I used TripleByte to recruit people for a startup without paying, by asking people to sign up for the test and screenshot the answers and collecting the scores at the end. I already had an inbound funnel, I just liked your test when I saw it the first time.
In 2015, candidates with the same schooling scored about 20 percentage points lower than 2020, the last candidates I used the test for. The number of questions doubled, and then the test got much easier, by eliminating more challenging programming questions and replacing them with questions with giveaway context. By comparison, according to my data, about N=43, until about 2018 being a senior versus a junior in college CS programs is worth about 10 percentage points on the test; going to Harvard instead of Berkeley is also worth about 10 percentage points. In 2020, the last tests had no predictive features.
I stopped using the test, because it became too easy and too noisy to be informative.
I recognize some of the coded language in the blog post. There are definitely more lucrative opportunities in recruiting for DEI. I don't know if it will last. If you're still jittery about the public-profile-by-default thing, which by the way, was totally irrelevant and overblown IMO, this may not be a pivot for you.
One thing I see in the data is that at MIT, women and men performed the same, controlling for seniority. This wasn't true at the 3 other universities that produced enough data to measure.
That said, what really is the best way to hire candidates? I'm not convinced having binders full of engineers is special, there are almost always more candidates than jobs, at least 5:1, in every non-credentialed industry vertical. Anyone who has worked at a jobs (or indeed any matching platform, like the Common App or Tinder) knows that.
Then there's this long thing about asymmetries or whatever, warble garble about missing information... It has never, not once been my experience that someone seeking a subordinate role at a typical private company with preferences like "pair programming" or whatever have ever been better than someone with no preferences at all.
Maybe it helps to engage in the vanity of whatever trendy workplace trend is hot for whatever vertical. But like, if you're being intellectually honest, if you thought pair programming was important, you'd pair program at TripleByte, but you don't, you know in your heart of hearts none of that shit matters, so why are you putting stuff like that into your search system?
Indeed and ZipRecruiter are ad arb companies. They don't care. Private universities lead, not lag, DEI at giant companies, so it's hard to see how to compete against them in that core business. It will still come down to a real defensible opinion.
Do you have more valuable inventory than ZipRecruiter for DEI candidates? Who knows. What an uninteresting question. Apple also hires people who just make shit up on their resumes, I know two - though they weren't engineers.
A recruiter once told me, if you see 30 applicants for a high level tech job on LinkedIn, not to worry about competition, because ~27 of them were likely to be unqualified trying-their-luck applicants, mostly from other countries where the company was not going to hire, and mostly with no relevant skills or experience.
They told me the actual number of qualified candidates was usually just a handful or less.
I doubt this is true in SV, but I don't live in SV.
> going to Harvard instead of Berkeley is also worth about 10 percentage points.
That's interesting to me. Most reputable rankings put Berkeley at #2 or #3, and Harvard is usually around #7 or #8 (and US News puts them at #16!). I wonder what the cause of that discrepancy is.
CS program rankings are largely based on their graduate programs. The parent comment is almost certainly talking about candidates with bachelors degrees. For undergraduate Berkeley has a 16% acceptance rate and Harvard has a 4% acceptance rate. This will likely act as a filter in combination with the quality of the undergraduate CS program, which could be better at either institution but is hard to get a metric for.
I was specifically looking at undergrad rankings. The acceptance rate is not a good filter because that's the acceptance rate for the entire school, not the CS program.
From what I can tell, anyone who is a Harvard student can declare the CS major if they pass the pre-reqs. At Berkeley, passing the pre-reqs doesn't get you into the major -- you still have to apply for a limited number of spots, which generally means having a 3.8 GPA or better in your pre-reqs. So effectively both programs have the same acceptance rate.
Pedigree filtering is most often myopic elitism, especially in a business context. Anecdotally, I avoided MIT and Harvard because of Boston's snow and traffic. CalTech is Pasadena: too frick'n hot. I didn't study at all for the SAT-I and aced the math section. (The SAT should be more like the JEE.) I went to an expensive, "top 50" public university where I liked the area. Also, it was more practical and rigorous academically than most Pac10s (Pac12s now) and Ivys because they had something to prove (no grade inflation at all, they don't care if you turn in homework or not, not much market for homework and tests, and proctored exams). I also didn't do a PhD because of the economic disincentive: if I put in the work and the cost, I could do an MD, JD, or PhD CS but there is no added benefit.
Most clinical doctors have social skills and crystalized intelligence from domain expertise, but aren't typically mistakable for particle physicists.
People don't necessarily require the proper sheepskins to possess fluid, crystallized, and/or other domains of intelligence AND the skills, personality, and experience relevant to excelling at a particular STEM role.
Pedigree is mostly used for social filtering and business leadership board packing, but if someone wanted to create an elite monoculture of staff, lacking in cognitive and personality diversity, by all means, go right ahead.
PS. I. Don't even get me started about a car full of CS IITians talking about the JEE and high-placers who seem normal on the way to snow country for snowboarding and gambling.
PS. II. I bombed an Apple interview for a mid-career role by being too intelligent and too maverick compared to the group of compliant, I hate to say, yuppies. It was an interview panel of about 9 people and they were just speechless. The other time I bombed an interview for being too smart was about 10 years before that when I was 20 at the old Borders bookstore in Palo Alto. Moral of the story: it's important to play dumb where appropriate because most people are relative-intelligence insecure.
PS. III. Sorry, reader, for the rambling and discontinuous thoughts. Absurd endocrine values of unknown etiology currently... doctor appointments pending. ):
I have a sneaking suspicion that being "too smart" isn't why you bombed those interviews. I know I’d never want you on my team after reading just this one post, can’t imagine having to suffer through a multi-hour interview too.
> I bombed an Apple interview for a mid-career role by being too intelligent and too maverick compared to the group of compliant, I hate to say, yuppies. It was an interview panel of about 9 people and they were just speechless.
Are you sure they were speechless due to your high intelligence and maverick qualities?
> PS. I. Don't even get me started about a car full of CS IITians talking about the JEE and high-placers who seem normal on the way to snow country for snowboarding and gambling.
In 2015, candidates with the same schooling scored about 20 percentage points lower than 2020, the last candidates I used the test for. The number of questions doubled, and then the test got much easier, by eliminating more challenging programming questions and replacing them with questions with giveaway context. By comparison, according to my data, about N=43, until about 2018 being a senior versus a junior in college CS programs is worth about 10 percentage points on the test; going to Harvard instead of Berkeley is also worth about 10 percentage points. In 2020, the last tests had no predictive features.
I stopped using the test, because it became too easy and too noisy to be informative.
I recognize some of the coded language in the blog post. There are definitely more lucrative opportunities in recruiting for DEI. I don't know if it will last. If you're still jittery about the public-profile-by-default thing, which by the way, was totally irrelevant and overblown IMO, this may not be a pivot for you.
One thing I see in the data is that at MIT, women and men performed the same, controlling for seniority. This wasn't true at the 3 other universities that produced enough data to measure.
That said, what really is the best way to hire candidates? I'm not convinced having binders full of engineers is special, there are almost always more candidates than jobs, at least 5:1, in every non-credentialed industry vertical. Anyone who has worked at a jobs (or indeed any matching platform, like the Common App or Tinder) knows that.
Then there's this long thing about asymmetries or whatever, warble garble about missing information... It has never, not once been my experience that someone seeking a subordinate role at a typical private company with preferences like "pair programming" or whatever have ever been better than someone with no preferences at all.
Maybe it helps to engage in the vanity of whatever trendy workplace trend is hot for whatever vertical. But like, if you're being intellectually honest, if you thought pair programming was important, you'd pair program at TripleByte, but you don't, you know in your heart of hearts none of that shit matters, so why are you putting stuff like that into your search system?
Indeed and ZipRecruiter are ad arb companies. They don't care. Private universities lead, not lag, DEI at giant companies, so it's hard to see how to compete against them in that core business. It will still come down to a real defensible opinion.
Do you have more valuable inventory than ZipRecruiter for DEI candidates? Who knows. What an uninteresting question. Apple also hires people who just make shit up on their resumes, I know two - though they weren't engineers.