AZA RASKIN, user experience chief, Mozilla Labs: There's two parts to that question; one is localization and internationalization, which is a problem that you need to think about and something that, again, Mozilla really helps with because we have people all over the world. I went to Rome a few weeks ago, got there late, we decided to put up in the very last minute [a message saying], "Hey, I'm in Rome, let's meet up," and people traveled 200 miles the next night to come down and have a discussion. These people in the community for [Mozilla] localizers, are way more passionate than we've ever seen anybody else about a project. Localization and internationalization is really important, and something we're looking at. In fact, somebody's already written a Japanese edition of Ubiquity.

The other side of your question is, a lot of really smart people spent a lot of time thinking about these problems, and that's very true. So there's two answers...One, yes, we can stand on their shoulders. Two, the open source world is arguably the largest and most advanced engineering department in the entire world. So there's a lot of leverage that we can hold here. And three, we're not actually attacking natural language processing full-on. We're being clever about it. We're cheating everywhere we can. We know what sorts of actions people take often, so you can start making really smart guesses. We're not actually solving the problem of all natural language; you can solve the problem of just a command. And as the community grows, we're going to see that expand. We're starting someplace very pragmatic, which is the Mozilla way -- make something to show that it works -- and then expand out there to hit your goals.

SCOTT FULTON: Well, let's think of the macro-history of the development of user interfaces. Realize that over the last couple of decades, we've really been in kind of a standstill in terms of what we've been able to represent graphically. We're getting "buttoned out" here. Even your other projects at Mozilla Labs besides Ubiquity have been looking into, can we make this simpler by getting rid of some of these tabs, and moving some of these buttons out of here, and creating controls that have more implicit functionality to them? Can we redefine things visually? Well, it seems that as we approach the problem of solving things lexically -- one command at a time, as you say -- don't we run the danger ten years from now of being "commanded out" here? Just having a litany -- more than a glossary, but an encyclopedia -- of other people's ways of processing the same commands, and eventually won't we almost be forced to attack that lexical problem full-on?

AZA RASKIN: I think a lot of what this comes down to, lexically speaking, is that there needs to be a tight feedback loop. When I'm talking to you -- this phone conversation may not be the best example, because it's hard to get immediate feedback of what you're thinking from your facial gestures -- but there's a lot of feedback that lets me know when I'm on the right track and when I'm not. There's a lot to learn in interface design from smart, bite-sized, immediate feedback, which helps. One of the big problems with GUIs is that it's hard to do sort of progressive enhancement. You either click the button or you don't; there's no generativeness in that process. Whereas language is a much more generative thing.

If you look at the history of literature, people are still saying arguably new things. Language is much more expressive...In five characters, you can express 26⁵ different things. How can you possibly reach that range of expressiveness with buttons, a billion things you could choose with buttons? That's very difficult.

• Raskin's next Mozilla experiment: an even smarter address bar by Scott Fulton
• Mozilla: Tell us how you see the future of Web browsing by Scott Fulton
• Concept video from Mozilla Labs gets developers thinking, talking by Scott Fulton