The reason some firms hire only top school kids is (apart from ego) they know that whatever test/puzzle/interview they give their decision in hiring candidates will still be fairly inaccurate.
However, good credentials (I'd normally include internships but it wasn't mentioned in the original article.) simply means that there's a higher chance of the kid having a better understanding of programming than another candidate coming from a less competitive pool. What the author does in order to suppress this statistical strategy is giving anecdotal evidence. Problem with that is for every example the author might come up, this community can produce 100 nontrivial counterexamples.
One other issue is most founders who do not come from top schools have a tendency to think that kids in these schools are simply giving so much money to have a cheesy-easy undergraduate life and are terribly spoiled (and that might be partly true). Yet the fact is those are the top schools because students have to produce work that is quantitatively and qualitatively superior to their counterparts in other schools. This is not because of everybody are geniuses in these schools but because the average student starts at a higher level so the academics can push the students further without overestimating their capacity from day 1.
Cheers for being brave enough to challenge the prevailing sentiment on HN about credentialism.
But, strong disagree.
In my experience as a hiring manager (which stretches back quite a ways), I've noticed no correlation between school and on-the-job performance. I'm sure MIT does indeed present more challenging CS/EE curricula, but the inference I draw after stipulating that is that CS/EE curricula quality just doesn't matter much in the real world. Perhaps we just do our most important learning in our first jobs, or, even better, throughout our career? Maybe memorizing MESI cache coherence in school is just less important than, say, having to learn on the job how to do a u/k copy in an ioctl handler because that's the only way to accomplish what your next dev task is?
And, you suggest that tests/puzzles/interviews are "fairly inaccurate". Well, all of technology hiring is inaccurate. Compared to other professions, the recruiting/onboarding process in the tech industry is amateurish across the board whether you ask people to design manhole covers or review Github pages. The answer to this is to improve the tests, and, I think, to evacuate as much of the subjective stuff (school and GPA, yes, but also "interviews" as much as possible) so you can make apples-apples comparisons to candidates and discover what factors really correlate to good performance.
I think if school attended influences on the job performance, that's a failure of the hiring/interview process. It is fine if school attended influences interview performance, but if it influences it too strongly that may allude to a failure of recruitment/screening process (failing to look at other indicators of performance, e.g., open source contributors, code samples, etc...)
I think when hiring (or vice-versa, making a choice of people I will work with) I look for some evidence that they've done something they're "not supposed to" have done. Growing up in a slum and yet attending Berkeley/Stanford/MIT/IIT Madras is such evidence -- higher education _is_ an equalizer -- but growing up in a wealthy suburb, going to a great prep school, and then and attending a top-tier school is insufficient.
Real open source contributions, serious undergrad research (even if done at an obscure liberal arts college), writing your own screen replacement "just because", building and open-sourcing an impressive piece of infrastructure at a previous job, are other examples of such evidence.
I have a correlation between school [performance] and on-the-job performance. I know you were probably comparing different schools, but within the same school there is definitely a difference. At my co-op job we always pulled from the same school[1], since we were on campus. We had to interview people basically every semester in order to keep up with the continual rotation. As we started dropping our GPA requirement in order to attract more people (we ended up with like a 2.5, school's co-op program required 3.0), the quality of hires significantly decreased in soft skills - self tasking, critical thinking, and creativity compared to when the GPA requirement was a 3.5.
Most hires were from their sophomore year, so they all had the same basic classes coming in and had a baseline of knowledge.
[1]There were a few exceptions, but they knew someone in our org before they interviewed
MIT people can be very good (ob: I am an MIT people, so I obviously have a huge bias here), but they can also be way too theoretical for development work. Not too long ago, it was entirely possible to get a CS/EE degree from MIT without doing any Java or C programming, which were the vast majority of industry use at the time.
PhDs, from any school, are also something I've gotten to be wary of. Of the 5 best coworkers I've ever had, all 5 had PhDs, but there is a lot of variance. Sometimes they spent a lot of time understanding one very specific problem very well, and people who have that problem should pay a premium for that talent, but if you don't have that problem they won't be worth it to you.
I would not call it unfortunate. After giving people from Ivy Leage schools a chance to work with me I've realized something. The amount of people who can ship software (not only write code) is basically the same everywhere. A college education does provide a lot of things, but the ability to ship software seems to not be included. Seems a lot of people out there are worried about the code being perfect, or their VIM setup being ideal, or anything else but shipping. I wish that somehow colleges of all levels would teach their students that the discipline to get shit done is worth more than anything else.
That depends on who you are hiring. If you're hiring Computer Science graduates than "getting shit done" is not a priority at all (at least in the meaning you imply). Computer Science graduates gets the discipline to be scientists not to be code ninjas or rockstars and it should stay that way.
On the other hand if you're meaning people with Computer/Software Engineering degrees then you're right.
Well, if Computer Science graduates are applying for Computer Programming positions, then I expect them to be able to ship software. I don't go around knocking doors asking for them. They come to me. By the way, people who ship code are not code ninjas or rockstars. They are called professionals.
> The reason some firms hire only top school kids is (apart from ego) they know that whatever test/puzzle/interview they give their decision in hiring candidates will still be fairly inaccurate.
For young firms in software industry, this is not true. Neither Google nor Facebook (disclosure: my employer) hires exclusively from top schools. These companies do spend more effort recruiting interns and new graduates from top Computer Science programs (which is a rather different set of schools than top schools in general), but the reason puzzles, participating in competitions like TopCoder or ACM ICPC, research, open source contributions, etc... matter is that they give smart people everywhere a chance to stand out and be noticed.
On the other hand, the technical interview process in these firms is known more for false negatives than false positives. These companies all explicitly track the relationship between interview scores and employee performance and tune the interview process accordingly.
Finance, management consulting, etc... firms do recruit exclusively from top schools -- but they have other reasons for doing so. If you're in technology business but only hire from top-tier schools and fail to setup a meaningful technical interview process, you're more than welcome to -- I would highly encourage all of my potential competitors to do so :-)
I made fairly silly choices as a high school student (as well as not-so-silly, e.g., interning at a startup junior and senior years of HS) which led to me having to "hack" my way through a higher education despite having high SAT scores: after high-school I attended a community college for a year, got a near 4.0 GPA, transfered to a university no one heard of (I had a chance to stay one more year and transfer to Berkeley or UCLA, but I jumped the gun out of fear that my performance was "illusory" as I've never had such a high GPA before) while interning at yet another startup. More traditional companies (think Cisco, Symantec, etc...) ignored my resume (unknown school, internships only at startups), but nearly everyone I actually wanted to work for gave me a shot and I had plenty of good offers. I'm now 29 and I don't feel my career suffered as a result of silly choices (e.g., getting modest grades in high school, transferring too soon from a community college) I made between ages of 14 and 19.
The reason some firms hire only top school kids is (apart from ego) they know that whatever test/puzzle/interview they give their decision in hiring candidates will still be fairly inaccurate.
However, good credentials (I'd normally include internships but it wasn't mentioned in the original article.) simply means that there's a higher chance of the kid having a better understanding of programming than another candidate coming from a less competitive pool. What the author does in order to suppress this statistical strategy is giving anecdotal evidence. Problem with that is for every example the author might come up, this community can produce 100 nontrivial counterexamples.
One other issue is most founders who do not come from top schools have a tendency to think that kids in these schools are simply giving so much money to have a cheesy-easy undergraduate life and are terribly spoiled (and that might be partly true). Yet the fact is those are the top schools because students have to produce work that is quantitatively and qualitatively superior to their counterparts in other schools. This is not because of everybody are geniuses in these schools but because the average student starts at a higher level so the academics can push the students further without overestimating their capacity from day 1.