4 ms·

I think many of the comments are missing a few important points: 1. Yes, 20% Asian hiring rate is pretty decent when you compare to the general Asian populatio

by thedz 10y ago

I think many of the comments are missing a few important points: 1. Yes, 20% Asian hiring rate is pretty decent when you compare to the general Asian population at large. However, the lawsuit specifically alleges that, in one example case, the qualified application pool was 73% Asian. In that case, it's extremely abnormal to hire 17 non-Asians and only 4 Asians. 2. Palantir is involved in government contracting, so there are also very specific regulations for compliance https://www.dol.gov/ofccp/aboutof.html https://www.dol.gov/ofccp/aboutof.html 3. The lawsuit is being brought on by the Department of Labor. IMO, generally government agencies are usually reluctant to initiate lawsuits that they are afraid of losing. 4. It might be malicious, it might not be. But Palantir was given multiple chances to correct their compliance issue: "The Labor Department sent Palantir a notice in October 2015 about its findings, according to the lawsuit. Both before and after that notice, labor regulators attempted to secure Palantir's voluntary compliance, the lawsuit said, but they did not succeed."

danso 10y ago
Has any other major tech company released numbers on what their hiring demographics look like compared to their qualified application pool? I've seen the demographics for employees in diversity reports, but not qualified-applicant pool numbers. Because, sure, perhaps many of the Asian applicants are poor enough at English that it outweighs whatever got them into the qualified-pool. But I would guess that would be a problem for every SV tech company. So how does Palantir's stats compare to theirs such that it warrants the DoL to file a lawsuit?
- jonathankoren 10y ago
  While I can't comment on the DOL's methodology because I don't know what it is, I've seen many SV tech companies essentially define the qualified-applicant pool in a way to explain systemic biases. The most obvious one, is "Unless you attended school X, you're unqualified," which quite frankly, is utter bullshit.
- Twirrim 10y ago
  Relying on referrals sure does punt the ball down the park to the employees from the employers. I wonder how much it allows the company to say "It's not our fault!"
derefr 10y ago
> However, the lawsuit specifically alleges that, in one example case, the qualified application pool was 73% Asian. In that case, it's extremely abnormal to hire 17 non-Asians and only 4 Asians. It's abnormal, sure, but that doesn't imply that the abnormality is Palantir's "fault." It might be that, say, one department of one major University, which happens to have an outsized Asian student population, heavily encourages its grads to apply to Palantir. As a result, Palantir would see one "general" applicant stream, from all walks of life, but with a self-selection effect filtering for people who are (in their own opinion) good matches for working at Palantir; and then one applicant stream of people who happened to go be in the Maths program at University X and didn't want to say no to some advisor's strong recommendations. The self-selected stream would be better hires than the "coerced into applying" stream for obvious reasons; and the "coerced into applying" stream would also likely be much larger than the self-selected stream. If the people in the "coerced into applying" stream, then, fit a certain profile, it would appear in the resulting data that people who fit that profile are worse candidates than you'd expect. Doing a causal factor analysis to the applicant graph, though, would immediately show that the contributing factor of being Asian (or whatever else) is "screened off" by the contributing factor of being an alumni of University X. Given that Palantir is full of the very people who can do this sort of analysis, though, I fully expect that if this is the case, they'll end up attempting this argument on a judge. Probably with amusing results.
kelukelugames 10y ago
Government agencies would not take a suit unless it's a slam dunk.
- JackFr 10y ago
  That's definitely not true. In many cases, corporations realize that settling is cheaper than fighting and winning -- and government agencies bank on that.
  - kelukelugames 10y ago
    That's definitely not true. Can you name many cases where the government filed frivolous suits? If you can't provide a list then have you thought about where that opinion came from? Government agencies are afraid of having a reputation like this so they don't touch questionable suits. Otherwise, agencies would receive even more frivolous complaints. Agencies are political. Settling weak cases looks bad to the voters. However, agencies are required to investigate every claim no matter how stupid. The people who hope to settle are ones with failed investigations. Those have to pay for their own lawyers, eg Pao vs Klein Perkins. On the other hand, companies have more incentive to settle when government agencies are suing because the case is tight.
morgante 10y ago
> However, the lawsuit specifically alleges that, in one example case, the qualified application pool was 73% Asian. My question is how the government is determining the "qualified applicant pool." Not everyone with a CS degree is qualified to develop software. In my experience, only a minority of CS graduates are qualified to program. I'm guessing there is some confounding variable here which predicts both ethnicity and poor qualifications. Perhaps being a master's student looking for a visa sponsor.
- dqv 10y ago
  An “Internet Applicant” is an individual who satisfies all four of the following criteria: * The individual submitted an expression of interest in employment through the Internet or related electronic data technologies; * The contractor considered the individual for employment in a particular position; * The individual’s expression of interest indicated that the individual possesses the basic qualifications for the position; and * The individual, at no point in the contractor’s selection process prior to receiving an offer of employment from the contractor, removed himself or herself from further consideration or otherwise indicated that he/she was no longer interested in the position. The “basic qualifications” which an applicant must possess means qualifications that the contractor advertised to potential applicants or criteria which the contractor established in advance. In addition, the qualifications must be: * Noncomparative features of a job seeker (e.g. three years’ experience in a particular position, rather than a comparative requirements such as being one of the top five among the candidates in years of experience); * Objective (e.g., a Bachelor’s degree in accounting, but not a technical degree from a good school); and * Relevant to performance of the particular position. https://www.dol.gov/ofccp/regs/compliance/faqs/iappfaqs.htm#Q2GI https://www.dol.gov/ofccp/regs/compliance/faqs/iappfaqs.htm#...
  - morgante 10y ago
    Thanks for linking. Do you know what the specific qualifications were in this case though? The problem is that you could find literally thousands of developers with a CS degree and 3+ years of experience who couldn't program their way out of a paper bag.
mattnewton 10y ago
I really think we need better statistics here. Palantir should definitely be making moves to make sure race isn't the dividing line, using something like more rigorous online tests and bias training for interviewers, but there are plenty of plausible explainations for this difference of Asian applicants vs Asian hires based on how they defined "Qualified". What if the job involved frequent client contact and you needed to asses a high English ability? That would no doubt discriminate against a lot of non native speakers who might have the requisite degree, but would they be qualified? There is also the issue of degrees or other certifications being usually useless to assess real world coding ability, and certainly useless to assess interview whiteboarding questions.

by thedz 10y ago

danso 10y ago

Has any other major tech company released numbers on what their hiring demographics look like compared to their qualified application pool? I've seen the demographics for employees in diversity reports, but not qualified-applicant pool numbers. Because, sure, perhaps many of the Asian applicants are poor enough at English that it outweighs whatever got them into the qualified-pool. But I would guess that would be a problem for every SV tech company. So how does Palantir's stats compare to theirs such that it warrants the DoL to file a lawsuit?

jonathankoren 10y ago
While I can't comment on the DOL's methodology because I don't know what it is, I've seen many SV tech companies essentially define the qualified-applicant pool in a way to explain systemic biases. The most obvious one, is "Unless you attended school X, you're unqualified," which quite frankly, is utter bullshit.
Twirrim 10y ago
Relying on referrals sure does punt the ball down the park to the employees from the employers. I wonder how much it allows the company to say "It's not our fault!"

derefr 10y ago

> However, the lawsuit specifically alleges that, in one example case, the qualified application pool was 73% Asian. In that case, it's extremely abnormal to hire 17 non-Asians and only 4 Asians. It's abnormal, sure, but that doesn't imply that the abnormality is Palantir's "fault." It might be that, say, one department of one major University, which happens to have an outsized Asian student population, heavily encourages its grads to apply to Palantir. As a result, Palantir would see one "general" applicant stream, from all walks of life, but with a self-selection effect filtering for people who are (in their own opinion) good matches for working at Palantir; and then one applicant stream of people who happened to go be in the Maths program at University X and didn't want to say no to some advisor's strong recommendations. The self-selected stream would be better hires than the "coerced into applying" stream for obvious reasons; and the "coerced into applying" stream would also likely be much larger than the self-selected stream. If the people in the "coerced into applying" stream, then, fit a certain profile, it would appear in the resulting data that people who fit that profile are worse candidates than you'd expect. Doing a causal factor analysis to the applicant graph, though, would immediately show that the contributing factor of being Asian (or whatever else) is "screened off" by the contributing factor of being an alumni of University X. Given that Palantir is full of the very people who can do this sort of analysis, though, I fully expect that if this is the case, they'll end up attempting this argument on a judge. Probably with amusing results.

kelukelugames 10y ago

Government agencies would not take a suit unless it's a slam dunk.

JackFr 10y ago
That's definitely not true. In many cases, corporations realize that settling is cheaper than fighting and winning -- and government agencies bank on that.
- kelukelugames 10y ago
  That's definitely not true. Can you name many cases where the government filed frivolous suits? If you can't provide a list then have you thought about where that opinion came from? Government agencies are afraid of having a reputation like this so they don't touch questionable suits. Otherwise, agencies would receive even more frivolous complaints. Agencies are political. Settling weak cases looks bad to the voters. However, agencies are required to investigate every claim no matter how stupid. The people who hope to settle are ones with failed investigations. Those have to pay for their own lawyers, eg Pao vs Klein Perkins. On the other hand, companies have more incentive to settle when government agencies are suing because the case is tight.

morgante 10y ago

> However, the lawsuit specifically alleges that, in one example case, the qualified application pool was 73% Asian. My question is how the government is determining the "qualified applicant pool." Not everyone with a CS degree is qualified to develop software. In my experience, only a minority of CS graduates are qualified to program. I'm guessing there is some confounding variable here which predicts both ethnicity and poor qualifications. Perhaps being a master's student looking for a visa sponsor.

dqv 10y ago
An “Internet Applicant” is an individual who satisfies all four of the following criteria: * The individual submitted an expression of interest in employment through the Internet or related electronic data technologies; * The contractor considered the individual for employment in a particular position; * The individual’s expression of interest indicated that the individual possesses the basic qualifications for the position; and * The individual, at no point in the contractor’s selection process prior to receiving an offer of employment from the contractor, removed himself or herself from further consideration or otherwise indicated that he/she was no longer interested in the position. The “basic qualifications” which an applicant must possess means qualifications that the contractor advertised to potential applicants or criteria which the contractor established in advance. In addition, the qualifications must be: * Noncomparative features of a job seeker (e.g. three years’ experience in a particular position, rather than a comparative requirements such as being one of the top five among the candidates in years of experience); * Objective (e.g., a Bachelor’s degree in accounting, but not a technical degree from a good school); and * Relevant to performance of the particular position. https://www.dol.gov/ofccp/regs/compliance/faqs/iappfaqs.htm#Q2GI https://www.dol.gov/ofccp/regs/compliance/faqs/iappfaqs.htm#...
- morgante 10y ago
  Thanks for linking. Do you know what the specific qualifications were in this case though? The problem is that you could find literally thousands of developers with a CS degree and 3+ years of experience who couldn't program their way out of a paper bag.

mattnewton 10y ago

I really think we need better statistics here. Palantir should definitely be making moves to make sure race isn't the dividing line, using something like more rigorous online tests and bias training for interviewers, but there are plenty of plausible explainations for this difference of Asian applicants vs Asian hires based on how they defined "Qualified". What if the job involved frequent client contact and you needed to asses a high English ability? That would no doubt discriminate against a lot of non native speakers who might have the requisite degree, but would they be qualified? There is also the issue of degrees or other certifications being usually useless to assess real world coding ability, and certainly useless to assess interview whiteboarding questions.