Commentary: The industry used to look to Red Hat to define open source success, but the cloud has made things more complicated.
We’re in a strange, somewhat unpredictable period in open source that has been caused perhaps by a lessening of Red Hat’s industry impact over the years. On Twitter, Brianna Wu asked men over 40 to comment on “structures [that] existed in your life to teach you how to be a good man.” Answers included things like Boy Scouts. A similar sort of question might be asked of developers and “open source structures…to teach you how to be a good open source citizen.”
When I got started in open source, the obvious answer to most every question was “Red Hat.” What’s the right way to build a business in open source? Look to Red Hat was the stock response. What’s the right way to advocate for code freedom in open source? Again, look to Red Hat.
SEE: 40+ open source and Linux terms you need to know (TechRepublic Premium)
Over the last 10 years, Red Hat has lost its standing as the center of the broader open source community, though not due to any fault of its own; instead, other institutions have displaced Red Hat’s authority without replacing it. We don’t seem to be the better for it.
How to disappear not so completely
AWS VP – and, perhaps more pertinently for this post, former Red Hat engineer – Matt Wilson correctly identified a variety of good open source things that have come from Red Hat over the years. And he just as correctly has asserted that Red Hat employees “like Kevin E. Martin are still advocating for FOSS graphics drivers, going on since 1998. And still scoring wins for the larger community.”
And yet, Red Hat no longer defines open source success, whether in business or code. The biggest open source contributor? That’s Google if we define it as lines of code contributed, or Microsoft, if we define it in terms of number of employees active on GitHub. Even the much maligned AWS has more active contributors to open source than Red Hat. You can run this analysis yourself using the open source project Fil Maj has created.
On the business side, it’s much the same. The big cloud providers not only contribute more code, they also make orders of magnitude more money with and from open source than Red Hat. Deep Discovery CTO Russell Jurney has argued that it’s this shift to cloud that has fundamentally devalued the open source ethos that reigned during the Red Hat years: “The shift to cloud computing reduces the number of companies making direct investments in open source by several orders of magnitude and concentrates control of companies’ indirect investments in the hands of a few intermediaries that lack the same incentives as individual companies which ensured their ethical participation.”
He might be right that the shift to cloud has destabilized the way open source got built before, but it’s hard to see how his argument holds up in light of the fact of mega investments in open source across the cloud landscape, particularly if we include companies that deliver services through the cloud like Netflix and Facebook. These companies that sell services, not software, have revenue models that make it much easier for them to be big contributors to open source. Open source entrepreneur and venture capitalist Peter Levine once famously argued that there would never be another Red Hat, to which the industry first took umbrage (“Of course there will be!”) and then showed indifference, because Red Hat was no longer the marker for success.
Teenage [open source] wasteland
Without a Red Hat as our accepted standard, as an industry we don’t seem to have a North Star. Or maybe we have several. Is that a bad thing?
It can sometimes be a messy thing. “[Y]ounger dev[elopers] today are about POSS – Post open source software. **** the license and governance, just commit to github.” Thus spake RedMonk co-founder James Governor, and thus has gone open source licensing on GitHub. It’s hard to ignore the continued decline of intentionally licensed open source code on GitHub, though it would be unwise to think this somehow means developers don’t care about openness in code. But that openness is generally about access, not distribution.
SEE: Master Linux and Docker before the next Linux adoption boom (TechRepublic Academy)
It’s also the case that while Red Hat set a standard, it was never universally followed. Way back in 2007, SugarCRM adopted CPAL, which required visible attribution or “badgeware” of the SugarCRM logo to prevent commercial forking. Over the years, there have been a host of so-called Open Core approaches to licensing. Though there has been plenty of discussion about such licensing approaches, customers haven’t seemed to pay much attention. Whether post-open source GitHub kids or quasi-open source companies, there no longer seems to be one accepted way to “do open source.”
I once thought this mattered. A lot. I’m no longer sure.
On the one hand, I liked the clarity of the Red Hat era. On the other hand, I love the experimentation of whatever this current era should be called. No, I don’t like each of these experiments, and you probably don’t either. I also have no idea where we’ll land. The new normal on the business front seems to require a fully managed cloud service to make it easy to consume open source, but we don’t seem to be anywhere near consensus on how it all should be licensed. Maybe that’s a good thing.
Or, as Percona’s Matt Yonkovit expressed it, “The evolution and changes in open source are good, scary, troubling, and welcomed all in one.” He’s right.
Disclosure: I work for MongoDB and used to work for AWS, but the views expressed herein are mine alone.