New Architect: What's your pet peeve about UI design on the Web?
Tim Bray: First, the lack of progress. Today's Web UI doesn't look much different than it did in 1994. Have we learned nothing? Second, the huge gulf between the visual access your desktop provides to your hard drive and the "query, hit Enter, look-at-lists-of-items" basis of most Web apps.
NA: Is the Web to blame for bad UI design? Or does the problem predate the Net?
TB: I think in general we took a step forward when we migrated from custom GUIs to the Web back in the '90s. The Web keeps things simple, and the notion of pieces of text doubling as control elements is very sound. Having said that, there are excellent pieces of desktop software UI and horrible examples of Web UI.
NA: Give me an example of a good UI.
TB: The interface at www.time.gov is one example of how to do things right. It uses a very small amount of screen real estate to provide a substantial number of choices in a way that is instantly obvious. What's not to like? I'm not going to point a finger at all the ways to do things wrong, that's Jakob Nielsen's job.
NA: What are your current projects or areas of research?
TB: The last three years of my life have been more or less totally dedicated to this problem. We are using the graphics engine in every modern browser to display visual maps—real maps, like cartographers draw—of complex data sets, so that access to shared information can be visual.
NA: There are a number of products out now—like Flash and, to a lesser extent, Java—that let designers produce more sophisticated interfaces for online applications. Do you see any of these furthering the cause of good UI, or are they adding to the problem?
TB: I think they are all steps in the right direction, but I think they share a fatal flaw—poor integration with the browser. We don't want to give up the basic browser interface (forward and backward, hyperlinked text). We do want interactivity and better use of visual tools. DHTML is no longer a fashionable buzz-phrase, but it's still the best way forward.
NA: There has been some difficulty with DHTML standardization between the various browsers, and particularly across OS platforms. Might this account for the appeal of Flash?
TB: As you say, there "has been," but the level of standardization has become immensely better in recent browser releases. If you're willing to bypass support of the Netscape 4 family, pretty much anything you're apt to find on a user desktop can be driven in a reliable way with only a very modest amount of browser sniffing. I repeat that Flash is too high a price to pay—the success of the Internet has been built around open, textual, non-proprietary message interchange.
NA: Has the nature of the browser and the HTTP protocol doomed UI on the Web? Or is there anything designers can start doing right now to improve usability?
TB: No, there's no underlying problem with the browser/HTTP architecture. We need to start sending XML to the browser and executing more code on the desktop, but we need to do it in a standards-based way and not have to rely on somebody else's proprietary engine.
NA: How might this be achieved? Presumably there would need to be some sort of standard language that could be parsed on the client side.
TB: It's here today; it's called JavaScript. Also in modern browsers, which expose a W3C-compliant Document Object Model, you can make the page dynamic with a modest amount of code in any programming language with a DOM binding: Java, C++, Perl, you name it. Browser development is not progressing very fast these days due to the fact that we've got a Microsoft monoculture. But I think that will change, because we're going to have a lot of non-PC devices using the Web, and then there's pressure from Gecko via AOL.
Still, most modern browsers are very standards-compliant and you can accomplish remarkable things with good clean JavaScript and the DOM. I encourage designers to make their sites more visual and interactive, and at the same time to future-proof themselves by playing by the rules.
