Is it time to scrap the Internet and start over?

More and more, Americans are treating their PC as a second TV. Few are ready to place a couch in front of their monitors, but millions are plopping down in front of Internet services such as YouTube, Joost, and television network websites to watch online shows and movies.

Streaming video will surge from 7.3 percent of all US consumer Internet traffic in 2006 to almost 33 percent by 2012, according to the market research firm IDC of Framingham, Mass.

But remember, that's just a percentage. Along with video's increasing share, total Internet traffic is expected to double every two years, industry analysts say.

Some warn that this rush will overwhelm service providers if they don't prepare for it. And the forecast for an ever-rising flood of data has some asking, is the Internet outdated?

"That's one way to put it," says Larry Roberts, who, in 1969, managed the Pentagon's APRAnet, the precursor to the Internet. "Another is that it's insufficient for the new kinds and new scale of today's transfers."

While some disagree with Mr. Roberts's characterization, the connection speeds reaching American homes are certainly behind those enjoyed by Japan, South Korea, and Sweden and could potentially limit Americans' online entertainment choices.

The Internet is perfectly tuned for e-mail, says Roberts. But 40 years ago, he and the many others who helped nurture today's commercial Web never imagined, nor planned for, streaming high-definition television shows to travel through the wires.

But that's where online usage has headed. Nearly 16 percent of American households with Internet access now watch TV shows online, according to a report released last week by TNS and the Conference Board, two research groups based in New York. And the online audience for entire episodes has doubled over the past year. Now, with the new TV season, networks are making even more of their prime-time shows available to Web viewers.

Online streaming has a lot of perks. The episodes are on-demand, so you don't have to worry about missing your favorite show. You watch them while they download, so you don't have to wait an hour for the whole file to finish transferring.

But unlike shows downloaded from Apple's iTunes store, any hiccup in Internet service can leave characters hanging in mid-sentence – or drag action scenes to a crawl.

"This is an important time because it's just before video is about to become a big problem," says Roberts, who is now chairman and founder of Anagran in Redwood City, Calif. Last month, this Internet hardware firm unveiled a new "flow router" that he says can better manage today's Web traffic. Instead of treating every one and zero as the same, the Anagran router gives priority to important data – perhaps streaming video – over smaller, less-urgent files such as e-mail.

Searching for ways to work around today's Web, British technology company CacheLogic proposes a hybrid system that would allow videos to flow into your computer from different sources. First, a video distributor would store multiple copies of a file on computers around the world, so that wherever the viewer is, the video doesn't have to travel far. This practice is already well adopted by large distributors. But CacheLogic wants to supplement this method by drawing from computers that have already watched the video, adding a legal peer-to-peer file-sharing element to the mix.

But bigger changes have to occur before computers can truly rival televisions. In particular, greater investment to deliver video to the masses. For example, 19 million Americans watched the season première of ABC's "Desperate Housewives" on TV this year. If all of those viewers had seen the program on the Internet in HD quality instead, it would require 4.5 terabytes of Internet capacity each second. That's triple the amount used on the entire Internet, CacheLogic reports.

While HD quality program files are massive, the problem delivering them to customers via the Web is not necessarily available bandwidth, says Len Bosack, cofounder of Cisco Systems and now chairman of his own start-up, XKL, in Redmond, Wash. Before the dotcom bubble burst in 2001, companies rushed to roll out millions of miles of fiber-optic cable all over the country. Plenty of that capacity remains dormant, he says.

Last month, XKL released a way for businesses to bypass telephone companies and tap directly into those cables, each consisting of about 200 strands of glass bound together.

"There's enough bandwidth in one [pair of glass strands] to allow the entire adult population of Los Angeles to talk to the entire adult population of New York," says Mr. Bosack, who himself was talking over an Internet phone connection. "That's just a piece of one cable. So what do you want to do with the rest of the capacity?"

The core "backbone" of the Internet is not the issue, most agree. But there's a huge discrepancy between how fast information gets to your city and how fast it enters your home.

Online speeds are only as fast as the slowest link. And many Internet service providers balk at the massive price tag attached to laying fiber-optic cables to every neighborhood doorstep.

Verizon FiOS, which trumpets itself as the only US "fiber-optics network straight to your home," hopes to reach 9 million houses by the end of the year. In all, the network will likely cost $18 billion. The telephone company aims to make up the investment over the long term by luring new customers with its faster speeds.

Still, capacity at a local level "has been a challenge, especially in the US," says Vinton Cerf, who won the Presidential Medal of Freedom in 2005 for his contribution in developing the Internet. In an e-mail, Mr. Cerf said he expects that as technology improves, Americans "can look forward to gigabit-per-second access speeds – and these already exist in countries like Japan, South Korea, Sweden, Hong Kong, Singapore, and elsewhere."

In particular, Japan enjoys broadband speeds that are 12.7 times faster and 12.3 times cheaper than the average connection in the US, according to the Information Technology and Innovation Foundation, a public-policy think tank in Washington. Analysts attribute Japan's advantage to two factors: population density – the country doesn't have to worry about wiring wide-open states such as Montana or Nebraska – and a larger willingness to swallow costs in the name of innovation.

In August, Japan's Communications Minister Yoshihide Suga announced a plan to imagine and develop a brand-new Internet – one that might replace the current Web. The American National Science Foundation (NSF) launched a similar "clean slate" endeavor in 2005.

"We're thinking, 15 years from now, what do we want the Internet to look like?" says David Clark, the effort's chief evangelist and a key architect of the Internet in the 1970s and '80s. The NSF is approaching its third wave of grants for researchers to develop and discuss long-term solutions to handling digital information.

"I'm not talking about speed or bandwidth," he says. "The private sector can handle that. We're looking for ways to make the Internet fundamentally safer and more manageable. That might require making an Internet from scratch. It might mean we'll have two parallel Internets. We don't know yet."