Hans Ulrich Obrist and Danny Hillis talking about the 10,000-year clock, Richard Feynman, Marvin Minsky, and Alan Kay (Taken with Instagram)
Great evening (if slightly overstuffed with a very interested audience in a modestly sized and not well ventilated space at For Your Art, across from LACMA in LA’s mid-Wilshire Miracle Mile).
Here, Obrist (who has spent many years collecting recorded interviews with some of the world’s most interesting and influential people) talks with Hillis about how he came up with the idea for a parallel-processing computer. A mathematical proof suggested such an approach was impossible.
Hillis ignored the proof (and later figured out where it was wrong) in creating the first Connection Machine, which in the late 1980s and early 1990s ruled in various iterations as the world’s fastest supercomputer before the company went bankrupt in the late 1990s. (Hillis mentioned that a young Stanford student named Sergey Brin would soon adapt his computing approach and go on to co-found a little startup called Google that later had some success).
Amusingly, Hillis said he was recently contacted by the researcher who maintains “the list,” which tracks the world’s fastest computers each year.
When Hillis created his first room-filling, multimillion-dollar Connection Machine in the late 1980s, it was rated at what was then a whopping 70 gigaflops (I believe a gigaflop is equal to 1 billion operations or calculations per second). The list’s keeper wrote Hillis to say he’d recently bought a Mac Powerbook with Retina display and had been able to clock that $2,200 laptop at, yes, 70 gigaflops. Moore’s Law is just merciless.
Hillis talked about hanging with Nobel winner Richard Feynman as they tried to prove that spaghetti always breaks in three, not two, and ended up with a kitchen full of broken spaghetti and a missed dinner.
But most of his conversation was focused on his singular hobby since the end of the Connection Machine: creating a mechanical clock capable of functioning for 10,000 years, or equal to about the entire span of human civilization so far.
I interviewed Hillis back in 1998 or so, not long after his Long Now project was featured on the cover of Wired magazine, backed by the enthusiastic support of interesting people such as Whole Earth Review’s Stewart Brand, musician Brian Eno (who gave the clock its poetic name) and Amazon CEO/founder Jeff Bezos (a major funder over the years). At the time, Hillis had gone to work as a so-called Disney Fellow, working as the Imagineer’s Imagineer under Bran Ferren and with Kay and artificial-intelligence pioneers Minsky and Seymour Papert.
As Hillis put it then, and now, he went from creating the world’s fastest machines to creating its slowest. I’m so delighted, all these years later, to hear an update on the fascinating project’s progress, which has moved to an actual remote mountain site in the desert Southwest (in Texas, I believe), where the inner caverns have been dug and test versions of the giant titanium and steel clock has been fabricated and assembled.
It’s not done, but when you’re working on a scale of 10,000 years, you’ve got a bit of time to get things together.
And, in a lovely coda to the evening, Hillis responded to an audience question by saying that having a long-, long-, long-term vision like the clock feeds is actually a very happy place to be. You understand that humans, for all the terrible things we do to each other and often to the planet, have managed to get through the short-term crises and complications we create and eventually move to a better place.
I take some comfort in that. We ARE a little further down the road than we were 10,000 years ago. With any luck, our descendants will be even further along 10,000 years from now.