Paul, John, and I went to this talk together. While the material was sound, the tone of the talk was a little to "us vs. them" for me. I was also trying to convince the Eclipse guys that Jython integration was a good idea. Fortunately, they were open minded enough to skip the attitude. John pointed out that if you used jython as the scripting language for Eclipse, then you could wire Eclipse as the debugger interface for Jython, assuming that Jython exposes some kind of debugger API. I don't know if it does, but that would be cool.

Paul was my second line manager when I worked at IBM. Now he's Mr. Eclipse, which cool. It was even cooler that he still remembered me after all these years. We got to catch up a bit on people that we know and so forth. We started talking about Eclipse, and he introduced me to John Wiegand, which was also cool. I told him about my experiences with Eclipse in the field, and how I've given a number of impromptu demos of Eclipse at SeaJUG. I talked with the two of them about two issues: 1) how to explain Eclipse to hackers (emacs/vi hackers in particular). I related how it takes a little explaining to get my hacker friends to understand why Eclipse is so powerful and has the potential for so much more. 2) Integrating Jython as a scripting language / interactive shell for Eclipse. This would be really cool because you get get at all the Eclipse internals so you could write scripts that would perform series of actions. More on this in the next post. It was good to catch up with Paul again, and I appreciated the chance to talk about the Eclipse project goals in person. I've been concerned about the level of outside participation in the core Eclipse components, and the guys told me that they want to encourage it, but that its hard code for people to get into because its the internals of the compiler. They have been getting good contributions for SWT.

He pointed out some examples of things that happened during the hardware paradigm shift. Companies from the old paradigm came into the new paradigm but tried to retain aspects of the old. In hardware an example of this was Compaq - they used the emerging commodity hardware base, but they also tried to add their own proprietary hardware. Eventually they gave up. In the software shift, the corresponding analogy would be IBM (WebSphere) and Apple (MacOS X), who have embraced open source but are also doing proprietary extensions

He also rehashed part of his ApacheCon Keynote "Watching the Alpha Geeks", and talked about open source killer apps being server side killer apps (Amazon, Google, etc). He pointed out that these apps breach some of the common assumptions about open source in terms of licensing (there are no binaries or redistribution). The value in these apps is not in the software but in the data. The claim is that we need to move beyond licensing and he offered three directions:

1. Commoditization of software - He reported that Amazon achieved a 10x cost reduction when the switch to linux. He also reported that eBay switch from Unix to .Net with no visible changes to eBay users, demonstrating that whole technology stacks (J2EE/.NET) are commodities which are easily replaced -- I'm not so sure about that.
2. User Level Customizable Systems and Architectures - Applications need to be updated constantly. He digressed at this point to claim that his requirement was a strengh for dynamic lnaguages which allow dynamic update to happen
3. Network enabled collaboration - His claim is that Usenet is the mother of open source, and points out that the software which was written and exchanged via UseNet (comp.sources.*) was not triggered by licensing, but by the existence of a network communication medium (UUCP/NNTP). He mused that 'patch' may have contributed as much to open source as anything else. His claim was that given a sufficiently large networked development community, open sourced like behavior will emerge.

His closing thoughts were: the platform for the future is web services, open source stuff will be aggregated and sold by subscriptions (the cable industry found that pay-per-view was not as profitable as subscriptions, and that people wanted to subscribe to packages of channels as opposed to single channels). Look for hidden open source business models (where open source is embedded in something else - like the BIND / DNS monopoly).

Open source is kind of bill of rights for software developers and users. What is the corresponding bill of rights for data (web services)?