Oscon 2006 - Tim O’Reilly’s Keynote

First I’d like to start by welcoming everybody to the new JasperForge. I’m encouraging to you to be an active particant - post in the forums, feel free to start your own blog, even start your own project. This site is all about the community and the more everybody contributes the more we all gain.

I’m up at the Executive Briefing at OSCON 2006 in Portland right now. Tim O’Reilly started out with a keynote on the current state of Open Source and Web 2.0. He highlighted a trend of companies that are open sourcing their framework but building a proprietary application on top of it. Examples include the folks behind Basecamp releasing Ruby on Rails as open source.

I couldn’t help but notice that all the examples Tim gave were companies that had a hosted application and released their development framework into open source. Tim claims that this is an increasingly comment trend for Web 2.0 companies, but I keep thinking his examples are really Software as a Service (SaaS) plays. Perhaps the distinction between the two is irrelevent at this point?

Another topic Tim discussed was open data - the ability to easily get at the data stored in hosted/Web 2.0/SaaS sites and do what you want with it. This is causing problems for many of the application developers, as they don’t really want competitors to be able to take all the data they have collected from their community and use it to bootstrap a new community.

All the open data exmaples were community sites, such as Flikr or Creative Commons. I’m more interested in applying the same thought process to more traditional structured data. There is a huge amount of data in the public domain (collected by both the public and private sectors) that should be more accessible to the public at large. Being able to download a CSV file that is hard to find on a government site isn’t good enough. People need to be able to find the data and then sift through to see it in a manner that is natural to them.

OSCON 2006 - The Ghost In The Machine

The second in a continuing series of posts from OSCON 2006. This time around I’m going to discuss a discussion panel featuring Jim Buckmaster of Craigslist, Chris DiBona of Google and Jeremy Zawodny of Yahoo!, moderated by Tim O’Reilly.

The discussion centered around the use of open source by the three sites and the relationship their companies have with open source. Some of the interesting points:

• Both Google and Yahoo! admit that they make changes to open source projects that they don’t make public. (Calm down, only projects whose licenses allow this.) Chris was blunt, saying they do this for competetive reasons.
• Chris stated the Google has built a ton of their own infrastructure, such as their own build system, their own command line argument parser, etc. Since so much of their code is built on top of this stuff it is hard for them to release as open source, as they have to strip out all this stuff.
• I think this is dodging the real issue - why did they reinvent all this stuff, and if it is so much better than what is in open source why don’t they release it as open source. Can their command line parser really be a competetive advantage? Is their build system more powerful than Maven?
• I was disappointed in the panelists’ view of licenses. They felt that open source licenses can be inconvienent and cause friction. The more I listened the more I felt that they were trying to rationalize not always stickly sticking to the various licenses

Another interesting point made was that when a commercial entity whose focus is an online application releases software as open source, there is no longer any incentive to enhance and maintain that code, as potential competitors may well be using it.

Finally, Jim Buckmaster of craigslist had an interesting tidbit. They are focusing on pages server per kilowatt hour, a stat I’ve never heard used before. They have had problems with being kicked out of co-location sites because of power consumption.