OpenCola-Have Some Code and a Smile

Distributed searching may save you from the info-glut.

By Paul Heltzel

April 9, 2001

Steve Jobs wooed Pepsi CEO John Sculley over to Apple with the question, "Do you want to change the world, or sell sugar water?" An enthusiastic startup called OpenCola hopes to do both.

A software company that happens to produce a soda, OpenCola uses peer-to-peer searching to find information on the Web. Its soft drink, OpenCola, helps to promote the company.

Peering Up to Search Together

Specifically, OpenCola Folders is a peer-to-peer network search utility, based on the idea of collaborative computing.

As you surf the Web, you drag things you find (and like) into a folder. From your choices, OpenCola learns your interests and crawls the Web for related pages, music or games.

The "cola" in OpenCola stands for Collaborative Object Lookup Architecture. Collaborative, because the software peeks into other people's OpenCola folders and makes choices for you based on what others have found useful. If you don't like what the software finds, you throw it out-and repeat.

The result should be "good stuff that is utterly surprising," says Cory Doctorow, the company's co-founder and chief evangelist. "It's like you meet someone at a party, and you hit it off really well. They recommend things to you that you've never heard of. And those things become the most important things in the world to you."

OpenCola acts as both client and server on your system. As in other distributed computing systems (see Virtual Supercomputers Sign Up for Business), the users of OpenCola each give a bit of their computing power to the good of the whole.

Finding interesting documents is a job handled by the client-servers (or "clervers"), which then break the work into smaller bits, essentially acting together as a distributed supercomputer.

The application should be ready for public release by this summer.

Opening the Folders

OpenCola relies on four distinct components. The first-and most visible to the user-is the OpenCola Folder. You put links-both to the Web and from your own computer-into the folder. A second component, the Relevance Queue, scans the linked pages and builds a set of search criteria.

The third component, the LinkRipper, pulls the links from the documents in your OpenCola Folder. OpenCola crawls the new links and a fourth component, the Relevance Engine, scores the new pages against your criteria. As you approve or reject pages, OpenCola updates your criteria, honing your search.

The software continues to crawl the Web but also looks at other users' OpenCola folders that are similar to your own. OpenCola sends you their links but-unlike peer-to-peer sharing software like Napster-not their files. OpenCola points to Napster's legal woes as proof that distributed searching beats file sharing, at least in the eyes of the courts.

Selling Sugar Water

Story continues below

And yes, the company already publishes a free cola recipe you can download and mess with. With the right ingredients-some oils, a little citric acid, some caffeine and caramel-just about anybody can make the stuff in a tub at home. (Remember, though, "caffeine can kill people in relatively small doses," warns OpenCola.)

While the analogy isn't perfect, the open-source cola gives Doctorow a hook for explaining his open-source software. "We published the formula, and you can download it and make derivative colas. The only requirement is that you tell people what you've done to it."