Just a reminder that I'll be at MoonBase ConFusion at the Detroit-Troy Marriott in Troy, Michigan this coming weekend. I don't think it's too late to get a room, and you can definitely still sign up to attend! I'm committed to do a reading Friday evening, panels on "Remaking Humanity" and "Singularity or Rapture?" on Saturday, an Evolution Q&A on Saturday, and a discussion of evolution with kids on Sunday. Oh, and there's a casual coffee hour with me on Sunday morning. From the schedule it sounds like there will be parties going on at all hours, too, so I'm hoping no one shows up for that one so I can bogart all the caffeine. I'll need it.
I'm also bringing along that ferocious grammar Nazi, Skatje. We'll both be blogging the con in addition to the talking and the wandering and the partying and the spectating and the hey hey glavin.
- Log in to post comments
Food for the Singularity/Rapture fire: Singularities and Nightmares by David Brin.
Last spring, I was playing around with some algebra and figured out that you can get a quasi-Moore's Law from two competing Poisson processes. If you say that "significant events" happen roughly uniformly throughout history, at a constant average rate, and if the fossils left behind by those events also decay with constant probability per unit time, then it's not too difficult to show that the average waiting time between known events increases exponentially back into the past. It's not exactly the graph that Kurzweil loves to trumpet, but it's awfully close: the illusion of accelerating change can arise from forgetfulness.
The other day, I returned to this model to see if I could jazz it up in any way. (Hey, it was either that or actually listen to other people talk.) The thought struck me that I might be able to get graphs which look exactly like Kurzweil's "Law of Accelerating Returns" from a simple requirement of scale invariance. It works like this:
Suppose you make a list of the 100 most significant inventions of the twentieth century and plot them on a timeline. Then you take a step back and make a list of the 100 most significant inventions of the millennium (1001 CE to 2001 CE), and plot them on a second timeline. If you postulate that the overall distribution of events should look statistically the same before and after this "coarse-graining" procedure, then — I believe — Kurzweil's "Law" follows very naturally, even though scale invariance is a rather uniformitarian postulate to make.
Anybody who works in network theory would probably think along the same lines, since for them, power-law behavior is naturally tied to notions of scale invariance.
Let me add that I don't necessarily agree with everything David Brin says, in the "Singularities and Nightmares" essay or elsewhere, but I do feel a mild obligation to give a little free publicity to anybody who calls me — for the record — "a very bright fellow". I also have open admiration for anybody who can infuriate subcultures as efficiently as he can: insulting Star Wars, saying that BASIC wasn't all bad, etc.
Dammit. Could you, like, give me a ride from MSU in East Lansing?
There's a tenspot in it for you or the kid.
Sorry, I won't have a car -- we're flying in.
PZ, I'm looking at the Con site, and they've misspelled your name here:
http://www.stilyagi.org/cons/2007/programgrid.php
Oh, well. At least you're not Steven Hawkins.
How fun! I am a A2 native and though I am just finding out - will DEFINITELY be there!!!
PZ, you are raising that girl right!