Guido van Rossum took the “benevolent” part of his job as Python’s Benevolent Dictator for Life so seriously, three months ago he stepped down from running the open source programming language that he created 29 years ago. Sitting in a conference room at San Francisco-based Dropbox, where he has worked his day job as an engineer for the past six years, he says he’s gone through the stages of denial and acceptance. The once-informal community has also done some grieving and is now wrestling with creating a constitution befitting one of the world’s top four languages.
Python creator Guido van Rossum recently received a Groundbreaker Award, presented by Oracle at its Code One conference in San Francisco.
Oracle
But does all this mean van Rossum is ready to put Python in his rearview mirror?
“Oh no, no no no. I’m very happy that I’m no longer involved in every painful day-to-day decision, that’s the main thing,” van Rossum says. Instead of mediating increasingly stressful mailing list debates that would metastasize to Twitter and Reddit, van Rossum is dedicating himself to other efforts to support the wildly successful language and community. One such activity is mentoring women apprentices to the core team of about 30 developers.
“Open source projects—especially large, well-known open source projects—have been slow in comparison to the rest of the tech industry in letting women into their teams,” says van Rossum, who was recently honored with a Groundbreaker Award, presented at Oracle Code One in San Francisco. Several years ago, he began tackling the diversity problem head-on, admitting in his keynote at PyCon 2015 in Montreal that there were no women among Python’s core developers, and that he was available to mentor women to fix that disparity. Software engineer Mariatta Wijaya from Indonesia eventually took him up on the offer, and two years later became the first woman on the core team.
How Open Source Stays Closed to Women
There’s an altruistic ethos to open source software (OSS), thanks to the characters who founded many of the projects and the fact that, well, it’s free for anyone to use, modify, or add to. We like to think that, beyond underpinning most software systems today, OSS is creating the future democratically. In that light, van Rossum’s observation that its diversity lags that of commercial enterprises reveals a weakness in the OSS ecosystem. He admits he himself hadn’t realized the problem until someone pointed it out at an open source conference.
“Different open source projects have reacted differently to that fact—some with denial, some with open arms, some with hostility,” he says. “I would say a pretty common reaction, especially when this first became an issue, was to say ‘Oh, but we don’t discriminate against women. If there are no women, they must not want to join our project.’ Or, ‘We only select on merit and skill set, and maybe there aren’t any qualified women to be found.’ That is all bullshit, because if you actually look at how a typical open source team decides on who gets to join that team as a core developer, who gets that commit bit, that status of ‘you are now one of us’—that is an imprecise process.”
The discussion about adding someone to the core developer team does not have a clear list of criteria. “It’s not like an engineering job interview. A lot of it is ‘Do we know this person? Have I seen their work?’ Team fit is always a criteria, but if you have a team that is all white males, it is going to be biased toward more white males,” he says.
PyCon blazed the trail, rectifying the dearth of women at their conferences by recruiting more female speakers. Since van Rossum’s invitation to mentor women, Python has added four women to the team of around 50 core developers, with one or two apprentices who might become core developers in the next year, he says.
But will adding women improve Python? After van Rossum’s wrenching decision to step down as Benevolent Dictator for Life, he seems to think so.
Deciding to Leave
His move came after some drama around Python Enhancement Proposal 572—“a fairly minor syntactic enhancement that I personally favor,” says van Rossum. “Quite a few people thought that the sky is falling, and this would be the end of Python as we know it. That felt all rather overdrawn.”
But with Python’s increasing popularity have come problems like “bike shedding” and “concern trolling”—common anti-patterns that fester in online communities. Contentious decisions faced endless regurgitation, accusations of favoritism flew, and nasty tweets still swirled well after decisions were finalized, he says: “I was at a point where I thought, this is causing me too much stress, and I am getting a little older, and I’ve got some health issues—I want to take a step back.”
He attributes the tensions both to growth and culture. “More and more people have heard about Python and as a result the number of people who join the mailing list and start debating stuff, whether they have good intentions or not, has increased,” van Rossum says. At the same time, “the social media and the current mood in politics seems to have given everyone the idea that they can just be nasty online all the time.”
Automating Problems Out of the Way
Python has gained fame for its ease of use, making it one of the top languages for data science and machine learning. But van Rossum doesn’t think it should be taught to everyone, and he’s not convinced, when pressed, that all efforts to teach programming are worthwhile: “We’re probably giving too many people a pretty shallow training in programming. A six-week or six-month boot camp is going to maybe make you an acceptable apprentice-level web developer, but in the end, there is not that much future in that. That’s like digging ditches. Sooner or later someone will develop a bulldozer for that field.”
That metaphorical bulldozer epitomizes how software development works. “Programming is particularly interesting in that you solve a problem by automating it out of the way,” he says. “And then you build something else on top of that automation. You keep building layers of abstraction, which is why very few people need to learn about binary numbers anymore. They are so many abstraction layers down the stack that they’re just not relevant.”
Similarly, he can see plenty of ways that Python will become less necessary as an alternative to learning to program in C, which requires manual memory allocation and has myriad validation issues, as different tools emerge.
“Honestly, a lot of machine learning is already in this area where there are incredibly powerful libraries or frameworks that do the heavy lifting. Those are developed by large teams of very detail-oriented programmers who have many, many years of expertise writing C and C++ up and down the stack,” he says. Those resulting systems are hugely productive for data scientists and others—but that doesn’t mean there aren’t still abundant annoyances to fix in the languages and tooling that tech companies building software at scale use.
Too Late to Pivot
For that reason, van Rossum isn’t planning to pivot off Python and try to solve some other quotidian problem.
“I think I’m too old for that,” he says. “I don’t want to fall into the trap of believing ‘Oh, I’ve had this tremendous success in this one field, so now I can use my fame and smarts to solve a completely different field where I have zero expertise.’”
And anyway, there’s plenty to do to create a more equitable core developer team for Python—one that reflects how diverse the wider Python community really is. After all, he says, “Men don’t have the monopoly on wisdom and technical expertise. Men certainly don’t have necessarily the best way of making good decisions.”