It’s 2019. Open source now powers most everything, from mobile to data infrastructure to pretty much all software we use (yes, even proprietary software is largely composed of open source under the hood). We have come a long, long way since open source really started to take off in the 2000s.
Except, sadly, in how we monetize open source—that model is stuck in the 1980s. This fact has been brought into focus by the endless Twitterizing (here and here and here and…) over MongoDB and AWS. If we strip the fight down to its bare bones, it’s not really about AWS “stealing” open source code—it’s really about the failure of the open source business community to grow out of antiquated licensing models.
SEE: Software licensing policy (Tech Pro Research)
Making money the old-fashioned way
I have spent roughly two decades in open source, most of that working for a range of open source startups that hoped to strike gold by giving away software. A few of those companies were acquired, but only MongoDB managed to reach escape velocity and IPO. When I left, the company was worth $1.6 billion—now it’s closer to $5 billion. It’s a fantastic company selling an amazing database. It just happens to not be the only company that wants to sell that database.
If we were talking about proprietary software, this wouldn’t be a discussion. No one can sell Salesforce but Salesforce. No one can sell Microsoft Office but Microsoft. Etc. Those are the rules.
Open source changes those rules. If I license my code under an open source license, anyone else is free to use it, so long as they abide by the license terms. To inhibit would-be competitors, for years companies like Alfresco, MySQL, and others licensed their software under the GPL or one of its variants, figuring no competitor in their right mind would use code that required them to open source their own code.
It was a way of turning copyleft even “lefter,” making it about keeping code open but effectively proprietary. It was the new proprietary, brought to life by mostly old-school enterprise software veterans who wanted to make a buck by selling Proprietary 2.0 (using open source as a marketing gimmick). It was, in sum, a way of prolonging proprietary licensing in a world that wanted to go cloud.
Your license is leaking cloud
Some would say this model failed because of the cloud. Analyst Ben Thompson, for example, has written, “The monetization model depended on the friction of on-premise software; once cloud computing is dominant, the economic model is much more dubious.” Why dubious? Because the licensing can’t compel a purchase.
Kyle Mitchell, for example, has argued that “Standard licensing has failed to keep copyleft up as a viable, general-purpose commercial tool.” Put another way, as Andrew Shafer summarized, “Cloud providers are monetizing OSS [open source projects] more effectively than the primary sources and some primary project sponsors feel they should be the only one entitled to monetize, so they are relicensing to claw back that right.”
Or put even more bluntly: Cloud vendors are selling what enterprises actually want.
SEE: Cloud migration decision tool (Tech Pro Research)
Going back to Thompson:
There is a secular shift in enterprise computing moving to the cloud, not because it is necessarily cheaper…but because performance, scalability, and availability are hard problems that have little to do with the core competency and point of differentiation of most companies….
This leaves MongoDB Inc. not unlike the record companies after the advent of downloads: what they sold was not software but rather the tools that made that software usable, but those tools are increasingly obsolete as computing moves to the cloud. And now AWS is selling what enterprises really want.
In MongoDB’s case, they sell a robust cloud offering called Atlas, geared toward smaller customers. But Atlas is an island in a sea of enterprise infrastructure when what AWS offers is a “Pangea,” connecting a vast array of enterprise services.
Which brings us back to open source.
Open source 3.0
As Thomas Dinsmore has correctly argued, “It’s impossible to argue that software should be open AND the originators have the sole right to monetize. If you believe in the latter, the solution is commercial software. With a license key.” This is where we are in 2019: Open source has eaten the world, but companies keep trying to monetize it in ways that would have looked proper in 2009 but today seem antiquated.
This conflict is made worse by the fact that AWS, Microsoft, and Google are so much better at turning software into the services that companies increasingly want. Sticking up a license wall to block them may buy a vendor a little more time to scrape together pennies, but the real money is in offering services at scale AND in sufficient breadth that enterprises don’t need a separate platform for each piece of software they’d like to use.
If we look through the eyes of vendors, MongoDB’s actions (relicensing under the SSPL to block cloud competitors) look completely reasonable. But if we look through the eyes of customers, AWS’ actions seem correct. This customer vantage point puts AWS in the driver’s seat, and perhaps suggests that they need to find a way to ensure competitors like MongoDB become partners. The world, after all, needs both: AWS can’t be the source of all great software services, open source or otherwise—it needs the MongoDBs of the world to create amazing things and then share in the proceeds.