InformationWeek: How has security been handled in Groove? At a recent security show, there was a lot of talk about the risks of using public wireless networks.

Ozzie: There are public domain tools you can download, sniffers from Internet sites. When you're at a conference or you're in an Internet cafe, you can look at the packets going by and see people's E-mail, see what Web sites they're going to, etcetera. It's pretty scary when you actually see it.

Some of it is technology but some of it is just administrator or user complacency--meaning the software is generally not designed to make it easy to figure out how to protect yourself, how to get it encrypted. Or it's very costly. Many corporations use these VPN systems, but they're very hard to use and they're very expensive to deploy.

With Groove security, we had a very basic belief back when we started working on this in 1997 that if you're going to make something secure, it has to be secure even in light of user indifference. We call it complacency-immune security.

From our vantage point, if you download the product, install it, and just start using it, we should never ask the user questions about keys and do you want to encrypt it or not. Everything should just be automatically encrypted that leaves your PC or even that gets stored on your PC's hard disk. So if you lose your laptop, you don't have to worry about having lost the data. And this has been very, very important to our customers. This is why we have an extremely healthy government market for our product, both on the defense side and in other areas of the public sector; it's very rare for them to find a product that not only works to bring people together across organizational boundaries, but also does so in a secure-by-design manner.

InformationWeek: To what extent has going with a peer-to-peer architecture benefited Groove?

Ozzie: It all started with the fact that when I was with Lotus working on Lotus Notes, which was a server-based product, we had a fundamental problem. Whenever one of our customers wanted to deploy a Notes application outside their organization for a business partner, that IT organization had to work with the other company's IT organization in order to agree how to securely setup things, how to exchange digital keys, how set up secure communication channels, how to authenticate users, and so on.

I came to the fundamental viewpoint that it wasn't possible to do what we wanted to do with server-based software because you can't control another company's IT organization. Different companies pick different directory technologies, different operating systems, all sorts of different infrastructures. And it's rare that you can mandate your IT organization's choices on another IT organization.

So, basically, we stepped way back and said, what is it about the technology that would be essential to getting it in the hands of people working across organizational boundaries, and that is, essentially, go all the way to peer-to-peer, and have the client software adapt to whatever infrastructure it happens to be on.

With Groove, because it's got this peer-to-peer architecture, I can download it, I can use it behind my company's firewall. You can be using it behind your company's firewall. It essentially adapts to the different companies' network and security environments, and it tunnels all of the information encrypted, in an automatic fashion. Peer-to-peer gave us the ability to be secure across organizational boundaries, and it also makes it very agile to different networks that are out there today by adapting to whatever network you happen to be connected to.

InformationWeek: Have we moved beyond the desktop metaphor to a shared desktop?

Ozzie: I guess what I would say, at a very high level, is I'm a very big believer in the rich desktop. We started many years ago with a belief in mainframes and decentralization of systems. Then we went through a PC era where end users were extremely pleased with the empowerment they got by having software on their desktops.

Then in the Web era, the pendulum swung back again to centralized control and centralized app deployment. And that was very good for many reasons. Particularly in enterprises, you get economies of scale by doing things in a centralized manner.

But I think the pendulum has swung back, or at least into the middle, where people are realizing that decentralization, in a world of ubiquitous networks, in a world of highly mobile individuals, this concept of decentralized systems and decentralization is of increasing relevance. And it's of relevance in our case to communications and collaboration between people. In the telecom world, I don't know if you've every heard of Skype, but that's an example of how decentralization and decentralized architectures are impacting telephony. Ubiquitous networks and rich clients can be very, very, very valuable.