Already, the latest Web browsers, including Firefox 3.5, Google's Chrome, and Apple's Safari, allow you to play video directly inside them, without the need for video-player plug-ins. This trend toward media-rich browsers will continue. The next Firefox browser will be able to play 3-D graphics, said Chris Blizzard, director of evangelism at the Mozilla Foundation (makers of Firefox) this morning at Technology Review's annual Emerging Technologies Conference (EmTech@MIT). With underlying software now able to run 30 to 40 times faster than in the past, "we are starting to see the pieces come together," he said. "This is something that is going to be delivered in Firefox, adding real-time accelerated 3-D rendering to the Web."

Among other things, this could allow 3-D video games based on common standards to move to the Web, threatening today's PC-based gaming market. (Google is also working on adding 3-D graphics to Chrome.) But for such transformations to happen, a significant fraction of Web users would have to be using the newest browsers, something Blizzard cautioned could take several years. "The most depressing thing is that the most widely used browser is still IE6," he said, referring to Microsoft's Internet Explorer Version 6, which was released eight years ago and does not support the latest Web technologies.

Blizzard also demonstrated the kinds of video technologies that Web developers can now easily build inside browsers--including adding face-recognition software atop a live video feed and then putting the recognized person's Twitter feed in a box over his head. This kind of creativity will become easier, and spread more widely across the Web, with the wider adoption of open-video standards and looser licensing of proprietary video archives. In general, rapid advances in browser technologies will allow a "much richer experience for users on the Web," Blizzard predicted.

This March, a company called OnLive promised a gaming technology that seemed almost too good to be true. The company said it could deliver graphics-heavy video games over the Internet to any computer or to a miniconsole hooked to a television. This includes games such as the first-person shooter Crysis, which is normally beyond the capabilities of anything short of a multi-thousand-dollar gaming machine.

Today at Technology Review's EmTech@MIT conference, OnLive founder and CEO Steve Perlman presented a live demo of the system in action.

OnLive has met with skepticism from hardcore gamers. The big question is whether the system can transmit high-end games over the Internet without serious lag, and many have said it can't be done. OnLive is currently in an open beta, which involves testing its technology on a variety of real networks and computers.

Though OnLive has developed its own compression technology, Perlman says that this is "just one piece of a complex problem."

The main issue, he suggests, is dealing with real-world network conditions. The company has spent the last seven years in stealth mode learning to do just this. Years ago, Perlman says, OnLive's technology worked perfectly under ideal network conditions. Since then, a lot of work has gone into addressing less-than-perfect conditions.

When streaming something like a video, a computer builds up a buffer to protect against network problems. The buffer buys some time to check whether the stream is flowing smoothly and to ask the server to resend any information that gets lost or corrupted along the way. In the case of a video game, which is inherently unpredictable, Perlman says that such a technique is out of the question.

Instead, OnLive's system uses perceptual science to keep the gaming experience smooth. The company's algorithms adapt what's shown so that it seems to be a complete image while the screen is moving, even if it wouldn't look that way if the picture were still. This allows some leeway for network hiccups. "Each frame may not look good, but we always deliver the data," Perlman says.

The company plans to launch to the public this winter.

Yesterday at EmTech's "From the Labs: Cool Innovations" session, professor of computer science Holly Yanco from the University of Massachusetts, Lowell, discussed her robotic wheelchair project. She first demonstrated the difficulty of using a standard robotic arm attachment for wheelchairs by showing an over screen shot of complicated joystick instructions, which, she pointed out, many people don't want to have to learn in order to command a robot to reach for an object. Instead, she is combining camera vision with touch screen technology, so that a camera will take a shot of objects in front of a shelf, for example, and display them on a touch screen. The user simply touches the object she wants on the screen and Yanco's software lets the robot reach for it. This intuitive approach, she says, will make robotic assistants more useful for people. "My students are very inspired by video games," says Yanco. Just as in video games, a more intuitive approach to the joystick tends to be more successful and result in a more enjoyable experience for the user.