Intel stoked the fires of file-sharing hype last week at its annual developer conference when CTO Pat Gelsinger declared that "peer-to-peer computing could be as important to the Internet's future as the Web browser was to its past."
But before you start trying to figure out the back-end technology of Napster or its decentralized file-sharing clones like Freenet and Gnutella, or before you join Intel's new peer-to-peer working group -- you might want to consider not just the message, but also the messenger. Intel is not some independent analyst, nor a disinterested Walter Mossberg. Rather, it's a chip manufacturing giant whose success depends on convincing the world to buy newer, faster computers. Moore's Law dictates that chip processing power will double every 18 months, but the best way to make us buy the new computers (with Intel inside), is to convince us that we need them.
That's where peer-to-peer comes in. First, consider file trading, which taxes your computer far more than some routine multitasking between your browser, e-mail and PowerPoint. If you use Napster, the more songs you save, the more work your computer has to do to find and play them; and every time someone grabs a few Beastie Boys' tracks from your hard drive, the computer acts like a server, processing the request, the search and certain aspects of the download. It all requires computing effort -- and could be speedier with more processing power.
But beyond the performance issues, widespread file-sharing and distributed computing bring up questions of interoperability and security, which could be addressed in new computing architecture and new chip designs. And wouldn't it be great for Intel if -- hooked on file-sharing and eager to shelve our plans for a new company server so that we might instead invest in desktop machines that could handle our personal work as well as a piece of the old server's job -- we all felt the need to buy a new computer with a flashy new processor?
There is, of course, nothing wrong with drumming up business, even if it's not exactly what you expect from a CTO purportedly sharing his technological vision with developers. And Gelsinger may be right when he says peer-to-peer networks may represent "the next computing frontier." Surely, the 22 million people who have downloaded Napster's software point to peer-to-peer's potential. And distributed networks like SETI@home, which use a million home computers to process signals from outer space as it looks for signs of intelligent life beyond Earth, could use more computing power.
But by making technological predictions that seem to be equal parts wishful thinking for Intel's future and reality, Gelsinger diminishes his own credibility. Come on Pat, we may not be as fast as your chips, but we're quick enough to recognize a sales pitch disguised in those predictions for the future.