When it first launched in 1991, Python “wasn’t lucrative,” remembers long-time Python community organizer Paul Everitt (now a Python and web developer advocate at JetBrains). “But we believed in it. The passion was there — we were doing good in the world.”
Yet surprisingly, Python traveled a bumpy early road on its way to becoming the world’s #1 most popular programming language, safely ensconced in the nonprofit Python Software Foundation that would help it grow through the years.
It’s a story Everitt will share in a soon-to-be-released documentary from Cult.Repo (formerly part of the tech-focused job platform Honeypot).
“We get so attracted to Python’s success,” Everitt told me this week — but what’s overlooked is “the story of the people, and the story of the Foundation, the story of what did happen and what didn’t happen.”
There’s passion, some trial and error, and a few truly hair-raising near misses. But what comes through is the message that community matters — and that a community’s values really can make a difference.
The Birth of a Community
After distributing Python for years through Usenet newsgroups, Python creator Guido van Rossum attended a breakthrough in-person meeting in 1994. “Obviously, we were emailing each other,” Everitt remembers, “but there was this physical event — 20 people, Gaithersburg, Maryland — at a windowless government office building.
“That’s when the community started,” he says with fondness in his voice. “As something separate from the language.”
It was at the National Institute of Standards and Technology, and Everitt laughs, remembering how it had seemed like such an abundance of space — in light of how massive the Python conferences became later.
“The people that got Guido in that room are the unsung heroes of Python,” Everitt says — future Python core developer Barry Warsaw and Michael McLay, then a NIST electronics engineer. Because they’d ultimately get van Rossum a job at the Corporation for National Research Initiatives — the long-running nonprofit (founded in 1986) focused on public-interest projects improving the world’s network-based technologies. Everitt remembers this as the beginning of “the engine that powered” Python’s early days — “Guido and team at CNRI, working for Bob Kahn, the co-creator of the internet!”
From here, Guido’s team brought on Roger Masse and Warsaw, who’d also been “bitten by the Python bug” at that same Python workshop at NIST — and the Python community grew stronger and stronger. “This team helped me create and maintain additional Python community infrastructure such as the python.org website, the CVS server, and the mailing lists for various Python Special Interest Groups,” Guido remembered on his “History of Python” blog.
Barry Warsaw even took over maintenance of a Python mailing list software created by John Viega — and then convinced the Free Software Foundation to adopt it as its official mailing list tool, GNU Mailman.
The Quicksand of Bylaws
“The Python workshops continued,” Guido wrote, “at first twice a year, but due to the growth and increased logistical efforts they soon evolved into yearly events.” But around this time, Guido also remembered that CNRI “tried to come up with a model to fund Python development more directly than via DARPA research grants.”
Everitt grins, remembering it as “The wrong model.”
Van Rossum’s site notes that their new Python consortium had “a minimum entrance fee of $20,000. However, apart from one group at Hewlett Packard, we didn’t get much traction, and eventually the consortium died of anemia.” Looking back, Everitt says it’s his opinion now — as the founder of these organizations — that this model “was never gonna work. It presumed that Python was going to grow into being a commercial success, where companies would want to pitch in.” Even if it could’ve worked someday, “It was way too early.”
So how do you organize a community? Van Rossum’s blog also remembers that even their first attempt to form a foundation “ended up in the quicksand of bylaw drafting.” Bob Kahn instead suggested forming the Python Software Activity, “which would not be an independent legal entity but simply a group of people working under CNRI’s legal (nonprofit) umbrella. The PSA was successful in rallying the energy of a large group of committed Python users, but its lack of independence limited its effectiveness.”
The world was moving all around them, and van Rossum wrote that “Eventually, in early 2000, the dot-com boom, which hadn’t quite collapsed yet, convinced me and three other members of the CNRI Python team (Warsaw, Jeremy Hylton and Fred Drake) to join BeOpen.com, a California startup that was recruiting open source developers.”
And as far as institutional support for the language, “The PSA just died,” Everitt says, “and there was nothing. Fortunately, they didn’t own any intellectual property.”
That One Last Question …
On the way out, there was one last question about how to license Python 1.6. Van Rossum remembers earlier versions used a modified MIT license, “with basically one sentence added where CNRI disclaimed most responsibilities. The 1.6 license, however, was a long, wordy piece of lawyerese crafted by CNRI’s lawyers.”
But since the Free Software Foundation was now using GNU Mailman, they were invested in Python remaining truly free. “We had several long wrestling discussions with Richard Stallman and Eben Moglen of the Free Software Foundation about some parts of this new license,” van Rossum wrote. “With the help of Eric Raymond, changes to the CNRI Python license were made that satisfied both the FSF and CNRI,” along with a coveted seal of approval from the Open Source Initiative.
And to this day, van Rossum wrote in 2019, “Only slight modifications were made to the text of the license to reflect the two successive changes of ownership, first BeOpen.com and then the Python Software Foundation, but in essence the handiwork of CNRI’s lawyers still stands.”
Though ironically, van Rossum wrote, in the end, BeOpen’s business plan “failed rather spectacularly.”
‘If Python Dies … ‘
Luckily, the core Python team was quickly hired by Paul Everitt’s Zope Corporation (an early player in web application servers). “We had just gotten our third round of venture capital money,” Everitt remembered, so his startup had obviously gotten a lot of value from Python. “And we’re like, ‘Listen, if Python dies … that’s a problem!'” (As Everitt says in the upcoming documentary, “If all the Python guys went their separate ways, Python wasn’t big enough to survive that at the time.”)
But instead, Python’s core team was now employed to work full-time on Python. And that’s also when the Python Software Foundation springs into existence. “We were standing amid the ashes of all these other things,” Everitt said wryly.
It was Everitt who joined with long-time open source advocate Greg Stein to vote the Python Software Foundation into existence. But Everitt acknowledges that “A lot had been done up to that point, where everybody was like, ‘This is what we’re going to do. Go do the stupid bureaucratic trick to spring it into existence!’ We had this basis of things that we wanted to codify and put a structure around, to make sure that it could continue.”
Community Owned?
Specifically, Zope had already been consulting the FSF, the OSI and the Software Freedom Law Center, and all those people,” Everitt remembers, “just understanding the legal regime of open source” — from copyright, to joint ownership rights, patent clauses, and contributor agreements. So he’d had a lot of experience.
But more conversations were held about structuring, and Everitt said it was “very much in his head” that the legal entity should be a 501(c) — a community-owned nonprofit to hold Python’s intellectual property rights. So when the Python team finally arrived, “I put into their employment agreement that we would own none of the intellectual property for what they did on Python. Because previous employers had gotten their names into the copyright.”
In 2002, Everitt moved to France, coinciding with the next era for Python, where “The people involved in the first third gave way to the people involved in the second third.” Because it was a natural decision to hold a conference — called PyCon — and as Everitt remembers it, “PyCon was the real engine for change. Because it made money.”
Finally, there would be a way to fund ongoing developer and infrastructure improvements.
‘A Secret Weapon’
“Python started being viewed as a secret weapon,” Everitt remembers (as the biggest companies grew “and needed to implement internet services fast … Developer productivity became more important than other factors.”) As companies scrambled to hire Python teams, they needed “a footprint” in the community, Everitt says, “which is manifested as the conference. And wouldn’t you really like to sponsor this conference, so that you look more attractive than the other people?”
With that money, the Python Software Foundation was now able to hire full-time professional staffers. And as Everitt notes, “Once you have a legal entity and you hire staff — they can go shake the trees for money. People like [PSF executive director] Eva Jodlowska were responsible for getting the big money from the Googles of the world — and not to sponsor the conference, but to sponsor ongoing infrastructure for Python.”
Hewlett Packard Enterprise funded a full-time developer for Python’s package repository, which also received a $170,000 grant by Mozilla’s Open Source Support program and a two-year grant from Bloomberg for a full-time project/community manager.
Everitt says one of the most gratifying parts of the story is Python’s commitment to diversity. But maybe that was there from the very beginning, Everitt suggests — an embrace of heterogeneous cultures he can trace all the way back to its roots in the Open Source movement of the 1990s, “which is by definition not really much of a monoculture.”
Everitt believes it’s this ongoing culture that ultimately brought in new and diverse cohorts that were well-positioned to then jump onto new trends like AI.
Cult.Repo says its Python documentary will be released August 28. (A sneak peek was shown in July at EuroPython in Prague.)
TRENDING STORIES