discussing this: https://techcrunch.com/2015/11/03/twitter-engineering-manage...

But unfortunately, an optimization system won't give you what you want unless you explicitly put what you want in the objective function, something our modern activists (probably including Cathy O'Neil) refuse to do.

Re: Mathematician Says Big Data Is Causing a ‘Silent Financial Crisis’

#100

Post by dragonwriter »

yummyfajitas wrote: post #97
dragonwriter wrote: post #91

Earlier quoted context omitted.

There is also "I am neutral to blacks and view them as, ceteris paribus , equally likely to pay back loans, but white people on average are not neutral toward them, and particularly are not neutral to living in close proximity to them, and the decrease in property values when black people are allowed in otherwise white neighborhoods makes both their default and the default of white people in the neighborhood more lik…

True, a model which predicts reality will in fact predict reality. I've never seen any evidence that your hypothesis is true, or even a model that would account for it, but I'm hardly an expert in the field. Assuming this is more than speculation/theoretical nitpicking, do you have citations on the effects you describe? If your theory is more than just idle speculation, this is a great way for non-racists to engage i…

> I've never seen any evidence that your hypothesis is true

If the key element of it wasn't true at one point, blockbusting would never have been a viable strategy. If you want to present arguments that race relations have changed in enough since then that whites no longer prefer living without blacks in the immediate vincinity so that such neighborhoods command a price premium and diversity causes price declines, I'm prepared to examine your evidence.

> If your theory is more than just idle speculation, this is a great way for non-racists to engage in house price arbitrage insofar as they wish to consume housing.

If you mean "white non-racists can get cheaper housing by voluntarily choosing to live where black people live", that's been a well-known way for non-racist whites to optimize home affordability since at least when my parents were young adults in the 1960s.

Of course, the downside in jurisdictions where there is a significant constituency of racist whites (which is, of course, any place where this strategy is particularly effective) is that non-racist whites then also get to experience the downside of the way racist whites distribute public services among different neighborhoods, which depending on personal utility functions may more than offset the value of the cost savings on the real estate itself. But, still, the strategy isn't new.

Post reply on HN