Thread by @sarahmei

A lot of folks are (rightly) taking issue with the “low barrier to entry” asserted in this tweet.

I want to explore something a little different - a blind spot that we in the tech industry have, that this reveals.

We often assume that the percentage of women graduating from computer science programs tracks the percentage of women entering the industry overall. As in, if 25% of CS grads are women, then 25% of the engineers entering the market that year are women.

One reason we make this assumption is that it’s one of few publicly-available hard numbers.

More significantly, though, large companies mostly hire from this group, so that’s the number they care about - & they’ve historically driven the larger diversity conversation.

The problem is, it’s never been true. There has always been another way into the industry besides through a CS program. Before code schools, that way was mostly self-study & open source, which is presumably the “low barrier to entry” that the original tweet refers to.

That barrier to entry is certainly low when compared to a four-year computer science or engineering degree. But historically, that avenue has only been open to white men, or to people willing to conceal their identity enough to be mistaken for white men.

So even before code schools, the CS grad numbers didn’t tell the whole story. If anything, they _overestimated_ the proportion of women in the industry, since almost all the people entering it via informal paths were not women.

Back before github, in the bad old days when open source contribution meant emailing a patch to a mailing list, I contributed to a number of projects using a carefully neutral handle.

Like many women, I created that handle because when I tried to contribute using my real name, my patches were ignored, downplayed, or nitpicked to death. I never got one actually accepted under my name.

Trouble is, I needed to make sure nobody ever connected real-me to that contributor - because if they did, it would be back to the defending & arguing & gatekeeping & nitpicking. So I couldn’t put that experience on my resume, or discuss it in interviews.

I had some close calls. One memorable interviewer directly accused me of lying when I forgot & mentioned I’d contributed to an Apache module.

He happened to work on the same one and, I guess, couldn’t fathom one of the other contributors being me.

As a result, that open source work didn’t give me the cachet it would have given someone operating under their own identity.

Eventually, after yet another string of thoughtless sexism on the mailing lists (“because there are no girls here anyway lol”) I quit altogether.

Has this improved recently? I suspect not, given the recent study that said pull requests from women were more likely to be accepted but ONLY if it was wasn’t obvious they were women.

So yeah. There are informal routes in to the industry with relatively low barriers to entry - and there always have been - but they’re really only open to men.

With the rise of code schools, the landscape shifted. Another entryway to the industry appeared, with a lower barriers to entry than a CS degree.

Are code school cohorts more gender-balanced than CS degree cohorts?

Yes, almost always.

Does that mean the structural issues that meant I couldn’t get patches accepted as a woman just...evaporated?

NOPE.

The biggest barrier to entry in software for non-white-men has never been credentials or knowledge or even experience.

It’s always been plain old discrimination, and neither Github nor code schools have fixed that part.

So now we have code schools, self-study/open source, AND CS degree programs as entryways to the field. By some counts, code school grads outnumber CS grads, but also have a harder time breaking in.

The landscape is way more complicated than it was even 10 years ago.

It makes less and less sense to rely on gender balance in CS degree cohorts to stand in for “entering class” gender balance overall.

I mean, it never _really_ made sense. But now it’s just indefensible. We need better data to really understand where we are. [fin]