How Discrimination Works (Copied From Chat)
I mean, I think racial discrimination and profiling as they now exist are mostly a side effect of living in a huge, atomised population where you can’t keep track of everyone you’re going to meet individually.
So, different groups of people are going to differ on averages for how they act - whether because of culture or incentives or w/e - but this doesn’t affect how you behave toward them under a certain number of people.
You don’t care about averages when you know every life detail of individuals.
Like, imagine you’re hosting a dinner party and you have 15 friends. Based on your years of interaction with them, you know Dave and Maria are going to start fights with the other participants, so you don’t invite them.
It doesn’t matter that Dave is a white guy or Maria is a Hispanic woman. You know them personally and, from knowing them, can predict what they, specifically, will do.
At this level, the demographic characteristics of people don’t matter /at all/. You’re using actual, individual knowledge to discriminate on who’s allowed at your event.
This is also the situation with the least false positives and negatives. Discriminating against groups will always leave you with a lot of false results because, while the groups may differ on averages, there’ll always be people in the tails you wanted to keep or exclude who you couldn’t.
So, why would anyone use a courser, looser, stupider proxy for figuring out if someone is actually an asshole? Well, costs.
I am told that people who aren’t me often have a hard time cultivating deep friendships with a hundred people which allow them to judge the character of every soul. IDK /why/ you guys aren’t on my level and think you should git gud, but I am accepting of your frailty.
But, anyway, that means the costs of personally tracking that many people are too high for you to discriminate on character. Instead, you want to discriminate on something that /correlates strongly/ with character.
It’s only a proxy, but it’ll still let you keep out a lot of the bad folks and keep in a lot of the good folks. Hopefully.
So, say you run a pub. You have a few regulars that you know things about and who’s character you can vouch for. Sometimes one of them has a bad day and yells at someone, but you know that this is a rare occurrence, so you just give them a stern talking to and leave them be.
The next day, a rando comes in. He’s had a bad day and he yells at someone. You ban him from the pub.
Why? Because /most/ people who randomly start yelling at other patrons are assholes. In the case of your regulars, you know them well enough to be aware that they aren’t personality-level assholes, even if they act like assholes once in a while.
Meanwhile, you have no way of knowing that this rando off the street is a firefighter with three cats and a heart of gold who just learned that his mother died.
You just don’t have that background because you can’t pay that much attention to that many people. Kicking anyone who becomes verbally abusive is a rule that /usually/ works, so it’s worth the costs when it fails. The rules are the rules (if you have nothing better than rules to work with).
This also means that a group having explicit rules is a sign that its members don’t know each other well enough or can’t agree well enough that they don’t need coarse-grained methods like these. Friend groups rarely have written rules; corporations usually do.
As you ascend the ladder of number-of-people-to-deal-with, you notice that the main constraint is how much time and attention you can spend on each person.
The dinner party host could spend hundreds of hours on each guest over the course of their friendships. The pub owner may have been able to watch the rando customer for half an hour before he started making a ruckus. What happens if you have five seconds?
Imagine you’re a bouncer for a club and have to decide who to let in and who to keep out. You’re the most important gatekeeper, so it’s on you to keep out as many assholes as possible - but you have to wave people in or tell them to fuck off /quickly/.
So, with just a few seconds to make your decision, you have to default to an even more visible proxy than the pub owner did.
Unfortunately, as proxies become more visible, they also tend to work less well, so you’ll be even more imperfect than before.
As you’re only human and have to work fast, you look people up and down and judge their hair and face and clothes. You form an intuitive impression of whether this person looks like The Right Sort.
Of course, this is going to be somewhat divergent from who is and isn’t an asshole and /notably/ correlated with social class. Whoops. But, well, you had five seconds, dude. What did you expect?
However, it’d be even /worse/ if you had effectively no seconds. When you get big enough for a bureaucracy with forms to fill out, you can’t even /look/ at people. You can only discriminate based on features someone can write down on a form.
So you exclude all the black men on the assumption that they’ll start fights. This is /ridiculously/ lossy. This is a false positive extravaganza. But, above a certain size, this is all you’re gonna get.
So, the obvious response to arriving at level 3 or 4 and realising that this profiling thing is out of control is to ban discrimination. However, I think that misses important facts about the situation illustrated above.
The first is that the discrimination is serving a purpose. Sometimes that’s avoiding fight-starters, but it could really be anything. It doesn’t matter much /why/ people value it, just that they do so strongly enough that they clearly put a lot of effort into it.
But something that falls out of that is that /people don’t stop/. As the proxies for discrimination become more stupid and useless, people /keep using them/.
They seem to be pretty desperate to find some way to filter who they let into their homes and pubs and clubs, so you should expect them to be willing to swallow a lot of costs to keep doing so.
So, when you start banning discrimination, what actually happens? What do people do?
Well, they seem to find the one method of discrimination that hasn’t been banned and can’t be banned: Price.
Most notably, in urban housing markets, zoning laws that make construction difficult and drive up prices until the poors move out. They aren’t a bug; they’re a feature.
“Has enough money to buy a house in San Francisco” is a very lossy proxy for someone’s propensity to start fights, get drunk, break windows, or otherwise be a bad neighbour. But there are too many people to filter to use a proxy that /isn’t/ super lossy.
Of course, this is a filter that’s way more costly than the others, because it comes with a massive deadweight loss - you’re paying a $300k premium to not live around poor people, with “poor people” being itself a proxy for the actual set of people you want to avoid. It’s not enough to /say/ you have this much money - you have to /actually pay it down/.
And that’s why I basically don’t think the Bay Area’s housing crisis is going away, unless there’s another white flight and suburbanisation becomes a thing again, which would /also/ be shit.
Like, there are people who want to go back to a time when houses weren’t ridiculously expensive, so they want zoning to go away. Meanwhile, there are people who want to go back to a time when cities didn’t have no-go areas, so they want to Zone All The Things.
I think that, as long as price discrimination is the order of the day, this tug of war isn’t going away.
No, I don’t have suggestions. I’m just the person who waves at problems suggestively and hopes someone else will solve them.
I can say that allowing people to discriminate on wealth-in-the-bank as its own metric would lead to basically the same distribution of “rich people /here/; poor people /there/” that we currently have without the massive deadweight losses.
But that’s still “making the current situation suck a little less”. We still have the problem of how all discrimination got super coarse-grained. If we could get as fine grained as the dinner party host, then we could bring salvation to the realm.
But, until then, we’re still haggling over ways to become marginally less fucked.
- proxy-service reblogged this from sinesalvatorem
- injygo liked this
- maximusrazel liked this
- trystimuli liked this
- aconiteherbalist reblogged this from sinesalvatorem
- neoliberalism-nightly liked this
- the-grey-tribe reblogged this from sinesalvatorem
- ageesen liked this
- imultifandomroleplay liked this
- lithnin reblogged this from sinesalvatorem
- hill-climber liked this
- kaminiwa liked this
- rangi42 liked this
- superroy liked this
- theworldisasecondarycarbon liked this
- falsestatistics liked this
- jwroen liked this
- mugasofer reblogged this from sinesalvatorem
- inositoltimemachine reblogged this from oktavia-von-gwwcendorff and added:
Interesting idea. It does seem vulnerable to people deliberately getting in arguments to run up the other person’s costs...
- shitifindon liked this
- mirrored-mazes liked this
- inspirehellfire liked this
- smart-elec liked this
- mitigatedchaos liked this
- neshselg liked this
- rareandradiant-maiden liked this
- type12error liked this
- 23rdhunter liked this
- abbygine liked this
- phenoct liked this
- bannableoffense liked this
- discoursedrome liked this
- oktavia-von-gwwcendorff reblogged this from sinesalvatorem and added:
For it to succeed, you only need, in theory, to do better than the existing system.The existing system, which results in...
- letinout liked this
- unfathomably-complex liked this
- urh2o2fantasy liked this
- nonevahed liked this
- yesharrypotterlover123blr reblogged this from sinesalvatorem and added:
The reason this probably doesn’t occur may well be due to morality. Crime insurance is efficient and non-judgemental,...
- radioactivecallista liked this
- poipoipoi-2016 reblogged this from sinesalvatorem and added:
So I agree with pretty much all of this. But two minor quibbles here. 1) The highest property crime rate in the US is...
- sinesalvatorem reblogged this from oktavia-von-gwwcendorff and added:
Two problems are how much surveillance you need to get in the right ballpark of who to charge (eg: knowing who stole...
- skeletontemple reblogged this from sinesalvatorem
- drethelin reblogged this from scientiststhesis
- drethelin liked this
- ilzolende liked this
- sabrinatristan liked this
- reasonableapproximation liked this
- alexanderrm liked this
- spiralingintocontrol liked this
- kelsbraintumbler liked this
- Show more notes