As processors become faster and memory becomes cheaper, software becomes correspondingly slower and more bloated, using up all available resources.It's truly hard to believe that there are still Robot Cultists out there who fancy Moore's Law is going to spit out a Robot God and end human history in "The Singularity," but sadly, so sadly, there are. Far from an acceleration of accelerating change, the computation-multimedia-industrial complex has looked to me to be cranking out stasis for landfill rather than building a toypile to tech-heaven for a long time now.
Monday, September 26, 2011
Learning from Lanier's Inverse Moore's Law
From his Half A Manifesto, now well over a decade old:
Subscribe to:
Post Comments (Atom)
13 comments:
You're one of the good guys Martin, but look how language wants to make you vulnerable to techno-transcendentalzing razzle-dazzle even though you aren't...
To wit, "supporters of the Singularity" -- just spend a few moments contemplating that string of words. "Supporters" of what, exactly, of what does there support presumably consist?
"Kurzweil's description of genome sequencing improvements is accurate: it kinda followed an exponential increase from 1990 to the early 2000s, even beating the early estimates of how long it would take to fully sequence the human genome. But 10 years later, we still don't have personalized genomic medicine. We still don't understand a lot about the genome." The fudge is signaled by "kinda" -- even before the evidence of "10 years later" came in we had plenty of reason to dismiss the framing as "exponential increase" as a perspectival effect mobilizing transcendental literary frames for their intuitive plausibility and peddling this as "sciency." To retroactively re-write this in the image of hypothesis failing the test is actually to give it far more than its due -- Kurzweil was always indulging rhetorical fun and games more appropraitely responded to on rhetorical than "technical" terms -- he was framing data in the service of a scam not contributing foresight via analysis.
"we already have supercomputers with the computational capacity of the brain" -- I'm not sure that word "capacity" means what it seems to want to mean in that sentence! You actually make the substance of that point later, but look what you concede at the level of framing.
Futurology is a discourse making normative claims not factual ones, its effects are promotional and subcultural not scientific. It is crucial to grasp this else even sensible skepticism can be mulched back into the futurological framework to fertilize the next crop of weeds.
Not sure if Moore's Law is still chugging along anymore. Compare a top end Desktop PC from two years ago with what you would get for the same money today.
And Kurzweil's even more hyperbolic Law of Accelerating Returns is completely bagged for disposal -- tho' sensible people found it a laugh out loud shell game from the start. Watching tee vee shows from a decade ago or more one finds the computational/multimedia furniture of their world completely familiar. Going back two decades that isn't quite so true. But clearly the accelerating acceleration of acceleration promoted and marketed in the belly of the irrationally exuberant beast is exposed in retrospect as arrant nonsense. Dot eyed futurological True Believers still foam at the mouth about racing upward ever quicker into Robo-Heaven, but everybody else sees we're stuck in the deep muddy.
> . . .speech recognition has been a stagnant field for a decade now. , .
You mean, no computer has yet unravelled the mystery of
the Pompatus of Love?
> Of course, Kurzweil should know better. . .
The Singularity bandwagon has swept up other folks you might
think should know better. Douglas Hofstadter has flirted with them
(and even hosted seminars for them), but has also kept a sensible
distance.
But this guy has an impressive resume and yet spends time
serving as an "advisor" to SIAI. (OTOH, his pal Stephen Wolfram
also has a bit of a reputation as being a. . .
well, you know -- a slightly flawed piece of earthenware,
also somewhat too wide at the top.):
http://en.wikipedia.org/wiki/Steve_Omohundro
(Cute, too:
http://steveomohundro.com/
http://www.flickr.com/photos/29246236@N02/2748347698/ )
So true! And, also, too: poop!
Hasn't SS run exposes of cryonics and drextechians and other Robot Cult flavored futurological flim flammery in the past?
> . . .he said that cryonics was not possible because of freeze fracturing. . .
Now here's a kind of "cryonics" I can believe in, that **depends**
on freeze fracturing:
http://www.timesonline.co.uk/tol/news/uk/scotland/article3780207.ece
http://www.youtube.com/watch?v=lIwEf8J2b34
http://www.youtube.com/watch?v=4HD5Gt80H6s
Don't know how soon it's likely to become available in the
U.S., though.
> . . .speech recognition has been a stagnant field for a decade now. . .
>
> . . .No computer can explain what it means to be revved up like a douche, either. . .
Apropos of which, a friend of mine commented:
"When context is limited, even humans have trouble. A good example,
extensively documented on the Net, is misheard song lyrics,
collectively called mondegreens, in honor of the pupil who transcribed
They have killed the Earl of Moray and laid him on the green
as
They have killed the Earl of Moray and Lady Mondegreen.
No machine would make that mistake; but our ability to make such a
mistake seems likely to be related to our ability to be right more
often than machines. The hearer knew nothing of the story except that
it had been made into a ballad; it is a fair strategy in such a
situation to thicken the plot by inventing a lady."
Post a Comment