Tuesday afternoon I visited Singularity University‘s graduate studies program to participate in a session on open source with Chris DiBona (Google Open Source Program) and Matt Mullenweg (WordPress). It was pretty interesting, though not in the way I expected — lots about contemporary licensing issues (which DiBona called roughly “the boring yet intellectually interesting part of his job”, a hilarious characterization in my book), not so much about how open source development will impact the future, nothing about an open source singularity. I sped through slides which include a scattershot of material for people interested in open source, a grand future, and not necessarily familiar with Creative Commons.
Hearing Mullenweg’s commitment to software freedom in person made me feel good about using WordPress. There was some discussion of network services and relatedly the AGPL. Mullenweg made a comment along the lines of silos like Facebook being less than ideal (not discussed, but BuddyPress, built on WordPress, as well as Elgg, used for SingularityU’s internal social network, are open replacements, though it seems to me that federation a la OpenMicroBlogging is needed).
DiBona indicated that Google doesn’t use AGPL’d software internally because it might cause them to share more than they’ve decided to (and they consciously decide to share a lot) while Mullenweg wondered whether complying with the AGPL would be difficult for WordPress deployers, including the question of whether one would need to share configuration files that include passwords. One could argue that such doubts are very self serving for Google and to a lesser extent WordPress.com (which use tons of free software and aren’t forced to share their improvements, though as mentioned, they both share lots), however, I hope that AGPL advocates (including me, with the caveat that I consider the importance of copyleft of whatever strength relative to release under any open license and non-licensing factors an open and understudied — consider possibilities for simulation, econ lab, and natural experiments — question, and I’m happy to change my mind) take them as strong signal that much more information on AGPL compliance is needed — sharing all source of a complex deployed web application is not often a simple thing.
Not explicitly but much more than tangentially related, probably the single most interesting thing said on the panel was Mullenweg saying that any internal WordPress.com developer can push changes to production at any time, and this happens 15-20 times a day, and he wishes he could do this for other deployments. My longstanding guess (not specific to WordPress) is that making deployment from revision control the preferred means of deployment would facilitate both more deployments running the latest changes as well as sharing their own.
I got a sense from questions asked by the students that the current Singularity University program might be more mainstream than the name implies. I understand that each student, or perhaps group of students, is to write a plan for using emerging technology to positively impact the lives of a billion people in one of the areas of health, climate change, or (I forgot the third area) in the next ten years. In any other context those parameters would sound very aggressive. Of course they could be met by first becoming rationalist jedi masters and then turning all available matter into computronium. Alas, ten years is a hurdle.
If the previous paragraph reads snarkily, it is not — I fully support the maximization of computation and the rationality of the same. In any case congratulations to all involved in SingularityU, in particular Bruce Klein, who I know has been working on the concept for a long time. It was also good to see Salim Ismail and David Orban. I’m especially happy to see that SingularityU is attempting to be as open as possible, not least this.