18 ms·

Discriminate, but do it well

56 points by agronaut 11y ago

robotresearcher 11y ago
TLDR: use priors to help estimate robustly under uncertainty. Don't use incorrect priors.
- erikpukinskis 11y ago
  Help eliminate robustly?
- skj 11y ago
  Unfortunately, all priors are incorrect.
idlewords 11y ago
This is a survey article of sorts built on shaky foundations. The most relevant paper to read in relation to this story is this one: "Over half of psychology studies fail reproducibility test." http://www.nature.com/news/over-half-of-psychology-studies-fail-reproducibility-test-1.18248 http://www.nature.com/news/over-half-of-psychology-studies-f... By the time the results get distilled into a Guardian article, the situation is probably hopeless.
- losvedir 11y ago
  Economics, too: https://www.timeshighereducation.com/news/papers-in-economics-not-reproducible https://www.timeshighereducation.com/news/papers-in-economic...
tux3 11y ago
There's a trade-off between evaluating everything on a case-by-case basis (which takes time and energy) and using general rules of thumb that allow quick decisions making, but is essentially discrimination. I like to reflect on my own judgement before I actually need to use it, deciding which rules will be acceptable, what situations call for carefulness, what ideas I should be wary of even if I'm in a horrible mood. The subconscious is a wonderful thing, but sometimes it's nice to take charge of one's own thoughts a little bit.
- jessriedel 11y ago
  It's not just about time and effort. There are situations where we simply do not have access to complete information, no matter how much time and effort we spend gathering it.
- tracker1 11y ago
  All judgements in favor of one person over another are discrimination. The problem is when discrimanation is for immaterial factors based on racial/gender attributes over those relevant to the environment. In some ways a person's culture can play a large role vs. physical attributes in a given work environment. I think it comes down to making at least an effort to treat everyone fairly and not making broad assumptions.
meowface 11y ago
Completely off-topic, but is the quote at the top grammatically correct? The comma looks unnecessary. I feel like it should be "We must make snap judgments on the basis of various shortcuts, or we’d be unable to function", not "We must make snap judgments, on the basis of various shortcuts, or we’d be unable to function". Is this just an error, or was the comma added to more closely reflect how the individual spoke?
- kaybe 11y ago
  It changes the meaning, I think. In the second case, it means that snap judgements must be made to function, and they're, incidentally, based on shortcuts (but you can leave out this part of the sentence), while in the first case 'on the basis of various shortcuts' is a very integral part of the information, and snap judgements must be made this way.
  - meowface 11y ago
    Good point. I think I agree.
- ScottBurson 11y ago
  I don't think I would have put a comma there, but I can't bring myself to actively object to its presence.
mswen 11y ago
In some ways statistical modeling is trying to do with rigor and honesty about error what we seem to do quite naturally in daily life. That is we make decisions based on observed and/or shared and taught patterns. Stereotypes, prejudice and discrimination are the crude, less substantiated and often mean-spirited cousins of statistical models. Models are useful as long as their limitations are well understood, and humans use them with caution and compassion.
- beccasanchez 11y ago
  Modern statistics and statistically modelling are actually just confirming the validity of the majority of widely held stereotypes. As statistics in the academy become more common and advanced, more and more old stereotypes are just confirmed to be accurate. This is considered a crisis right now in social psychology, for instance, because there are almost no stereotypes that research has debunked. Again and again, research confirms stereotypes, and shows that humans in the aggregate are very good at creating accurate inferences. Stereotypes create injustice when they lead to prejudice--failing to give an individual a chance because they belong to a stereotyped group--but prejudice is wrong because outliers deserve a chance, not because stereotypes are incorrect. How can liberalism persist in the face of so much science that dismantles its most sacred beliefs?
  - nommm-nommm 11y ago
    >there are almost no stereotypes that research has debunked. Again and again, research confirms stereotypes, and shows that humans in the aggregate are very good at creating accurate inferences. I'm calling BS on this unless you give some specific examples. Not only this but confirming stereotypes does not mean that "humans in the aggregate are very good at creating accurate inferences." For example if (group) have the stereotype of being dumb that doesn't mean that others are good at making accurate inferences, it may just mean that others are good at denying (group) access to education and the same standard of living.
    - beccasanchez 11y ago
mwsherman 11y ago
Biases are turtles all the way down, one learns. The only “solution” I’ve found for myself is not to “correct” them, but to have a lot less conviction about my own correctness. Big error bars.
- duaneb 11y ago
  Well said.
oneplease 11y ago
I'm not sure I get the point of this article. On the teacher example, of course we discriminate based on skill, no one's disputing that. I could not find a deeper point than discriminate by legitimate metrics but not by skin color, race, etc which is fairly obvious.
- Afforess 11y ago
  The point of the article is that using outrage and shame culture to resolve discrimination can be just as bad as acts of discrimination themselves. The act of shaming a person can make you feel you have done your 'good deed for the day', and allow you to unconsciously commit your own biases against others. The article suggests that instead of shaming or attacking biases we observe in others, we should reflect on ourselves first, and fix our own misdeeds and mistakes. The entire article could be summarized from a verse in the (NKJV) Bible: Why do you look at the speck in your brother's eye, but fail to notice the beam in your own eye?
  - trowawee 11y ago
    Because my brother is actually being a racist, which is a beam in his eye, whereas I am calling him out for being a racist, which, if we accept that there's something wrong with "outrage and shame culture" (which, tbh, I mostly see used interchangeably with "PC" as a catchall putdown by people upset that they can't just talk shit about minorities and women without consequence like they used to be able to do in the good ol' days), is a speck in my eye.
    - VanillaLime 11y ago
      Is that really what you got from the article? I feel like the author makes a pretty valid point: everyone pattern-matches people at first impression. If you decide to climb on a tower of moral virtue because you think you've consciously decided to stop pattern-matching on the basis of race/gender and thus don't introspect about your (many) other subconscious biases, you're really not doing anyone a favor.
brenschluss 11y ago
I read the title, saw the photo of the author, and made a snap judgment that the article was going to be a pseudo-intellectual, covertly conservative argument, such as: "Why race is actually a thing, sometimes" or "Why gender pay imbalances may be sometimes helpful" - and my snap judgment was right.
- klipt 11y ago
  What's funny about discrimination is some is overt, but much more is covert. Overt is mostly frowned upon, although e.g. car insurance companies still openly charge men more. I'm pretty sure that's only allowed because men aren't the "protected gender" when it comes to discrimination.
  - nommm-nommm 11y ago
    The term you are looking for is protected class. When gender based discrimination is outlawed men are included as well. That being said Montana outlaws gender based discrimination in auto insurance as does the EU. Where its allowed it is allowed because nobody got around to outlawing it. That can change. Until Obamacare many US states allowed health insurance companies to charge women higher premiums because women go to the doctor more. Obamacare outlawed that. I think gender discrimination is auto insurance will end eventually.
asdfzxc 11y ago
That article didn't say shit. It just asked me to go read a book..

56 points by agronaut 11y ago

robotresearcher 11y ago

TLDR: use priors to help estimate robustly under uncertainty. Don't use incorrect priors.

erikpukinskis 11y ago
Help eliminate robustly?
skj 11y ago
Unfortunately, all priors are incorrect.

idlewords 11y ago

This is a survey article of sorts built on shaky foundations. The most relevant paper to read in relation to this story is this one: "Over half of psychology studies fail reproducibility test." http://www.nature.com/news/over-half-of-psychology-studies-fail-reproducibility-test-1.18248 http://www.nature.com/news/over-half-of-psychology-studies-f... By the time the results get distilled into a Guardian article, the situation is probably hopeless.

losvedir 11y ago
Economics, too: https://www.timeshighereducation.com/news/papers-in-economics-not-reproducible https://www.timeshighereducation.com/news/papers-in-economic...

tux3 11y ago

There's a trade-off between evaluating everything on a case-by-case basis (which takes time and energy) and using general rules of thumb that allow quick decisions making, but is essentially discrimination. I like to reflect on my own judgement before I actually need to use it, deciding which rules will be acceptable, what situations call for carefulness, what ideas I should be wary of even if I'm in a horrible mood. The subconscious is a wonderful thing, but sometimes it's nice to take charge of one's own thoughts a little bit.

jessriedel 11y ago
It's not just about time and effort. There are situations where we simply do not have access to complete information, no matter how much time and effort we spend gathering it.
tracker1 11y ago
All judgements in favor of one person over another are discrimination. The problem is when discrimanation is for immaterial factors based on racial/gender attributes over those relevant to the environment. In some ways a person's culture can play a large role vs. physical attributes in a given work environment. I think it comes down to making at least an effort to treat everyone fairly and not making broad assumptions.

meowface 11y ago

Completely off-topic, but is the quote at the top grammatically correct? The comma looks unnecessary. I feel like it should be "We must make snap judgments on the basis of various shortcuts, or we’d be unable to function", not "We must make snap judgments, on the basis of various shortcuts, or we’d be unable to function". Is this just an error, or was the comma added to more closely reflect how the individual spoke?

kaybe 11y ago
It changes the meaning, I think. In the second case, it means that snap judgements must be made to function, and they're, incidentally, based on shortcuts (but you can leave out this part of the sentence), while in the first case 'on the basis of various shortcuts' is a very integral part of the information, and snap judgements must be made this way.
- meowface 11y ago
  Good point. I think I agree.
ScottBurson 11y ago
I don't think I would have put a comma there, but I can't bring myself to actively object to its presence.

mswen 11y ago

In some ways statistical modeling is trying to do with rigor and honesty about error what we seem to do quite naturally in daily life. That is we make decisions based on observed and/or shared and taught patterns. Stereotypes, prejudice and discrimination are the crude, less substantiated and often mean-spirited cousins of statistical models. Models are useful as long as their limitations are well understood, and humans use them with caution and compassion.

beccasanchez 11y ago
Modern statistics and statistically modelling are actually just confirming the validity of the majority of widely held stereotypes. As statistics in the academy become more common and advanced, more and more old stereotypes are just confirmed to be accurate. This is considered a crisis right now in social psychology, for instance, because there are almost no stereotypes that research has debunked. Again and again, research confirms stereotypes, and shows that humans in the aggregate are very good at creating accurate inferences. Stereotypes create injustice when they lead to prejudice--failing to give an individual a chance because they belong to a stereotyped group--but prejudice is wrong because outliers deserve a chance, not because stereotypes are incorrect. How can liberalism persist in the face of so much science that dismantles its most sacred beliefs?
- nommm-nommm 11y ago
  >there are almost no stereotypes that research has debunked. Again and again, research confirms stereotypes, and shows that humans in the aggregate are very good at creating accurate inferences. I'm calling BS on this unless you give some specific examples. Not only this but confirming stereotypes does not mean that "humans in the aggregate are very good at creating accurate inferences." For example if (group) have the stereotype of being dumb that doesn't mean that others are good at making accurate inferences, it may just mean that others are good at denying (group) access to education and the same standard of living.
  - beccasanchez 11y ago

mwsherman 11y ago

Biases are turtles all the way down, one learns. The only “solution” I’ve found for myself is not to “correct” them, but to have a lot less conviction about my own correctness. Big error bars.

duaneb 11y ago
Well said.

oneplease 11y ago

I'm not sure I get the point of this article. On the teacher example, of course we discriminate based on skill, no one's disputing that. I could not find a deeper point than discriminate by legitimate metrics but not by skin color, race, etc which is fairly obvious.

Afforess 11y ago
The point of the article is that using outrage and shame culture to resolve discrimination can be just as bad as acts of discrimination themselves. The act of shaming a person can make you feel you have done your 'good deed for the day', and allow you to unconsciously commit your own biases against others. The article suggests that instead of shaming or attacking biases we observe in others, we should reflect on ourselves first, and fix our own misdeeds and mistakes. The entire article could be summarized from a verse in the (NKJV) Bible: Why do you look at the speck in your brother's eye, but fail to notice the beam in your own eye?
- trowawee 11y ago
  Because my brother is actually being a racist, which is a beam in his eye, whereas I am calling him out for being a racist, which, if we accept that there's something wrong with "outrage and shame culture" (which, tbh, I mostly see used interchangeably with "PC" as a catchall putdown by people upset that they can't just talk shit about minorities and women without consequence like they used to be able to do in the good ol' days), is a speck in my eye.
  - VanillaLime 11y ago
    Is that really what you got from the article? I feel like the author makes a pretty valid point: everyone pattern-matches people at first impression. If you decide to climb on a tower of moral virtue because you think you've consciously decided to stop pattern-matching on the basis of race/gender and thus don't introspect about your (many) other subconscious biases, you're really not doing anyone a favor.

brenschluss 11y ago

I read the title, saw the photo of the author, and made a snap judgment that the article was going to be a pseudo-intellectual, covertly conservative argument, such as: "Why race is actually a thing, sometimes" or "Why gender pay imbalances may be sometimes helpful" - and my snap judgment was right.

klipt 11y ago
What's funny about discrimination is some is overt, but much more is covert. Overt is mostly frowned upon, although e.g. car insurance companies still openly charge men more. I'm pretty sure that's only allowed because men aren't the "protected gender" when it comes to discrimination.
- nommm-nommm 11y ago
  The term you are looking for is protected class. When gender based discrimination is outlawed men are included as well. That being said Montana outlaws gender based discrimination in auto insurance as does the EU. Where its allowed it is allowed because nobody got around to outlawing it. That can change. Until Obamacare many US states allowed health insurance companies to charge women higher premiums because women go to the doctor more. Obamacare outlawed that. I think gender discrimination is auto insurance will end eventually.

asdfzxc 11y ago

That article didn't say shit. It just asked me to go read a book..