Q: Does that same third-world mindset apply to open systems as well? Is the U.S. open systems movement more of a revolution than an evolution?
The real fervent push for open system really got started in Europe. It originated in North America, but the real force behind it really got strong in Europe. A lot of it is about the user community banding together to do the right thing when suppliers don't do the right thing.
The Apache Web server, for example, was a whole lot of people who built Web sites and got together because none of the Web server providers were doing the right thing.
Q: Are the Europeans a bit ahead in terms of these grassroots efforts, as well? Is Europe faster to adopt new techniques and technologies?
It almost feels like the U.S. took capitalism too far. It's kind of like we compete ourselves to death, so instead of having one good cell phone infrastructure, we have half a dozen really mediocre ones.
A lot of the stuff that has worked well in Europe and Asia is entirely because it has been a government-regulated thing. In the case of NTT DoCoMo, for example, when it first got started it was a semi-regulated, quasi-government monopoly thing. They had to figure out how to make money off of the bits, so they pushed hard to focus on the bits.
A U.S. company, on the other hand, would not want to marginalize it just as a bit-pipe, but would want to be a full-service provider and then try to get into things in which they really didn't have any corporate confidence.
Q: A lot of high-tech companies in the U.S. seem to be moving from a decentralized computing and networking approach back to a more centralized architecture. This would seem to be better from a security standpoint, but does this have any adverse impact on the open source development?
I don't think it makes a whole lot of difference. Certainly, manageability and security are a lot easier when it's more centralized. A lot of open source is about building large clusters, which works really well in a centralized kind of way. A lot of the issues are about the ebb and flow of network bandwidth. When network bandwidth isn't great, you kind of have to put as much out to the edge as you can.
As network bandwidth gets better, it actually works to make stuff more centralized. There's really no one answer that makes any sense without knowledge of the context.
Q: Are software development and development concepts changing to any great extent as we move more toward pervasive and higher speed networks and software as a service environments?
Well, we've kind of been there for quite a few years now. Software as a service hasn't been a technology problem, but it has been a business-model problem. We've been talking about doing software as a service for close to a decade I think.
Q: It's a tough sell?
Yeah. It's a tough sell when you ask if the software as a service that you sell should be external or internal. Do the external providers actually do a good job?
Q: In terms of the technology and network infrastructures, will it be easier for open source development once technologies like WiMAX become widely available?
As networks get better, it is certainly easier to deploy services centrally and to do software as a service, and that's going to work for some things and not for others.
If you have to do expense reports then you don't need a lot of bandwidth for that. In terms of videogames, the network is important if the games are collaborative. However, with most games, most of the application is close to you and no network change is going to make a big difference. So, it's a pretty uneven landscape.
Q: Java has been around for a while now, but how is it changing to adapt to new technologies and dynamics in the enterprise market?
A lot of this ties back to the whole open source effort. We put a lot of emphasis on working collaboratively with the development community, and Java has gone off in a lot of directions form smart cards to supercomputers and there is a great diversity of applications that people build and the interconnect.
The word "application" in my mind stopped making sense a long time ago, because these things have become pretty sophisticated assemblages of components that run here, there and everyplace else. You really have to think of software as an ecology and not just deconstructed components.
Next page: What is your impression of Google?