The entertainment industry will be thrilled: Computer scientists say they can slash the time it takes to download from peer-to-peer networks.
But there's a catch: It wouldn't make much difference for popular downloads.
According to this article, more than half of all downloads from peer-to-peer networks fail. Plus, it can take 24 hours to download 100 megs.
David G. Andersen, an assistant professor of computer science at Carnegie Mellon; Michael Kaminsky of Intel Research Pittsburgh; and Purdue University grad student Himabindu Pucha developed the new approach, called similarity-enhanced transfer or SET. They claim it could reduce a five-minute download to about two minutes.
SET's complicated, so I won't try to reproduce the explanation here. If you're really curious, you can download a .pdf of their research paper for free. Basically, SET is a hybrid technology that matches statistically similar chunks of data. It speeds things up by pulling that data from systems with the most resources for uploading the chunks.
The research group plans to use SET for sharing research papers and software. But, since they're releasing the source code, we'll no doubt soon see other applications - and we hope they'll be more ambitious than piracy.