[Gibbon1]

The problem is the algorithms are designed to put people in a group and then the groups are treated differently. That's the mechanism underpinning racism. And other disparities. You take the current garbage and the algorithm spits it back out.

Current garbage looks like this. I know three people that got popped for drugs. One served time and has a felony record. One had it reduced to a misdemeanor and did 10 hours of community service. And the third has no record at all.

You can probably guess the socioeconomic status each of those three and you'd guess correctly. And I'm sure the algorithm will also guess correctly.

[ThA0x2]

>The problem is the algorithms are designed to put people in a group and then the groups are treated differently.

That's not the problem, that's literally the solution.

>That's the mechanism underpinning racism. And other disparities.

Not at all. The mechanism underpinning racism is believing that race accounts for differences in human character or ability and that a particular race is superior to others.

Disparities are not inherently racist.

>Current garbage looks like this. I know three people that got popped for drugs. One served time and has a felony record. One had it reduced to a misdemeanor and did 10 hours of community service. And the third has no record at all.

One had a long rap sheet and previously was in jail, one had a few priors, and the third had a clean record.

You can probably guess the socioeconomic status each of those three and you'd guess correctly.

We can both make up dumb anecdotes all day.

>And I'm sure the algorithm will also guess correctly.

The algorithm will correctly guess which of the three is more likely to flee, be a public threat, etc. Sounds like the algorithm is working.

[colejohnson66]

> Not at all. The mechanism underpinning racism is believing that race accounts for differences in human character or ability and that a particular race is superior to others. Disparities are not inherently racist.

No. Racism is discriminating on the basis of race. Hatred of a race is just a form of racism.

The definition of racism[0] includes:

> (n) Discrimination or prejudice based on race.

If I’m a hiring manager and I turn a black person down simply because “blacks are more likely to commit crimes”, it’s racism; I’m discriminating simply on the basis of a stereotype about a race.

Even if your stereotype is based on statistics, it can be racist. Not all black people commit crimes, so saying “because statistically, your race commits more crimes, we’re going to keep a closer eye on you,” is racism. The race doesn’t commit the crime; the person does.

Keeping a closer watch on a group of people because of race leads to statistically more reporting of crimes committed by that race even if it’s not true (less police presence leads to less crimes being reported and visa versa).

Not to mention that generalizing a single statistic to an entire race ignores every other factor that causes it. Systematic racism is a thing, and because of it, blacks are statistically poorer than their fellow whites. Being poor leads to people doing things that a richer person wouldn’t do, such as stealing because they literally have no money to pay for it. Drug habits can form and getting off them is hard when society shuns you because you’re “wasting your money on drugs instead of food.” But if you’re rich enough to afford food, drugs, and insurance,[a] one could go to a doctor for help.

Simply put, it’s very easy to be racist without realizing it; People like easy to digest facts. But being willfully ignorant about the whole story isn’t right.

[0]: https://www.wordnik.com/words/racism

[a]: I’m not going to even get into how abhorrent the idea of needing health insurance is, and how we fine people for not having it

[ThA0x2]

>No. Racism is discriminating on the basis of race. Hatred of a race is just a form of racism.

No, you are wrong and this is a straw man. A random definition of racism is not the mechanism underpinning racism.

>If I’m a hiring manager and I turn a black person down simply because “blacks are more likely to commit crimes”, it’s racism; I’m discriminating simply on the basis of a stereotype about a race.

No one said this is not racism, definitely not me.

>Even if your stereotype is based on statistics, it can be racist. Not all black people commit crimes, so saying “because statistically, your race commits more crimes, we’re going to keep a closer eye on you,” is racism. The race doesn’t commit the crime; the person does.

Not all men drive terribly and get in wrecks, so saying "because statistically, your gender wrecks more, we're going to charge you more", is sexism (which is illegal). The gender doesn't wreck cars; the person does.

Again, no one stated that race should be a category upon which these algorithms work. The fact of the matter is, even if the algorithm was 100% accurate and did not factor in race, there will still be inequity, which people will decry racist, even if it's 100% accurate and correct.

>Keeping a closer watch on a group of people because of race leads to statistically more reporting of crimes committed by that race even if it’s not true (less police presence leads to less crimes being reported and visa versa).

No one, definitely not me, has suggested that race should be a factor in these algorithms. Also, these groups are "closely watched" more because they commit more crimes. Police presence does not lead to people committing violent crimes like murder and rape more; it's the other way around. Police are in these communities more because they commit more crimes, and more crimes are reported there.

>Not to mention that generalizing a single statistic to an entire race ignores every other factor that causes it.

Strawman, strawman, strawman. No one has stated race should be a factor.

>Systematic racism is a thing, and because of it, blacks are statistically poorer than their fellow whites. Being poor leads to people doing things that a richer person wouldn’t do, such as stealing because they literally have no money to pay for it.

Systemic racism no longer exists. Indeed there are some inequities caused by previous systemic racism that was in place (Jim Crow), but these systemically racist systems no longer exist. There are more poor White people in total than poor Black people, but Black people commit more violent crimes in total. They are WAY overrepresented per capita. Being poor does not cause rape or murder. If it did, White people and poor Asians should have their proportional share, but they don't.

Poverty can explain things like shoplifting, petty theft. It does not explain rape, murder, etc.

>Drug habits can form and getting off them is hard when society shuns you because you’re “wasting your money on drugs instead of food.”

Non sequitor.

>Simply put, it’s very easy to be racist without realizing it; People like easy to digest facts. But being willfully ignorant about the whole story isn’t right.

Not really, no. Some easy to digest facts are just that, facts. They're inconvenient because they go against certain narratives, so they're immediately labelled "racist", which is the point I originally made. QED.

[CogitoCogito]

> Again, no one stated that race should be a category upon which these algorithms work.

Many variables that a priori have nothing to do with race (e.g. where you live, how much money you make, etc.) are highly correlated with race. Allowing those variables is basically the same thing as allowing race into the models. I don't want to pretend like controlling for something like this isn't possible, but I rarely see it considered (and quite rarely do I even see the problem acknowledged).

> Not really, no. Some easy to digest facts are just that, facts. They're inconvenient because they go against certain narratives, so they're immediately labelled "racist", which is the point I originally made.

What are examples these facts you refer to?

[ThA0x2]

>Many variables that a priori have nothing to do with race (e.g. where you live, how much money you make, etc.) are highly correlated with race. Allowing those variables is basically the same thing as allowing race into the models. I don't want to pretend like controlling for something like this isn't possible, but I rarely see it considered (and quite rarely do I even see the problem acknowledged).

They may be correlated with race, but correlation != causation. Poor people shoplift more than the well off, but that has nothing to do with race, it has to do with the culture.

>What are examples these facts you refer to?

Blacks commit sexual assaults and rapes at a way higher level than Whites and other minorities, even when accounting for income.

[CogitoCogito]

> They may be correlated with race, but correlation != causation. Poor people shoplift more than the well off, but that has nothing to do with race, it has to do with the culture.

What I said had nothing to do with correlation vs causation. What I said was that "variables that a priori have nothing to do with race (e.g. where you live, how much money you make, etc.) are highly correlated with race." I'm speaking of signals and information theory. The point is allowing variables that combine to be equivalent to allowing the variable of race ends up being the same thing.

Do you understand my point?

> Blacks commit sexual assaults and rapes at a way higher level than Whites and other minorities, even when accounting for income.

Presumably you mean that blacks are _convicted_ (or maybe arrested) for sexual assaults and rapes at a higher level (since obviously what they do are what they are convicted of doing are not the same thing).

Anyway alright fine I do believe that. Do you believe such a general fact has anything to do with individuals and their likely future actions? How? Why?

[Gibbon1]

> One had a long rap sheet and previously was in jail, one had a few priors, and the third had a clean record.

Nope. You lose.