HUMAN 1
I am assuming that the computer program did not factor in race at all and instead focused on actual circumstances which black people just tend to be worse off in. Then ‘ProPublica’ got outraged when doing simple data mining because working to actually resolve and understand issues has been well outside of the American left’s wheelhouse for decades.
HUMAN 2
The algorithm is literally biased
Yeah, you can say you have a completely race-blind algorithm, but if it’s blind to racism impacting the data, it’s going to have a result that suffers from racism as well.
For example, asking, ‘Was one of your parents ever sent to jail or prison?’ is really closely akin to asking whether someone’s grandparent was a slave before forcing them to pay a poll tax or take a literacy test. The question may not be inherently racist, but the question it’s asking is addressing a reality that was racist and affected people disproportionately.
If you’re white, your parents are less likely to have been arrested by police for smoking weed in the 1970s. If you’re white, you’re more likely to have gotten off with a warning when you got in a fight in high school than prosecuted for a felony. If you’re white, you probably have a social network that can provide you with a job more easily because your family, friends, neighborhood, and classmates were allowed to inherit and increase their wealth.
An algorithm that perpetuates systemic biases probably is not a well-designed one.
The mathematicians who want to save democracy – With algorithms in hand, scientists are looking to make elections in the United States more representative.
HUMAN 1
Maximum Compact Districts, such that the average compactness is as high as possible. It’s a pure mathematical problem, and for any given territory / state / whatever to divide, there is one way to do it. Nothing a modern computer can’t solve in short order. Easy, simple, non-manipulable.It will never happen.
HUMAN 2
That doesn’t really improve representation though. If anything, minorities would have fewer representatives, since a legitimate reason for drawing oddly shaped districts is to gather minorities in several different areas so they can have a representative. It’s a side effect of having a very large representative to vote ratio. Decrease the number of votes per representative e and it becomes less of a problem
HUMAN 1
Once you start trying to factor stuff like that in, the game is lost. If you gerrymander for one reason, no matter how noble, 3000 other “good” reasons will line up, and then you have maps that are messed up like today, which is the fundamental problem we are trying to solve.
Yeah, but if you do ‘race blind’ districts, that doesn’t mean anything except that the districts are blind to everything involving race, including racism and the continuing effects of racist policies.
This goes back to the issue of why ‘I don’t see race’ is bad as an outlook. Not only do lots of other actors obviously continue to see race and try to actively extend racism in new or disguised forms, but also the legacy of racism continues by sheer inertia.
So if you’re not looking at how segregation by law and redlining by policy has affected where people could live and still does affect people, you’re allowing it to perpetuate frictionlessly.
Even if tomorrow we were able to wake up and everyone was literally race-blind, you’d still have the effects of government-endorsed wealth theft disproportionately affecting people previously identified as a minority group that would impact race-blind measures of who deserves a loan, the quality of education (based on property values and therefore revenue from taxes), and so on.
It’s definitely not simple, but just because your algorithm (or law) doesn’t identify a particular factor doesn’t prevent it from influencing the results.
If instead of geographically compact districts, your goal was districts as close to the median household income as possible, that might result in a Democratic bias. Or if instead of voter ID, you wanted to protect the integrity of elections by having voting locations in population areas no less than 250,000 people, or no less than 5,000 people per square mile, that’s going to result in a Democratic bias by making it less likely for rural voters to participate because they just happen to be older, whiter, and more conservative than the typical voter.
Putting your fingers in your ears doesn’t stop the sound, just your ability to hear it.
HUMAN 1
And if you do “race aware” districts, here come the 3000 other issues that want you to make sure the districts are likewise “aware” to their needs and interests. And you have a mess. Like today.You want to gerrymander for nobel reasons. But there is no way to keep it limited to such nobel reasons, even assuming we could agree on what “nobel” is. You are giving up the entire game at the first step.
“Compactness” is already a principle that we draw against, so I am not changing the principle, just maximizing the principle.
Gerrymandering is having districts drawn for political ends, to benefit a political party. Therefore, you could have an algorithm maximize for compactness and spit out any number of districts that are perfectly compact but through randomness benefit one party versus another, and through political maneuvering be chosen by one party against another.
if you read the article, that’s one of the problems with compactness. There’s lots of ways to define it, and it’s almost certainly going to be used to the detriment of the representation of cities due to cities already being dense and compact. You can cheat sparse, rural areas much easier.
The approach by Nicholas Stephanopoulos at the University of Chicago, aimed apparently at getting Justice Anthony Kennedy on board seems like a pretty good one to me. I don’t really care what your order of operations is if the result is getting a sum of parts different than the whole.
North Carolina could have been less stupid and obvious about trying to disenfranchise black people, but if you’re turning a 50-50 preference statewide into a 70-30 preference in your legislature, obviously there are shenanigans there.
HUMAN 3
Can you give an example of another interest that hypothetically could be taken into consideration like ethnicity is being?
Median wealth would be a good one. Age. Native born population. Religious preference.
There’s lots of things you could look at or try to factor in, fairly or unfairly. But the interest is demonstrated to be gerrymandering when it uses First Past the Post to waste the votes of the people whose party is out of power.
HUMAN 4
And if slavery never happened the districts would be different still. And if the British didnt attack even more different. Race blind is the best we can do if you want equal opportunity moving forward. Any deviation from that opens Pandora’s box to a million other factions and issues that groups of different people will never agree on.
Right, if the United States, and one region in particular, didn’t have a history of actively disenfranchising people, then you probably wouldn’t need to consider that history when looking at voting districts.
But as it turns out, when you say, ‘It’s been a few years. We can probably let the old Confederacy off the hook when it comes to getting their voting districts drawn and the standards for who and how people can vote,’ they do have a tendency to go right back to disenfranchising the sorts of people they continue not to like very much.
HUMAN 4
Sure if by “one region in particular” you mean nearly half of the entire country at the time. Disenfranchisement can happen for many reasons. Allowing voting districts based on race doesnt automatically give them a voice. Maybe more integration would be useful, so race wasnt as large of a factor in districting.
In North Carolina and Texas, race was the central factor in districting, as courts have consistently found. The only argument the states have made is they were trying to target Democrats and race was the best proxy.
This problem isn’t limited to the South. Wisconsin did similar work targeting early voting and creating ID restrictions. But again, they said, ‘Hey we’re doing this thing over here totally unrelated to diminishing minority voter power over there. Oh look! This thing has a hugely disproportionate effect on them in just that way. What a surprising coincidence.’
Voting districts based solely on race make Packing easier, so in that way, they’re a problem. But it’s definitely one of the important checks that should be done when examining the effects of voter restriction measures because those funny coincidences pop up again and again.
HUMAN 5
Do you really believe “if tomorrow we were able to wake up and everyone was literally race-blind” then we would still have “problems” from racism?It is as though you yourself can’t look at the world and judge people except if you know the colour of their skin.
Economists can measure inequality, and it’s measured in a race-blind way, through the gini coefficient. It doesn’t matter what people look like as to how we measure it. The most important consequence of this is that we can see the effects of certain policies on inequality even if we are completely race-blind – meaning we can make completely rational decisions, even if we are completely race-blind.
From a race-blind perspective, racist policies look like policies which are ineffective, disadvantaging people seemingly at random – so we would still oppose them.
‘Do you really believe “if tomorrow we were able to wake up and everyone was literally race-blind” then we would still have “problems” from racism?’
Yeah. For example, if Harvard admissions didn’t factor in race at all, you still have problems of who can afford it and who can get in through legacy admissions and who knows influential people personally who can help out. That all remains when the other conceptions of race disappear, and then they perpetuate themselves.
If you’re saying, ‘Let’s measure inequality in a race-blind way and then invest in things that disproportionately benefit the least wealthy’, OK. I’m down for that. But then you’re essentially talking about reparations. And by the way, I think that’s a really good practical approach but you get there a lot more quickly by looking at the historical and current forms of discrimination and correcting for them.
Like, one of the problems a lot of minority neighborhoods have is a lack of access to banks for savings accounts or check cashing. A good solution to that would be using Post Offices as a federal banking system. That idea can be framed as race-blind, but the effect will be disproportionately felt among people who have been victims of redlining in the past. Etc.
So maybe we disagree on some semantics here, but I’m saying the transformation from injustice to justice isn’t simply stopping the injustice. You have healed someone by virtue of stopping your beating of them. You have to actively do the things that get them better.
I think that’s best done by looking at where you wounded them. You’re saying there may be other people with injuries, but in America, I think the determination of the triage is going to be pretty similar regardless of what process we’re using.
HUMAN 5
“If you’re saying, ‘Let’s measure inequality in a race-blind way and then invest in things that disproportionately benefit the least wealthy’ ”This is what I’m saying, so I think we do agree mostly, but I also think that it’s important to avoid (what could be percieved as) racist policy that benefits people even if they’re rich and hurts them even if they’re poor, based on their race which is only a proxy for inequality. That is, ideally, the laws themselves should be written in a race-neutral way.
Of course, you end up massively benefitting minorities with these laws, but you also help a few other underprivileged people as a side-effect.
‘it’s important to avoid (what could be perceived as) racist policy that benefits people even if they’re rich and hurts them even if they’re poor, based on their race which is only a proxy for inequality.’
Yes, but the United States has literally never had the problem of helping black people too much and often had the problem of policies specifically designed to disadvantage them or avoid advantaging them. Look at the GI Bill or Social Security being originally designed to exclude black Americans, or how minimum wage carved out exceptions for tipped laborers.
We agree that Will Smith’s family and Barack Obama’s family have persevered and achieved success and no longer need help for certain things, especially wealth-based. But Chris Rock still gets pulled over for driving a nice car. Thabo Sefolosha still gets his leg broken for sassing police. James Blake still gets tackled for standing while black.
For a very long time, law and policy were specifically written in a racist way, and still today policy is enacted in a racist way. If your car pulls a bit to the right and you want to keep it going straight, it’s not enough to keep your wheel straight. You’ve got to turn the wheel back the other way.
The help a person gets shouldn’t be dependent on their race, but if you aren’t looking at historically racist policies and directly addressing them in the opposite way, you’re going to be much less effective.
That is, it’s not enough going forward to just pretend the past doesn’t exist.
ADJ 