SAN FRANCISCO (Business 2.0) - Dmitry Shapiro wants to reinvent television – and he's counting on help from the millions of fans of Internet video.

Shapiro's formula is part Napster (Research), part Amazon.com (Research), and part YouTube, the popular site where users view each others' homemade videos. By combining peer-to-peer file-sharing technology, collaborative filtering, and the popularity of viral video clips, Shapiro is hoping to create a killer app for Internet video.

On Wednesday, Veoh, Shapiro's San Diego-based startup, launched a Web-based version of its video-sharing software to compete with the likes of YouTube, iFilm, and vSocial, which each let users watch, post, and share videos online. Whether or not Shapiro ever catches up to this already-crowded pack, he does have a few ideas that may very well point to how we will all one day be watching TV -- and not just over the Internet.

Veoh started last fall with a peer-to-peer video file-sharing application. It attracted all of 30,000 users who swapped a total of 250,000 files -- a test run that showed the limits of trying to get users to download software. Now that Veoh's on the Web, Shapiro is betting that it can become a serious distribution platform for Internet video.

While YouTube and other video sites may serve up millions of free short video clips a day, often the videos are jerky or low-quality, even over broadband connections. But Shapiro does a few things differently. For one, he has a way to deliver DVD-quality, and even high-definition, videos to your laptop. For another, his system allows the people who create the videos in the first place to either charge for downloads or to share in advertising revenue. And finally, using advanced vector math, he plans to employ a sophisticated recommendation engine that will target shows and advertising to you.

Veoh's success is predicated on its ability to attract the best video content from both amateurs and professionals. It still has a long way to go on that front. Right now, on Veoh's beta website, there are a few amusing user-created shorts featuring Lego characters, a bunch of skateboarding clips, and some soft-core bikini videos.

More interesting is what you can do with the videos. You can rate them, embed them in your blog, create your own curated series of Veoh shows, or download higher-quality versions of them. To do so, Veoh has users download its file-sharing software, which can download video much faster than a Web browser.

Veoh is not just relying on users to contribute content, it's also relying on their computers to distribute that content efficiently. Bandwidth is expensive, and serving up high-quality video can cost 65 cents per gigabyte or more. By using peer-to-peer software and having users download the video from other users' computers, Veoh can do it for a few pennies per gigabyte.

Shapiro is hoping to persuade Hollywood to open its vaults, before their TV and movie content ends up on illegal file-sharing networks. "The big media companies are clearly nervous," notes Shapiro. "They realize they need to do something." Some TV shows are now on sale at iTunes, to be sure, but a much greater number are on file-sharing networks.

Veoh offers what could be a compelling business proposition. For one thing, it has a mechanism to pull copyrighted material off the network to prevent piracy. Any video publisher on Veoh can choose to charge for downloads and keep 70 percent of the revenue, with Veoh getting the rest.

Or publishers can offer their content for free and split the advertising revenue, with Veoh taking a larger cut since it plans on selling the ads. "We've modeled that most of our revenues will come from advertising," says Shapiro.

To sell advertising, Shapiro is using what he calls "crazy math" from Veoh's chief scientist, Ted Dunning, who created the music recommendation engine for what's now Yahoo (Research) Music. If a critical mass of people start using Veoh, Dunning's software will track what they search for, what they preview, what they download, what they watch, how many times they watch it, what ratings they give it, which friends they send it to, what those friends are watching, and so on.

Based on all of that data it will start making recommendations. Sometimes those recommendations will be videos. But sometimes they will be advertisements. And viewers might not even know the difference.

"We look at advertising as content," says Shapiro. "Targeted advertising is potent content." As an example, he shows me a clip of a blogger driving around in a Porsche in Germany. It's not an ad, but it might as well be one. What if, suggests Shapiro, after viewing it, you then got the option of viewing longer, official Porsche driving videos?

If users could control the advertising like that, they might watch more of it. And advertisers might pay more to reach them – and even create videos just to reach that demographic. If Shapiro or one of his competitors (or Google (Research)) could pull something like that off, well, then I'd say that would pretty much qualify as reinventing television. Wouldn't you?

___________________________________________

For 25 hot startups, click here.

For more from Business 2.0, click here.