Roughly 26 years after Linux pushed open source into mainstream enterprise adoption, we're still debating how big a role open source should play -- at least, some people are.
Developers aren't in that group: Open source has become part of the "furniture" for them, essential to data infrastructure (Hadoop, Spark, and so on), mobile (Android), operations (Docker, Kubernetes), machine learning (TensorFlow, Mahout), and more.
The PR world, however, didn't get the memo. Canonical, the company behind Ubuntu, sent out a note this past week arguing that companies that "don't jump on the open source bandwagon" risk being "left behind." SAS's PR team, by contrast, put out a report offering helpful guidelines: Make sure your software portfolio contains no more than 40 percent open source or face potential exposure to all sorts of security, cost, and other problems.
Neither is correct.
SAS silliness
Let's hope, for starters, that RapidMiner CMO Tom Wentworth is right and the SAS broadside against open source is the product of a rogue field marketing team, and not the company's official position. Either way, it's embarrassing.
In the paper, SAS correctly argues that open source versus proprietary software is not an either/or decision. Both can be useful. The company's credibility comes into question, however, when it advises enterprises to maintain a balance of 60 percent proprietary, 40 percent open source (no more, please!). How they arrived at this bizarre conclusion is hard to fathom, except that SAS sells more than $1 billion worth of proprietary software every year and presumably would like to continue, despite a clear trend toward open-source-powered analytics.
Indeed, unfortunately for SAS, enterprises already seem to have the SAS ratios wrong.
In a Burtch Works survey of over 1,100 quant pros, 61.3 percent prefer open source R or Python to SAS, and only 38.6 percent opting for SAS, with that percentage growing for open source options every year.
Worse for SAS, a variety of open source data infrastructure and analytics tools threaten to encroach on its bastions in data management, business intelligence, and analytics. As Cloudera co-founder Mike Olson has boldly stated, "No dominant platform-level software infrastructure has emerged in the last 10 years in closed-source, proprietary form." Nearly all innovation in data infrastructure is happening in open source, not proprietary software. That's a tide SAS can try to fight with white papers, but it would do better to join by embracing open source in its product suite.
Besides, it's not as if developers are going to read that whitepaper. They're too busy downloading and using open source software.
Understating adoption
Which is where Canonical's own silly PR takes center stage. To be fair, it started with saber-rattling from Linux Foundation chief Jim Zemlin, who thundered down at Google Next attendees: embrace "shared innovation" or you "will fail." Canonical, however, made open source a matter of life and death, decreeing in its subject line that enterprises must "Adapt [to an open source world] or Die."
Subtle -- and sort of irrelevant. After all, though Canonical touts, "In 2016, 65 percent of companies were already leveraging open source software to speed application development and innovation," the number is actually 100 percent or very close to it. (SAS, for its part, put the percentage of open source adopters at a mere 25 percent, which is pathetically wrong.)
That's right: All or nearly all enterprises already use open source. Such adoption may not be on the CIO's radar, but that's hardly a good indicator of what's really happening within an enterprise. The real folks to ask are the developers, and they're too busy downloading open source code to answer. They are, however, asking each other for tips and tricks all the time on Stack Overflow and other such forums, each of which is riddled with discussions on various open source projects.
Zemlin's job, in other words, isn't to convince companies to adopt open source, but rather to provide a home for the nurturing of open source projects, so they're worthy of adoption. Similarly, Canonical can focus on contributing code rather than spooking enterprises into adopting more.
And SAS? Well, it should probably start with 40 percent open source adoption and grow from there.