Hacker News Reader

Transparency about our interviewing process: http://bjk5.com/post/3340326040/in-any-language-you-want-kha... (which comes from the hiring cultures at Fog Creek and Google, see http://www.joelonsoftware.com/articles/GuerrillaInterviewing...).

Transparency about all past internships, their expectation and accomplishments: http://bjk5.com/post/36067210182/khan-academy-internship-sum... http://bjk5.com/post/8826207372/khan-academy-internship-summ... http://bjk5.com/post/15500332273/khan-academy-internship-fal...

Transparency from previous interns about their work, which I believe speaks for itself: http://david-hu.com/2011/11/02/how-khan-academy-is-using-mac... http://dylanv.org/2012/09/14/the-khan-academy-internship-exp... http://jamie-wong.com/2012/08/22/what-i-did-at-khan-academy/

If any piece of evidence from the above points to an organization that values quotas over excellence in interns and the work expected of them, well please raise the alarm.

We need to get past our default response being a knee jerk assumption of discrimination.

Fogcreek and Google's interview process both seem to yield a far larger number of male employees than females yes?

David, Dylan and Jamie are all men yes?

If any piece of evidence from the above points to an organization that values quotas over excellence in interns and the work expected of them, well please raise the alarm.

No one is accusing you of using quotas and no is saying you do not value excellence in interns. What's being asked is why was there such a steep increase in the number of female interns accepted? Were there more female applicants? Were interview questions changed? How did this result come about?

And what I'm asking you is: can you come up with any possible explanations for an increase in women devs over time that should be celebrated, especially at an organization that values equal education for everyone, without first wondering if our interview questions have changed?

Can you come up with any possible explanation period? You're dodging my question.

An increase in women in tech majors and a company mission that strongly appeals to both women and men.

I'm pretty sure most universities have a company mission that appeals to both men and women and they haven't suddenly seen a 2:1 f/m ratio in CS majors. What makes Khan so different?

He might have a point in that a lot of women seem to aspire to be teachers. For example when I got my maths degree, 99% (estimated) of the women sitting in the lectures with me were studying to become teachers. Perhaps working for Khan Academy somehow seems close enough to teaching.

What % of the applicant pool was women?

Some hard numbers would help this discussion.

Why is an explanation required? Any developer worth his or her salt knows that, statistically, something like this is bound to happen over enough companies and enough years, even if the trend is so heavily skewed the other way.

Is this a 3 sigma deviation? Maybe. Or could it be that, on average, women might be drawn to educational software more than other types? Doesn't sound unreasonable.

Regardless, your insistence on an explanation shows that you're still stuck in the bad old days.

As a hiring manager for over a decade, I can't recall the last time I received a resume from a qualified female applicant for a full-time software position.

Internships tend to be a bit less one-sided, but the skew is still quite high.

This sucks. I want to receive those resumes, for what I hope are obvious reasons, but I don't ever receive them.

Given that, I can't help but wonder whether:

1) You have a completely different applicant pool

or

2) You somehow changed your application process in a way that resulted in a 2:1 female:male ratio.

I don't think those are terrible questions to have. The answers could be enlightening.

I don't think you have to feel attacked, isn't it natural to ask what is going on if you produce an outlier? It could be genuine curiosity.

In the same vein it is interesting to ask why most companies receive more male applications. Without asking these questions, how can there ever be an improvement (except for random luck)?

One thing Fog Creek did was we hunted harder for qualified women applicants, which meant we actually had to go to female tech recruiting events.

The problem is that there are few female developers to start with, and they are all heavily recruited by large software companies (Google, Microsoft, etc) already. The larger companies have big budgets to devote to this and encourage the very small number of women developers to go work there.

If you just sit back and wait for more women to apply, they are going to get picked up by the bigger companies before you even get a chance to interview them.

It's also difficult to attract that small group of women devs to apply to work at your company when there are no women developers there currently. The larger companies already have a bunch of women developers, so it can suggest to future candidates that these companies are great places for them to work (i.e. they won't be discriminated against at these companies). It may be that your small company is also a great place, it's just that you don't have any women to vouch for that.

So in Fog Creek's case, we increased the number of women applicants by making it our goal to get more women to apply. That's the only thing we changed. Our interview process stayed exactly the same.

And that's fine but did you get a 2 to 1 f/m ratio after you increased your female recruitment? That's what I'm trying to get to the bottom of.

Why does an investigation need to be launched?

Having worked with Kamens, including interviewing the same candidates, I can't even entertain the idea that the interviews were rigged in some way, which is you abhorrent implication.

We see 2:1 male to female ratios all the time, and no one bats an eye. But when it gets flipped we need to figure out why?

Regardless of how it came to be, this is a win for our industry, as it sets an example that the numbers don't need to be so heavily skewed towards men.

We see 2:1 male to female ratios all the time, and no one bats an eye. But when it gets flipped we need to figure out why?

Simple probability. A 2:1 male environment is completely reasonable in a gender neutral process, whereas a 2:1 female environment is exceedingly unlikely to occur under such a process.

http://news.ycombinator.com/item?id=5233153

"it's well known that minority students are admitted to prestigious universities with far lower SAT and GPA's that their white/Asian counterparts"

SATs have been criticised for having inherent socioeconomic bias. The validity of both SATs and GPAs, and their appropriateness as gauges of student intelligence, future performance, learning, and appropriateness for use in admissions have all been questioned.

Do you have a better criterion with which to measure?

How about proven ability to overcome adversity?

Well A) How do you plan on quantitatively measuring that so that all candidates are equally considered? and B) What does that have to do with writing software?

LockeWatts: What does the SAT have to do with writing software? I'd say being able to overcome adversity, i.e. being able to do more with what you've been dealt than others believe you can, is usually going to be a stronger indicator of future performance than is a test score. Furthermore, I'd argue in software development especially, you need to be able to overcome people's initial compunctions, and you need to be resourceful and able to figure out how to succeed when it seems like the chips are stacked against you.

As to how you test for that: why not ask? The SAT and ACT both have an essay portion. Why not ask candidates to write an essay, or tell you in person, about how they've overcome adversity? Also, how is asking interviewees how to solve a programming problem more quantitative than asking them how they've dealt with adversity?

"I'd say being able to overcome adversity, i.e. being able to do more with what you've been dealt than others believe you can, is usually going to be a stronger indicator of future performance than is a test score."

Well you can say that, but I haven't seen any studies that would indicate as such. However, multiple studies have shown that logical reasoning, critical thinking, and mathematical aptitude all lead to better programmers. These are things the SAT attempts to select for.

"why not ask? The SAT and ACT both have an essay portion."

A much discussed and argued about section, precisely due to it's inability to be quantitatively and statistically analyzed.

Admissions officers at multiple schools indicated they gave the essay portion little to no consideration for precisely this reason.

"Why not ask candidates to write an essay, or tell you in person, about how they've overcome adversity?"

To colleges or jobs? For jobs, simply because it's a less efficient method.

"Also, how is asking interviewees how to solve a programming problem more quantitative than asking them how they've dealt with adversity?"

You can measure things about this. Candidate A took N minutes to solve the problem ideally, Candidate B took N + 5 minutes. All other things being equal, candidate A is superior to candidate B.

Furthermore, you can look at things such as programming fundamentals that are easily quantifiable, such as being able to correctly create a log(n) versus n search algorithm, or being able to correctly indicate why log(n) might be superior to n.

Finally, 'adversity' hasn't even been defined in this conversation, so I'm arguing against an ethereal concept. Please define it for me.

Adversity: a state, condition, or instance of serious or continued difficulty or adverse fortune (see adverse)[1]

One of the biggest problems with relying purely on quantitative data is that you bias the results to those who are better at achieving quantitative goals. EDIT: and your results are inevitably skewed by the test. Are you saying communication and writing capability are unimportant, simply because they're difficult to quantify?

What you're arguing for is, given a choice of bad quantitative tools and OK to mediocre qualitative ones, using the quantitative tools as they're the only way you can measure things. This is a similar argument as to why grades are more important than actual teacher feedback, despite some teachers giving 95% of the class an A, and others giving 20%.

The SAT has a history of discrimination, and, in addition, there are proven stereotyping effects: give self-identifying Asian women a math test, and they'll perform better when you make them identify with being Asian then when you make them identify with being a women[2].

Given how much energy has been spent recruiting me (having me fly out to interviews, organizing tech talks at my school, having employees at the company take me to lunch, etc.), and given that every company I interviewed with asked something along the lines of "tell me about a project you've done recently," which also is a subjective, non-quantifiable question, it hardly seems like an unreasonable burden to ask, as part of the regular interview questions, how a candidate has overcome adversity.

And admissions officers don't give weight to the SAT essay because they have their own essays with which to judge a candidate with. And the SAT just provides a quantitative essay score, whereas colleges can see and evaluate multiple written pieces. And they do often ask questions like how you've overcome adversity. Which makes sense, as they want students who won't give up when facing difficulties.

[1] http://www.merriam-webster.com/dictionary/adversity

[2] http://icos.groups.si.umich.edu/shihpaper.html

"Why not ask candidates to write an essay, or tell you in person, about how they've overcome adversity?"

Wouldn't such a system be very easy to attack, how would you avoid the judges being biased, or the claim that they could be biased? Since judging an essay would probably not be an automated thing at this stage of AI development?

Most colleges (especially private colleges) already require you to write essays. Having one on adversity is also, for certain colleges, already in place. I'm not talking about automated judging, or even explicit scoring (besides figuring out hire/not hire, or next round or not), i.e. no, this can't really be used to exactly replace the SAT, but it could be used as an evaluative supplement.

Didn't know that, I never actually had to deal with the US admissions system.

There seems to be an assumption here that Khan Academy made some change that resulted in this outcome, instead of it being a natural occurence.