The General Public License has long been the preferred license for open-source businesses, but new analysis suggests that Apache-style licensing may yield more adoption and money.
I have spent years advocating the GNU General Public License as the optimal open-source license for commercial open source.
Roughly nine years after I first became a fan of the GPL, I think I've been wrong.
My admiration for the GPL mostly stemmed from its ability to mimic, but then invert, proprietary licensing. The GPL is like opening a cannister of radioactive waste: while your competitors can touch it, you're dead certain that they won't.
Given that openness is increasingly a winning business model--if not the winning business model, as Red Hat executive Michael Tiemann argues--one has to wonder if pretending to be open through the GPL accomplishes as much as fully opening up through Apache-style licensing would.
Open-source luminary Eric Raymond is pretty clear on this point:
I think we live in a...universe...in which the GPL is unnecessary rather than futile. Mind you, I am not claiming the GPL is entirely useless. It's a signaling behavior, like wearing a crucifix or yarmulke or pentagram; it helps build trust groups. But it has costs, too.
It creates a lot of needless fear from potential allies and users who suspect they won't be able to control their exposure, if they let it in...Is the GPL's utility as a form of in-group signaling worth the degree to which fear and uncertainty about it slows down open-source adoption? Increasingly, I think the answer is no.
The GPL may be a community-building signaling device, but it is also a confession of fear and weakness. To believe that it matters, you have to believe that you live in a...universe where closed-source development is such an attractive proposition that you have to punish people for trying to move to it.
In other words, if openness works (in the Jamesian, pragmatic way), why not give it free rein, rather than hedging our open-source bets to the point of obviating their efficacy?
Equally important, we may not be getting the "protection" we seek from the GPL, anyway, as the GPL becomes the new BSD in the cloud, as Linus Torvalds recently commented to me in an e-mail:
AGPL/GPLv3 anti-ASP/TiVo language doesn't "protect" anything. There is no upside to pushing freeloaders away.
Sun Microsystems CEO Jonathan Schwartz rightly identifies adoption, not protection of freedom, as a key open-source benefit: open source provides an efficient way to distribute software to the maximum audience at the minimum price. With this in mind, unfettered Apache-style licensing would be the ideal license to maximize adoption, despite likely being the worst way to directly monetize software.
So long, however, as one's business either monetizes software indirectly (i.e., Google with its advertising model) or adds to the open-source components with commercial extensions (i.e., IBM with proprietary software, services, and hardware add-ons), then a company should be able to reap a bounteous harvest from its open-source seeds.
In sum, the GPL may well be an excellent capitalist tool, but Apache licensing could well be even better.
Disclosure: My company uses the GPL, not an Apache license.
Follow me on Twitter @mjasay.