I’m still kicking around the ideas suggested by Tim Bowden’s post, which suggested that the GPL is a better licensing choice than BSD for vendors establishing commercial dominance around an open source project.
If you were to draw up a list of the most successful commercial open source vendors, I believe they would all be based on either the L/GPL or the MPL. Certainly, taking Tim’s central point about M&A valuations for open source vendors as the yard stick, then the largest open source M&As have all involved copyleft licenses (although Ian Skerrett believes this has as much to do with trademarks as it does license, and of course the brand is significant).
With community-led projects, vendors more often that not face a choice of two business models: support services (Covalent) or proprietary extensions (EnterpriseDB). Both models enable commercial businesses to flourish, although neither does so as quickly as the captive model has done via dual-licensing (MySQL) or mandatory subscriptions and copyright control (Red Hat Enterprise Linux).
The dual licensing approach, in particular, gives the open source vendor control over the ‘open market’ for support and services around a captive open source project, and has proved popular for customers concerned about the implications of adopting GPL software. Meanwhile Red Hat’s model is as close as you can get to a dual-licensing approach in a community-led GPL project.
Arguably it is these models, rather than the license itself, that has enabled the vendors to establish dominance in a particular market (although more research is needed into the difference between captive GPL projects created by commercial operations and commercial operations that have emerged from community GPL projects).
Meanwhile, earlier this week I met up for a chat with Gianugo Rabellino, the CEO of Sourcesense, during which he made the point that what is good for the vendor is not necessarily good for the customer. As an ASF member and Apache committer, Gainugo leans towards community- rather than captive-open source software and points out that for every MySQL there will be several open source vendors that don’t make it.
The failure of open source start-ups is an aspect of the industry that open source has not yet had to face up to. There have been isolated incidents of vendors falling by the wayside, and it is likely there will be more, although most investors will look to get some return from their investment, rather than just letting it dwindle away.
So far M&A activity around open source vendors has been driven by the value extant in the open source model (although you could argue about XenSource) but what happens if and when open source vendors are acquired by a vendor in order to kill an open source alternative or to take advantage of dual licensing to take a project proprietary?
It is certainly theoretically possible for the dual licensing approach to be used to remove the freedoms and flexibility that attracted customers to open source software in the fist place. While the original code would remain GPL, unless a community springs up to support it, it is effectively moribund. Of course the BSD license specifically allows this situation to occur, but the fact that projects are community-led rather than captive enables them to survive and thrive while vendors come and go (see PostgreSQL).
According to Gianugo, if an open source vendor shake-out occurs, the potential advantages of community-led projects will come in to play for customers. The community-led project arguably offers a better model for ensuring the long-term availability of code and the creation of a contestable market for support and services. If customers find that the dual license approach isn’t the safety net they thought it was, the argument goes, they could migrate towards community-led projects.
Wavemaker CEO, Christopher Keene’s The Silverado Rules for Open Source Success points out that license strategy is one of many factors influencing success for VC-backed commercial open source vendors (although he does recommend a dual license strategy based on the GPL).
Ian and Tim continue the discussion in the comments on Ian’s blog.
Open Source Licensing: Obsolete or Of Importance?, Redmonk’s Stephen O’Grady.
Thinking Legal for Open Source Success: Trademarks & Licenses, Mark Radcliffe.