January 12, 2005

How Soon We Forget (Technology)

xmas-tree.jpg (187837 bytes)I just got home from a very snowy (nearly 3 meters of snowfall in 3 days) weekend up in the Sequoia National Park - where the really big trees are.  The world's tallest species of tree grows along the California coast, including around my house in Santa Cruz.  But the biggest species of tree, in terms of mass, lives up in the Sierra Nevada, particularly in Sequoia National Park.  And when I say "big", I mean big.  Here's a photo of my wife and her Christmas tree.  It probably sprouted while Rome was an empire.  And it's not nearly the largest, or oldest, tree in the vicinity!

These trees exist because they (or rather their ancestors) are highly successful innovators.  When these trees come up with a useful evolutionary trait they don't shoot themselves in the foot (root?) by discarding the new trait simply because it might change the status quo.

On the other hand we humans tend to institutionalize the old and endow it with an enormous power to resist change.

When I got back from the mountains (and snow) I read with dismay of the continuing Media-Luddite attack on BitTorrent.

BitTorrent is the latest target in the sights of the movie and recording industry.  I'd hate to see BitTorrent be squished - I routinely use it to download open source distributions.

BitTorrent is an interesting bit of technology - it operates through a kind of mutual assistance between receivers: a receiver that has obtained a file block from the originator makes that block available to other receivers, thus reducing the burden on the originator and also spreading the traffic load across the net so that there is no traffic hot-spot focused on the originator.

In the IP multicast world reliable multicast streaming and file transfers have been around for decades, including protocols that use mutual assistance among receivers.

But IP multicast has become a lost technology - the MBONE, a worldwide multicast overlay on top of the internet, has faded.  And cross-ISP dissemination of IP multicast routing information never really happened.

There has been talk about the development of new media distribution networks based on BitTorrent.  It would be a much more efficient use of net resources if that media distribution were to occur using IP multicast.

Could it work - yes.  How do I know this?  Because back in 1995 we built it and it worked.  We could efficiently move DVD quality audio/video over the net, even over multi-vendor, multi-provider networks.  And yes, even in that earlier and more innocent era we did try to build-in mechanisms to honor the rights of copyright owners.  (By-the-way, the first movie we watched over our lab net was BladeRunner off of one of those now ancient LP sized video disks.)

IP multicast is not the only lost technology.  A once promising technique for reliability and security, capability based hardware and operating systems, also seems to have faded.

I have my own favorite lost technologies:

  • A video conferencing system that begins with an exchange of photos of the participants in a set of standard poses.  Once the conference begins, rather than actually moving images of the participants, morphing directives are sent so that the pre-exchanged poses can be morphed in real-time to mimic the actual motions of the people.  The system could be tweaked so that a participant could give the morphing commands a kind of stereotypical inflection: in the Italian dialect the morphing would be extravagant while in the English dialect the morphing would be reserved.

  • A resource advertisement and detection system that uses "network pheromones", a kind of packet containing resource advertisements that is wafted, aggregated, and relayed around the net.

So what are the points I'm trying to make?

First: Innovation isn't just the art of the new idea, it is also the resurrection of the old and seemingly lost idea.

Second: New ideas must be given a fair chance to succeed and not be nipped in the bud simply because some pre-existing interest claims that it might be harmed.  (See my proposed First Law of the Internet.)

Posted by karl at January 12, 2005 1:23 AM