Wednesday the Twenty-Third of December, Two Thousand and Nine
"...and it's hotter than reality by far."
So, Avatar. Jesus, Avatar. Part of it was the 3D, but let me tell you: I do not remember the last time I was so deeply immersed in a film. I was shaking for a good hour afterwards. The real world didn't look real enough after that movie. It was only the second CGI-heavy movie I have seen where the effects were truly unobtrusive—and the first came out six months ago. The story itself was clichéd and laden with unfortunate implications, as a thousand bloggers have already pointed out. You know what other story was clichéd and laden with unfortunate implications? Yeah, I went there: I think Avatar could be the 21st century's Star Wars.
And the thing is, I'm not really making a value judgement here. I'm honestly not sure how good a film it was. (Not sure how good Star Wars was either, for that matter.) But it was an immensely powerful film, more viscerally and immediately powerful than any other piece of fiction in any other medium I can think of off the top of my head. And I think it will be recognized in hindsight as a very, very important film.
In a couple months, when everyone who wants to see it has seen it (and when I have seen it again), I'd be interested in discussing it in more depth on the blog, including the obvious Avatar vs District 9 debate. Before then, if you're curious, hit me up on IM or something.
Monday the Twenty-First of December, Two Thousand and Nine
To be fair, plenty of humans don't either.
Speaking of which: Wow. Google Voice does not understand my father. "It's, Ron Shaw, I wasn't sex not drive, Florida, and I'm hoping to get back with the space."
Sunday the Twentieth of December, Two Thousand and Nine
I can't believe I wasted half a day of my rapid Wilkinson on this.
So I've just read The Age of Spiritual Machines, by Ray Kurzweil, and wow. It's always so weird watching someone very, very intelligent being completely wrong about everything. He could have gotten away with a lot more wrongness if he hadn't tried justifying his wrongness with more wrongness. The whole beginning of the book is all about how evolution is an "intelligent process" leading inexorably to a higher state of order and therefore to a species capable of creating technology. This complete misunderstanding of evolution doesn't really have anything to do with his point, but it certainly makes it harder to read. I had to keep throwing the book aside and yelling at it. Look, Ray, evolution does not have goals, much less intelligence. Humans are not the pinnacle of evolution, though horseshoe crabs might be. Technology use is a major evolutionary boon to a species (though consider all the drawbacks that go with it—long childhoods, large heads, higher rates of death in childbirth, etc., all of which must come into play before technology can exist) but then so is the ability to breathe fire, and that doesn't seem to be an inevitable result of evolution.
Anyway, I don't want to quibble about how wrong he is about his tangential arguments or we'll be here all day. The main thrust of his argument is that computers are growing exponentially more powerful (a trend which he claims predates and will outlast Moore's law) and that it's only a matter of time before computers are more intelligent than humans. But that's founded on the idea that human brains are just sufficiently advanced computers, which is, quite simply, balls. The world's fastest computer today peaks at around 2 × 1015FLOPS (floating-point operations per second), and distributed computers are up to 8 × 1015—my conscious brain can manage about 1, tops, and that's only due to extensive use of memory caching. (When you ask me to add 7 + 6, my pattern recognition software recognizes that it's equivalent to 7 + 5 + 1, so my brain pulls the cached equations "7 + 5 = 12" and "12 + 1 = 13" from memory to solve the problem. It's actually more efficient than the method a computer would use, which only underlines how incredibly slow my brain is.) FLOPS, in short, are a completely meaningless measure of human brain power. We don't actually have a good measurement of brain power, besides the horrifically flawed but semi-useful concept of IQ, and in fact we haven't even got a good definitition of human intelligence. But it is obvious to me that the human brain is not like a computer, and I'm not even sure (and people smarter than me continue to debate) whether the brain is even a Turing machine—that is, whether the very notion of "computation" is even relevant to the brain. And the thing is, Kurzweil knows all this. He even addresses some of this. He just chooses to ignore it when it contradicts his theories.
There are a whole bunch of logical fallacies that Kurzweil uses that I don't even have names for. Like this one: "Technology α requires technologies β, γ and δ. If we have already discovered β and γ, then α is inevitable!" In other words: "Look, we've got plastic and blinky lights, right, so now all we need are a couple Heisenberg compensators and we'll have a transporter in every home!" Another fun example is when he puts forward the following (paraphrased): "Imagine a computer prints the phrase 'I'm bored; play with me' on the screen. Is that evidence of intelligence? What if instead of printing it to the screen, it says it aloud, using a synthesized voice? Is that intelligence? Taking it one step further, what if the computer was never programmed to say that phrase, but came up with it on its own? Is that intelligence?" My rephrasing of that would be: "Imagine a horse, capable of running about 25 miles per hour. Is that fast? Imagine now it was a thoroughbred horse, and could run at 30 miles per hour. Is that fast? Okay, let's take it one small step further: imagine that the horse was actually the starship Enterprise, capable of going warp 9.6..." (Well, I had fun with the first Star Trek metaphor, so I figured I'd run with it.)
I don't know, maybe eventually artificial neural networks will become more promising as computer-based models of the human brain. But that research isn't directly based on Moore's law (though it does depend on it in part—neural networks, in their current state, consume IMMENSE amounts of computer resources) and I see no evidence that it's accelerating exponentially. But that's the only computer technology currently on the table that I can see coming anywhere close to the human brain... eventually. And there are still a lot of unknowns there.
Anyway, the book is fundamentally a list of predictions, so let's look at some. First, let me just go over Kurzweil's earlier predictions for 1999—the year this book was written—which he takes to be almost 100% true. He uses the claim that he was right about all of his earlier predictions to justify his predictions for the 21st century, so it's worth examining a few of them:
"A computer will defeat the human chess champion around 1998." This is one of the better predictions he made, and unlike most of the others, it was based on actual data—he noted that computers were advancing up the chess rankings faster than humans, and plotted the convergence.
The information economy will lead to "sustained economic growth and prosperity". Heh. Remember 1999, when people actually thought that?
"A worldwide information network... will emerge." This would have been a fantastic prediction a few decades earlier, but in 1988, when the Internet was already widespread and the Web was only two years in the future, it's not so impressive.
"The three technologies for a translating telephone... will each exist in sufficient quantity for a first generation system by the late 1990s." Hilarious. Don't you remember back in 1999 how great speaker-independent speech recognition and machine translation were? The scary thing is he actually thinks this came true.
He talks a lot about how awesome speech recognition software (which he sells, incidentally) and machine translation are, so just for fun, let's demonstrate that technology in 2009. I'll take one of my transcribed Google Voice voicemails and run it through Google Translate. The original message was, by my transliteration:
Hey, it's me, just calling to say hello. We're at San Francisco and getting ready to board in ten to fifteen minutes, heading home. Don't need to call back, obviously; my phone'll be off. Um, hope you had a good weekend, and hope everything's going well, and I'll catch up with you soon. Bye bye. I took the Google Voice transcription of that, ran it through Google Translate into Spanish, then translated it back to English myself (writing out any untranslated words as they would be pronounced by a Spanish speech synthesizer):
Ey, it's me, simply calling to greet. In the case in San Francisco and it is preparing to support and ten fifteen minutes to house. I got a return call. Obviously, my phone. I'm going. We hope that you had a rapid Wilkinson. Opay all goes well and I will become to the day with you all soon bee-ay bee-ay. I don't know if this link will last, but at the moment you can hear it spoken by a speech synthesizer here (or here, if you want to hear the Spanish). Gosh, I wonder why translating telephones never caught on...
Right, so if that's 1999, how'd he do in 2009?
"Supercomputers match at least the hardware capacity of the human brain—20 million billion calculations per second... A $1000 personal computer can perform about a trillion calculations per second." Supercomputers are an order of magnitude behind that prediction, and his repeated claim that the human brain has a capacity of 20 petaflops, as I said before, completely absurd. Teraflop-plus personal computers are out there, but that's just Moore's law at work, not something new. Anyway, over the course of the last decade, we've largely stopped noticing the increase in processing power. Software expands to take use of the resources available—when Office 2007 came out, for example, it took longer to load on a contemporary computer than Office 2003 did in 2003. Software (which does not improve exponentially, by the way) often doesn't use the hardware very efficiently—this is particularly true on multi-core computers. Also, we're moving more and more towards smaller but less powerful computers—netbooks and smartphones use hugely underpowered processors compared to desktops. So, while true, it's not very useful.
"Personal computers with high-resolution visual displays come in a range of sizes, from those small enough to be embedded in clothing and jewelry up to the size of a thin book". How exactly do you put a high-resolution visual display in a piece of jewelry, anyway? I mean, make the resolution as high as you want, you're not going to browse the web on a half-inch screen. Anyway, this seems to imply he believed that desktops would be obsolete by 2009, which is a bit silly in retrospect. Yes, we have computers the size of a thin book and smaller, but we don't use those as our primary computers. Big screens and big keyboards are just too damn important.
"Cables are disappearing. Communication between components uses short-distance wireless technology. High-speed wireless communication provides access to the Web." This incredibly thick Cables To Go catalog begs to differ. (I got on their mailing list somehow.) Bluetooth and other wireless technologies are grand, but nine times out of ten it's cheaper and easier to run a few feet of cable. And for high-quality video and audio, cables are a must. As far as wireless Internet goes, the cellular phone networks are now the only real national option, and they're still much, much slower than wired connections. Most people connect either by wired or short-range wireless (i.e. WiFi) connection. Maybe in ten more years.
"The majority of text is created using continuous speech recognition... but keyboards are still used. CSR is very accurate, far more so than the human transcriptionists who were used up until a few years ago." Oh, snap! Mirabai, you just got CALLED OUT by Ray Kurzweil! Seriously, speech recognition software is miserable compared to human speech recognition. Humans can (unconsciously!) select a single speaker at a crowded party and understand them clearly, even if it's a complete stranger talking about something the listener doesn't know much about. Speech recognition software may have a 95% accuracy rate under ideal conditions, but that's like me saying I can beat an Olympic sprinter as long as they're running backwards up a hill through sand. Take them out of the ideal conditions, and they get destroyed. Hell, I can't even ask my car to turn the radio off if the heater's turned up too high, because the fan noise confuses it.
"Most routine business transactions... take place between a human and a virtual personality." Thank God, no.
You know what? There are a lot more predictions. They're pretty much all wrong. Many of the wrong guesses are based on ideas that seemed promising in 1999 and then turned out not to be. Others are based on the theory that technology alone is required to make an idea happen, leaving aside the questions of practicality, cost, and above all consumer desire—videophones and automated highway systems being classic examples (both of which have sunk more futurists than just Kurzweil). Predicting the future is hard! Most people who try are wrong! It's usually easier to see problems coming than guess what the solutions will be. As an occasional writer of science fiction, I am absolutely guilty of making very wrong guesses about how the future will unfold—that comes with the territory. It's not Kurzweil's incorrect predictions that bother me, it's his incorrect assumptions. His whole theory is based on a foundation of wishful thinking and bollocks. And he should know better! Grrr!
I'm definitely not a synaesthete—and I'd give both legs to be one—but I have noticed that sometimes, when I'm reading something, the writing style provokes an almost sensory reaction. It usually happens when I'm reading something that, in some intangible way or another, reminds me of something I've written before. It's like it brings up all the emotions and memories I have tangled up in my own writing and the life experience that fuels the writing, all at once, too quickly to disentangle the threads.
Back in the French Revolution, there was one Occitan revolutionary who was known by the nom de guerre "Toulouse", after his home town. He rose to prominence after his stirring "Nothing Left" speech, in which he decried the way the aristocracy had taken everything from the French people, leaving them with nothing. As the Revolution turned into the Reign of Terror, he sided with the Enragés and carved out a niche for himself in rural France, well-insulated from Robespierre and the other Jacobins in Paris, where he and his followers formed a commune built on the most radical ideals of the Revolution. However, as time went on, Toulouse, like so many of his contemporaries, became a petty dictator. After his second-in-command questioned the direction that the commune was going, Toulouse had him executed.
On hearing about this, one of Toulouse's followers turned to a friend and said, "But, citoyen, how does executing his lieutenant help the cause of freedom for all France?"
His friend scoffed and replied, "Pah! Freedom's just another word for 'Nothing Left' Toulouse!"
