Off by about 200 Years

[oneirophrenia]

Google fellas say it will take another 300 years to index all the world's information

And I quote:

"We did a math exercise and the answer was 300 years," CNet quotes
Schmidt as saying. "The answer is it's going to be a very long time."

He estimated that out of the five million terabytes of information
(no, we don't know how he counted it all up, either) in the world, a
mere 170 terabytes have been indexed so far.

What kind of bonehead "math exercise" was this, you dingdong? I rarely criticize Google, 'cause them boys is smart and stuff, but this is just ridiculous. "All the information in the human world" will NEVER be indexed. Why? Because we keep making more every nanosecond.

The technology for indexing information is growing exponentially. Which means if we made no more new info, we'd be able to index it in much less than 300 years just by dint of ever-growing bandwidth, processor speeds, etc. But we're also using that exponentially-growing tech bed to generate more information, as well. Exponentially. It's nothing but exponents all the way down, people....

Google's Eric Schmidt needs to read Kurzweil's latest book. Hell, I might just buy him a copy and send it to him....

Anyway, I wonder what the Googleplex will think of this when it becomes spontaneously self-aware sometime in the next twentyfive years.

Comments

[fkmedocrity.livejournal.com]

Wow. That was a monumentally stupid thing for him to say! Why is it that some phenomenally smart people can come to the most illogical conclusions?

[rapier1.livejournal.com]

Spontaneous self-awareness is unlikley. In fact, its probably nearly impossible without necessary external pressures demanding it on a regenerative fault-prone system. It may be possible to create a self-aware system - or at least one that can pass a turing test - but the chance of it happening without anyone trying to make it happen is so close to zero as to be zero.

Likewise, Kurzweil does a lot of handwaving. He makes the fundamental mistake of reifying Moore's Law (which is hardly a law as much as an ipso facto observation) and extrapolating from there. Its like that old saw where you take a single bacteria and let it divide for a few days and you end up with so many bacteria that it weighs more than the planet. Sure it works mathematically but inescapable laws of physics and biology keep getting in the way. It important to keep in mind that Moore wasn't stating a technical maxim but an *economic* one and its the economics will be the breaks (as they have been in the past). Eventually its very likely that diminishing returns will simply make it unprofitable to put more tranistors on a chips or develop more compute power.

Even Moore said that no law of exonential growth can last forever - forever can be postponed of course - but not forever.

[oneirophrenia.livejournal.com]

Yeah...just wait until some kid--or Google sysop--releases a genetic-algorithm-powered worm into the Google databases in 2020. :)

Kurzweil's so-called "Law of Accelerating Returns" is, also, primarily powered via economic development factors. It predicts a certain accelerating development curve but doesn't, in itself, indicate the particular *causes* for said development curve--though this is something Kurzweil himself addresses in his latest book and in a number of articles. It's all economic. More bang for your buck. Customers demand better computers for their money, and this economic imperative ultimately drives processor manufacturers into their development cycles--for *personal computers.* What Kurzweil doesn't really address is the fact that the US government's own security agendas enforce continual development of better computing technology as well...just not for common public use (at least initially). Who do you think is bankrolling most of the quantum-encryption development currently (aside from universities operating under private industry development grants)?

Eh. Conventional two-dimensional, silicon-based transistor platforms are rapidly reaching the physical limit of miniaturization. There's already plenty of research on going beyond that....Three-dimensional processor arrays. Photoelectronic circuits. Quantum computing. There're already small models of all of these functioning at various research facilities around the world, and right now they're in their initial development stages--certainly not ready for "public consumption" yet. But that really doesn't mean anything in the long-run: the second the technologies become even *vaguely* cost-effective (I don't give it any more than ten years) to mass-produce every single microchip manufacturer on the planet will be scrambling to build fiberoptic-based circuit production lines and the world's best 10gHz Pentium 10's will be relegated to Ray's attic.