Stimulated in part by a controversial Eric Raymond blog post, open source thought leaders like Matt Asay are publicly questioning the contemporary merits of the GNU General Public License (GPL). In fairness, Matt has long been a GPL advocate. And it's never bad to question whether past practices make sense going forward. All that said, the notion that open source vendors should simply abandon the GPL in favor of liberal licenses like BSD/Apache is myopic and wrong-headed, in my opinion.
Raymond (author of The Cathedral and the Bazaar) posits that, in a world transitioning to open source development and distribution, the GPL is unnecessary. Further, he implies that those who use the GPL for protection against "misbehavior" create needless friction and reduce the efficiency of open source adoption. Raymond speculates that those licensors will ultimately be penalized in the same way closed source licensors will be penalized – it'll just take a little longer. Raymond goes on to conclude that, in a world of open source believers, the GPL is a "confession of fear and weakness".
Asay's thesis is similar but more grounded. He points to the use of liberally-licensed open source software by vendors like Google and Facebook as evidence that frictionless licenses stimulate broad organic adoption. His belief is that, had MySQL and Linux been licensed under the Affero GPL (a version of the GPL that expands the definition of "distribution" to include network-based apps), neither of these technologies would be used today by the likes of Google. So if open source software vendors want to achieve the broadest possible IP adoption, they should remove all friction and go straight to a liberal license like BSD/Apache.
While I find Eric Raymond's post intellectually and logically engaging, that's about as far as it goes. Raymond has framed the discussion in an "ideal world" context, where actors are motivated by binary alternatives. The real world – where entrepreneurs must make multi-dimensional, bet-the-business decisions against a backdrop of gray (i.e., neither black nor white) – is magnitudes more complex than the one implied in Raymond's post. Thus, Raymond is theoretically safe but pragmatically unhelpful.
Asay approaches the open source license choice from a practitioner's perspective. His longstanding position, with which I wholeheartedly agree, is that most open source vendors cannot make sustainable revenues by exclusively farming their downloader communities (exceptions like Red Hat are few and far between). Vendors must have complementary "pay-for" products, regardless of which open source licenses they use. Given this understanding, it's critical for vendors to weigh the trade-offs of greater control (GPL) vs. higher adoption (BSD/Apache). But they are trade-offs!
In commercial applications, the GPL offers two potent levers:
- It preserves a vendor's ability, via dual licensing, to acquire revenue from closed source ISVs until a competent, liberally-licensed open source competitor enters the market.
- It prevents closed-source competitors from ripping off high-value open source IP and exploiting it to their own competitive advantage. Further, it prevents all non-GPL competitors from hammering an open source vendor over the head with its own innovations.
As Matt points out, the above benefits come at the cost of lower adoption – the GPL is a higher friction license than BSD/Apache. So if a vendor's business model and market strategy are such that the benefits of higher adoption outweigh the benefits of greater protection, then I'd concur that BSD/Apache is the better choice of license for that vendor.
Practically speaking, however, many open source projects are sponsored by early stage software companies that are trying to break into highly competitive markets. They want to harness the disruptive nature of open source to get their unique innovations into the hands of developers, early adopters and evangelists. They don't have a portfolio of pay-for products ready to roll out. For these companies, the GPL is the best way to go because it gives vendors leverage to make near-term ISV revenues and provides them cover from hostile competitors while they build out their market beachheads.