I read a description -- sorry, don't remember where -- that finally made me feel like what happened made sense. Though who knows if any outsider will ever understand what happened.
Reading between the lines, Gardner basically claimed Angwin -- who was the EIC -- didn't do the job. Angwin wanted to do journalism, but that's not the EIC job. Her job was to hire people to do journalism, and to set up the (prioritization, themes, etc) in which journalism was to be done. Angwin did that poorly. [1] Her job also was to set up technical infra. It was claimed she didn't. etc.
From another article, "Angwin acknowledged to CJR that 'meetings are not my favorite thing.'" [2] . That makes me believe the firing of Angwin was justified; the EIC job is literally meetings.
This is a pattern that is repeated in startups: a technical founder often doesn't understand the job is not to be the best engineer on the team, but rather to hire managers, set priorities, coordinate with the rest of the business, etc.
And who gets to decide what "the objective state of the world" is?
There is real value in reading about the same topic from differing perspectives. Not all writing is or aspires to be purely fact-based journalism, and that's okay.
It's reasonable that The Markup wants to hire group of people with diverse viewpoints on how "powerful institutions are using technology in ways that impact people and society". I would be surprised and underwhelmed if they had e.g. only software engineers or only consumer rights activists writing.
The issue is their assumption that only non-white, non-males have interesting things to say about how "powerful institutions are using technology in ways that impact people and society".
I mean, I'd argue that journalism as it should be is focused on truthfulness, accuracy, objectivity, impartiality, fairness and public accountability - all standards one would assume are important given The Markup's stated mission - but your point about not all writing aspiring to be fact-based is well taken.
Why has it become acceptable to assume so much about someone's view of the world based on their race, gender and sexual identity? Why has our definition of "diversity" become so narrow and stereotypical?
If you want to hire people with diverse viewpoints, then look for people with diverse viewpoints. It's incredibly easy to fill a room with people who look like a Benetton poster but who all read the same books in the same colleges and all rigidly adhere to the same dogma. It's incredibly easy to use tokenism to justify your own prejudices. It's difficult to build a genuinely diverse collaboration based on profound and meaningful differences of perspective.
> Why has it become acceptable to assume so much about someone's view of the world based on their race, gender and sexual identity? Why has our definition of "diversity" become so narrow and stereotypical?
US Supreme Court jurisprudence more or less. No one actually cares about diversity as a goal. The political coalition that passed the Civil Rights Act looked around, saw they’d won and wanted more stuff for their constituents. Affirmative action was one of the ways to get more stuff. It’s also illegal under a plain reading of equal rights. For reasons I’m unqualified to opine on a quota system is too blatant legally (also explicit points based systems with group X getting Y extra points) so the Supreme Court went with diversity as an acceptable workaround to get to the desired goal of getting more stuff for the desired interest groups. This all happened in the context of university admissions and the buzzword spread from the case law.
Same as free speech or state’s rights people who genuinely believe in them are rare compared to those who use them as talking points when convenient and drop them when not.
Yeah, no one actually cares difficult, deeply rooted social and cultural problems that affect millions of people. It's clearly just about getting more stuff.
Yes, that’s broadly right. The number of principled ideologues is always very low compared to those who see that an ideology has something for them and then follow it. I’m not suggesting these people are self-aware hypocrites that’s just how human psychology works.
In under a generation people go from arguing that everyone deserves equal rights and dignity to campaigning for explicit racial privileges for their side, or from supporting the right of the federal government to seize slaves who escape and travel to a free state to yakking about state’s rights to justify Jim Crow, or the one we’re in the trailing end of right now, the transition from free speech being a left wing value to a right wing one. When the right had the establishment the left was all about people’s right to speak and be heard. Now that they’re in charge it’s safe spaces and hate speech. Or look at the collapse of the anti-war movement once Obama was elected. Bush bombed Muslims for eight years, Obama bombed Muslims for eight years. Bush destroyed Iraq and had no real effect in Afghanistan as it comes pre destroyed. Obama destroyer Libya and cape quite close with Syria before his good sense overcame the presence of Hilary Clinton in his cabinet.
Politics is overwhelmingly about sides, coalitions and spoils. Insofar as it’s about ideology most people have 1-3 things they really care about and the rest they’re pretty happy to modify to fit in with their coalition.
...the two categories I mentioned -- software developer and consumer rights activists -- are not race, gender, or sexual identity...
In fact, the whole thrust of my comment was that it's sometimes desirable to have "a genuinely diverse collaboration based on profound and meaningful differences of perspective", and the last paragraph of my post calls out the "nice looking poster" approach.
What point are you trying to make? It sounds to me like we're largely agreeing.
Whether a bias is good or bad depends on the observer. If you agree with it, it's good; otherwise it's not.
The parent post expressed hope they were hiring writers who would 'report on the objective state of the world, not pursue a particular perspective'. I was pointing out that these are not opposites.
We often think of 'objective' as the opposite of 'biased', but it isn't: it's substituting a personal bias with an impersonal one. This certainly has it's place - it's critical in the sciences - but objectivity has limits when you're dealing with people.
A key strength for any good journalist is the ability to contextualize a story for their readers - not just provide objective facts, but answer the question "What does this mean to me?" A journalists' background and personal perspective - their bias - are relevant because they inform that context.
yes, there is much confusion in this area. IMO values determine what is considered relevant for any description of reality - values that may be ubiquitous (in which case the description is 'objective'), or personal/limited to a few (in which case the description is 'biased').
Or so my amateurish thinking goes. I'd be much interested in a more learned perspective on this, but haven't really found anything yet.
Seems pretty clear cut that Sue Gardner was making hiring decisions based on practices prohibited by the U.S. Equal Employment Opportunity Commission. The CJR article says she sent them the spreadsheet herself...
One thing wasn't clear to me from this article: did being non-white or non-male improve or hurt a prospective employees' score? I'm assuming in this case being 'non' raised the score, but the article doesn't state one way or another...
Myers-Briggs personality test?! Isn't that like horoscope? The last time I took one in a company was administrated by a consultant paddling some psycho analysis things to HR. We all had a good laugh, like reading our palms to pigeonhole us into artificial compartments.
No, it’s nothing like a horoscope. It’s not the best we can do as far as empirically validated personality tests but it’s results are, to use the psychometric terms, reliable and valid. One test will agree with another and the scores are useful for prediction.
Comparison with the best in class personality test.
> The big five versus the big four: the relationship between the Myers-Briggs Type Indicator (MBTI) and NEO-PI five factor model of personality
> Abstract
This study sets out examine the relationship between two personality measures—most popularly used measure in the consultancy and training world (the Myers-Briggs Type Indicator) and one of the most heavily used measures in the academic research area on personality (the five factor NEO-PI). One hundred and sixty adults completed the NEO-PI and the MBTI. The NEO-PI Agreeableness score was correlated only with the thinking-feeling (T-F) dimension; the NEO-PI Conscientiousness score was correlated with both thinking-feeling and judging-perceiving (J-P) dimension; the NEO-PI Extraversion score was strongly correlated with the extraversion-introversion (E-I) dimensions, while the Neuroticism score from the NEO-PI was not related to any MBTI subscale score. The openness dimension was correlated with all four especially sensing-intuitive. These results were related to two other similar comparative studies. Results are discussed in terms of recent criticisms of the MBTI.
Hi, psychology student here. It's not very good in a psychometric sense, but its validity problems aren't really a concern when it comes to idiographic data ie: "helpful tools for getting to know the people with whom she was creating a new company".
It's not the devil we memed it into, there are better tools for categorizing and rejecting candidates based on objective-ish traits but this wasn't the purpose according to thr article.
The MMPI-2 is the big reference right now. Legally you'll probably need a licensed psychologist to administer it (depending on your state/country), but that would be true of most useful tests.
The MBTI is mostly useless, because it's a binary schema. Knowing whether someone is slightly more or slightly less introverted than average has extremely limited utility; knowing that someone is exceptionally introverted or exceptionally extroverted is often highly useful.
Isabel Myers explicitly argued that the direction of preference is more significant than the degree of preference, which runs entirely counter to the basic principles of modern psychology.
Whatever merits MBTI may or may not have, one thing is abundantly clear - we have far better personality metrics available.
This spreadsheet looks basically fine to me? They're taking the information they'd normally use to make decisions ("understands the tech industry" etc) and using a tool to be more consistent across candidates.
With race/sex/class those are after the computation of the candidate's total score, which makes me think they're being used for affirmative action in a pretty normal way.
This spreadsheet is what I dread serving on interview panels. It has column headers that are hard to understand but gives a score to that header. Hopefully there’s a companion doc describing the criteria and the meaning between a 1-5.
Without such definitions, this is just a way to ouija board bias into results because you can sum factors and sort candidates.
I’ve found that people like to put more importance on numbers like this than they should and will naturally try to create cutoffs.
The class column is concerning as a hiring criteria, but also that it’s just based on superficial review of info and imputing. Someone going to Wesleyan and studying abroad might be extremely poor on scholarship. Probably not, but might be. Trying to guess class by reading someone’s LinkedIn and then using that guess for something like hiring is confusing to me. That doesn’t seem like good decision making since the “data” aren’t factual.
There’s a term called scientism [0] which is when something is cosmetically like science but kind of bullshit. This smells like whatever the equivalent is for data where the data are all made up bullshit, but then a reasonable looking method is applied to it. Maybe it’s “data scientism.”
There's something about this spreadsheet that I don't get.
The article gives examples of categories graded on a 1-to-5 scale:
> The criteria [for the "social class" column] are: “1) very poor family background, 2) working class, 3) middle class, 4) slightly upper middle, and 5) super rich, super privileged.”
> A key for the column marked “Famous” describes the criteria as: “1) has only published in small media, or hasn’t published very much 2) has published regularly in ordinary-sized media 3) has published a LOT in ordinary-sized or larger media, or has a social media presence exceeding 10K, 4) has published a lot in very popular media AND has a social media presence bigger than 10K 5) practically a household name.”
These do not look like evenly spaced points to me. In particular, the gap between a 4 and a 5 appears to be much larger, in both cases, than the gaps between other adjacent numbers such as 2 and 3.
And that is fine; there's nothing saying your scale needs to use uniform spacing.
But if you look at the spreadsheet, it starts with a bunch of columns assessing various metrics. Then there's a column labeled "Total", the value of which is just the sum of each entry in the preceding columns.
This makes no sense. If your numbers come from a non-uniform scale, adding two numbers together is an error. With the sharp dropoff from 4 to 5, it should be much easier to have a 3 in two categories than to have a 5 in one and a 1 (the minimum! You're guaranteed to be at least this good!) in another.
The "famous" and "social class" columns aren't among the ones that get summed. (Presumably because higher social class is bad rather than good.) But I tend to suspect that the scales for the summed columns were assigned similarly.
And the summed columns do show direct evidence of a very similar problem -- they are color-coded. Each cell is assigned a color ranging from deep red (bad) to white (neutral) to deep green (great). In most of these summed columns, a 3 is neutral, a 4 is light green, and so on, exactly what you'd expect. But in "Competent technology end user", a 5 is deep green, and a 4 is already light red. In "explanatory, helpful to readers", a 3 is deep red. This suggests again that the columns aren't being measured in similar scales, and that adding them makes no sense.
34 comments
[ 3.1 ms ] story [ 67.0 ms ] threadReading between the lines, Gardner basically claimed Angwin -- who was the EIC -- didn't do the job. Angwin wanted to do journalism, but that's not the EIC job. Her job was to hire people to do journalism, and to set up the (prioritization, themes, etc) in which journalism was to be done. Angwin did that poorly. [1] Her job also was to set up technical infra. It was claimed she didn't. etc.
From another article, "Angwin acknowledged to CJR that 'meetings are not my favorite thing.'" [2] . That makes me believe the firing of Angwin was justified; the EIC job is literally meetings.
This is a pattern that is repeated in startups: a technical founder often doesn't understand the job is not to be the best engineer on the team, but rather to hire managers, set priorities, coordinate with the rest of the business, etc.
[1] > "Hiring was slow. Recruitment was slow. Even as of this month, we didn’t have stories banked. We didn’t have editorial processes in place to accept and develop pieces. We hadn’t developed areas of coverage. We still lacked an editorial..." — Jeff Larson https://link.medium.com/oPwISQSD9V https://medium.com/@jeff_larson/about-the-markup-6adc6a77810...
[2] https://www.cjr.org/analysis/the-markup.php
There is real value in reading about the same topic from differing perspectives. Not all writing is or aspires to be purely fact-based journalism, and that's okay.
It's reasonable that The Markup wants to hire group of people with diverse viewpoints on how "powerful institutions are using technology in ways that impact people and society". I would be surprised and underwhelmed if they had e.g. only software engineers or only consumer rights activists writing.
The issue is their assumption that only non-white, non-males have interesting things to say about how "powerful institutions are using technology in ways that impact people and society".
If you want to hire people with diverse viewpoints, then look for people with diverse viewpoints. It's incredibly easy to fill a room with people who look like a Benetton poster but who all read the same books in the same colleges and all rigidly adhere to the same dogma. It's incredibly easy to use tokenism to justify your own prejudices. It's difficult to build a genuinely diverse collaboration based on profound and meaningful differences of perspective.
US Supreme Court jurisprudence more or less. No one actually cares about diversity as a goal. The political coalition that passed the Civil Rights Act looked around, saw they’d won and wanted more stuff for their constituents. Affirmative action was one of the ways to get more stuff. It’s also illegal under a plain reading of equal rights. For reasons I’m unqualified to opine on a quota system is too blatant legally (also explicit points based systems with group X getting Y extra points) so the Supreme Court went with diversity as an acceptable workaround to get to the desired goal of getting more stuff for the desired interest groups. This all happened in the context of university admissions and the buzzword spread from the case law.
Same as free speech or state’s rights people who genuinely believe in them are rare compared to those who use them as talking points when convenient and drop them when not.
In under a generation people go from arguing that everyone deserves equal rights and dignity to campaigning for explicit racial privileges for their side, or from supporting the right of the federal government to seize slaves who escape and travel to a free state to yakking about state’s rights to justify Jim Crow, or the one we’re in the trailing end of right now, the transition from free speech being a left wing value to a right wing one. When the right had the establishment the left was all about people’s right to speak and be heard. Now that they’re in charge it’s safe spaces and hate speech. Or look at the collapse of the anti-war movement once Obama was elected. Bush bombed Muslims for eight years, Obama bombed Muslims for eight years. Bush destroyed Iraq and had no real effect in Afghanistan as it comes pre destroyed. Obama destroyer Libya and cape quite close with Syria before his good sense overcame the presence of Hilary Clinton in his cabinet.
Politics is overwhelmingly about sides, coalitions and spoils. Insofar as it’s about ideology most people have 1-3 things they really care about and the rest they’re pretty happy to modify to fit in with their coalition.
In fact, the whole thrust of my comment was that it's sometimes desirable to have "a genuinely diverse collaboration based on profound and meaningful differences of perspective", and the last paragraph of my post calls out the "nice looking poster" approach.
What point are you trying to make? It sounds to me like we're largely agreeing.
The parent post expressed hope they were hiring writers who would 'report on the objective state of the world, not pursue a particular perspective'. I was pointing out that these are not opposites.
We often think of 'objective' as the opposite of 'biased', but it isn't: it's substituting a personal bias with an impersonal one. This certainly has it's place - it's critical in the sciences - but objectivity has limits when you're dealing with people.
A key strength for any good journalist is the ability to contextualize a story for their readers - not just provide objective facts, but answer the question "What does this mean to me?" A journalists' background and personal perspective - their bias - are relevant because they inform that context.
Or so my amateurish thinking goes. I'd be much interested in a more learned perspective on this, but haven't really found anything yet.
https://www.eeoc.gov/laws/practices/index.cfm
EDIT: I guess EEOC doesn't apply to businesses smaller than 15 people. The 'ol Discrimination Loophole for Small Businesses!
"Non-male"
How does this spreadsheet not represent practices prohibited by the EEOC?
https://www.eeoc.gov/laws/practices/index.cfm
https://employment.law.tulane.edu/blog/what-is-an-equal-oppo...
Comparison with the best in class personality test.
> The big five versus the big four: the relationship between the Myers-Briggs Type Indicator (MBTI) and NEO-PI five factor model of personality
> Abstract This study sets out examine the relationship between two personality measures—most popularly used measure in the consultancy and training world (the Myers-Briggs Type Indicator) and one of the most heavily used measures in the academic research area on personality (the five factor NEO-PI). One hundred and sixty adults completed the NEO-PI and the MBTI. The NEO-PI Agreeableness score was correlated only with the thinking-feeling (T-F) dimension; the NEO-PI Conscientiousness score was correlated with both thinking-feeling and judging-perceiving (J-P) dimension; the NEO-PI Extraversion score was strongly correlated with the extraversion-introversion (E-I) dimensions, while the Neuroticism score from the NEO-PI was not related to any MBTI subscale score. The openness dimension was correlated with all four especially sensing-intuitive. These results were related to two other similar comparative studies. Results are discussed in terms of recent criticisms of the MBTI.
https://www.sciencedirect.com/science/article/pii/0191886996...
It's not the devil we memed it into, there are better tools for categorizing and rejecting candidates based on objective-ish traits but this wasn't the purpose according to thr article.
Isabel Myers explicitly argued that the direction of preference is more significant than the degree of preference, which runs entirely counter to the basic principles of modern psychology.
Whatever merits MBTI may or may not have, one thing is abundantly clear - we have far better personality metrics available.
With race/sex/class those are after the computation of the candidate's total score, which makes me think they're being used for affirmative action in a pretty normal way.
Without such definitions, this is just a way to ouija board bias into results because you can sum factors and sort candidates.
I’ve found that people like to put more importance on numbers like this than they should and will naturally try to create cutoffs.
The class column is concerning as a hiring criteria, but also that it’s just based on superficial review of info and imputing. Someone going to Wesleyan and studying abroad might be extremely poor on scholarship. Probably not, but might be. Trying to guess class by reading someone’s LinkedIn and then using that guess for something like hiring is confusing to me. That doesn’t seem like good decision making since the “data” aren’t factual.
There’s a term called scientism [0] which is when something is cosmetically like science but kind of bullshit. This smells like whatever the equivalent is for data where the data are all made up bullshit, but then a reasonable looking method is applied to it. Maybe it’s “data scientism.”
[0] https://en.wikipedia.org/wiki/Scientism
The article gives examples of categories graded on a 1-to-5 scale:
> The criteria [for the "social class" column] are: “1) very poor family background, 2) working class, 3) middle class, 4) slightly upper middle, and 5) super rich, super privileged.”
> A key for the column marked “Famous” describes the criteria as: “1) has only published in small media, or hasn’t published very much 2) has published regularly in ordinary-sized media 3) has published a LOT in ordinary-sized or larger media, or has a social media presence exceeding 10K, 4) has published a lot in very popular media AND has a social media presence bigger than 10K 5) practically a household name.”
These do not look like evenly spaced points to me. In particular, the gap between a 4 and a 5 appears to be much larger, in both cases, than the gaps between other adjacent numbers such as 2 and 3.
And that is fine; there's nothing saying your scale needs to use uniform spacing.
But if you look at the spreadsheet, it starts with a bunch of columns assessing various metrics. Then there's a column labeled "Total", the value of which is just the sum of each entry in the preceding columns.
This makes no sense. If your numbers come from a non-uniform scale, adding two numbers together is an error. With the sharp dropoff from 4 to 5, it should be much easier to have a 3 in two categories than to have a 5 in one and a 1 (the minimum! You're guaranteed to be at least this good!) in another.
The "famous" and "social class" columns aren't among the ones that get summed. (Presumably because higher social class is bad rather than good.) But I tend to suspect that the scales for the summed columns were assigned similarly.
And the summed columns do show direct evidence of a very similar problem -- they are color-coded. Each cell is assigned a color ranging from deep red (bad) to white (neutral) to deep green (great). In most of these summed columns, a 3 is neutral, a 4 is light green, and so on, exactly what you'd expect. But in "Competent technology end user", a 5 is deep green, and a 4 is already light red. In "explanatory, helpful to readers", a 3 is deep red. This suggests again that the columns aren't being measured in similar scales, and that adding them makes no sense.