Archive: March 2008

I’ve read a ton about the debate over torture, but in a piece in today’s New York Times Week In Review, Scott Shane makes a new and excellent point: The government has spent virtually nothing studying the sciences of influence and persuasion and how they apply to interrogation.

Shane points out that well-known fact that the Army Field Manual explicitly advises against torture; it offers instead a set of observations about human motivation that an interrogator can exploit. (“People tend to want to talk when they are under stress and respond to kindness and understanding during trying circumstances,” for example.) This is the sort of behavioral psychology we’ve all learned at the foot of prime-time police procedurals. But, as Shane points out, an understanding of this stuff isn’t reflected at the highest levels of government, which is either a symptom — or a cause — of the bigger problem, which is that the feds don’t avail themselves of the truckloads of research that’s been done in recent decades, partly by corporations eager to get people to buy, y’know, $300 prestressed jeans.

As Shane writes …

… the manual’s inherited wisdom has not been updated to reflect decades of corporate analysis of how to influence consumers. Behavioral economists have dissected decisionmaking, and academic psychologists have studied political persuasion, but their lessons have not informed the interrogator’s art either. Nor has there been a systematic effort to analyze the successes and blunders of the interrogations carried out since the attacks of 2001.

Steven M. Kleinman, a colonel in the Air Force reserve and a veteran interrogator in Iraq and elsewhere, says the government spends billions on spy satellites but almost nothing on studying interrogation. This is true, he said, despite a broad consensus that interrogation might be the best source of information on an elusive, low-tech, stateless foe like Al Qaeda.

“We need to bring scientific standards for interrogation to the same level of sophistication that we bring to satellite imagery and intercepting communications,” said Mr. Kleinman, who has studied the American interrogation programs used for high-level German and Japanese prisoners during World War II, which he judges superior to those developed since 2001.

Kleinman suggests “a new intelligence agency or subagency devoted solely to interrogation — sponsoring research, conducting training and building a team of sophisticated interrogators with linguistic and psychological skills.” That sounds like a great idea. When I wrote a story a year ago for the New York Times Magazine about the intelligence agencies — and the difficulty they’re having getting their superlegacied and legally “airgapped” databases to talk to one another — virtually everyone I talked to agreed that high-tech spying was great, but secondary to good old-fashioned cultivation of sources. If you want to understand the makeup of terrorist threats, you need to just, well, talk to a lot of people, which requires intelligence experts fluent in the languages and psychology — not the black arts of torture.

I admit, I’m a sucker for brain-scanning experiments. But this one is particularly intriguiging: A group of scientists scanned several jazz keyboardists while they improvised solos. The finding? The parts of the musicians’ brains that monitor their performance shut down, while the sections that organize “self-initiated thoughts and behaviors” were highly activated.

Soloing is, of course, one of the more spontaneous and creative moments in music. You have to follow the basic structure of the song while, on the fly, generating a new melody that picks up on — and plays off of — the individual performances of the other instruments. But the idea that improvising requires you to sort of stop scrutinizing yourself is incredibly interesting.

As a piece in Scientific Blogging notes …

… the researchers found that much of the change between improvisation and memorization occurred in the prefrontal cortex, the region of the frontal lobe of the brain that helps us think and problem-solve and that provides a sense of self. Interestingly, the large portion responsible for monitoring one’s performance (dorsolateral prefrontal cortex) shuts down completely during improvisation, while the much smaller, centrally located region at the foremost part of the brain (medial prefrontal cortex) increases in activity. The medial prefrontal cortex is involved in self-initiated thoughts and behaviors, and is very active when a person describes an event that has happened to him or makes up a story. The researchers explain that, just as over-thinking a jump shot can cause a basketball player to fall out of the zone and perform poorly, the suppression of inhibitory, self-monitoring brain mechanisms helps to promote the free flow of novel ideas and impulses. While this brain pattern is unusual, it resembles the pattern seen in people when they are dreaming.

That latter point is particularly lovely. I have to say, this corresponds perfectly with my own experience of improvising. I’ve played guitar, harmonica, and a bunch of other stringed instruments — banjo, etc. — for 20 years, and I often find that improvising for a half-hour or so at the end of the day is the single best way to clear my brain. I’ve always thought that this was because I work in words, and by the end of a long work day I crave doing something that’s completely nonverbal; and instrumental noodling perfectly fits that bill.

But now I’m wondering — maybe the deeper reason I enjoy it so much is that improvising shuts down my brain’s near-constant self-surveillance. My job, like many white-collar jobs, involves a lot of socializing (over the phone, anyway, in interviews), and enforced workplace socializing requires constant self-awareness, self-scrutiny and inhibition. This is pretty exhausting to maintain all day long. So maybe what I like about improvising on an instrument is that it frees me having to pay attention to myself.

I’d imagine Daniel Levitin would have some smart things to say about this.

(Thanks to Music Thing for this one!)

My comments are down again. I need help!

Apparently the spambots are slamming my comment script so rapidly that Pair, my hosting service, keeps on disabling the script. That’s why comments aren’t working right now.

Anyone have any advice on dealing with this? I currently use Autoban (version 1.2.3), which is supposed to throttle any spambots that try to post zillions of time in a row. I’ve been using it for about a year now, with great success.

But something’s changed in the spambots’ behavior, I guess. Anyone with smart advice, feel free to email me — I’m all ears!

So, you’re living in Florida, and you just suffered through a massive blackout. Want advance warning of the next one? Then go to the website of the Pacific Northwest National Laboratory, and download the “Grid Monitor” — a screensaver that shows you the stability of the power grid, in real time, via a series of totally gnarly graphics, such as the Oscillatory Mode Graph above.

You can actually watch the grid begin to buckle and collapse when a blackout is approaching. Since grid collapses occur randomly and very infrequently, you’d have to be staring at your screensaver 24 hours a day, but hey: Maybe you’ll get lucky! On the other hand, you can also set the screensaver to give off a warning sound when the power in the grid fluctuates too wildly — an impending sign of a blackout.

Pacific Northwest also created this nice PDF pamphlet that explains how the Grid Monitor works. It contains this rather metaphorically lovely passage:

In reality, the AC electric power signal is the sum of innumerable sub-signals. The 60 Hz AC signal is actually a complex accumulation of many elements such as random noise, mechanical vibratory dynamics of generators producing the power, damping effects, and even self-induced oscillatory dynamics of the transmission grid. It acts like a tremendous bed of interconnected springs and weights.

This is fascinating to watch: A team of Japanese researchers have created “shockwave traffic jams” that replicate the dynamics of real-world highways.

For 15 years, researchers have known that traffic jams can emerge out of the blue. All it takes is for one driver to momentarily slow down, at which point the person behind him hits the brakes, forcing the person behind him to hit the brakes even harder, and so on, and so on. One teensy butterfly flaps its wings, and pretty soon the whole damn interstate’s a mess. If you’re in a helicopter, you can watch the “shockwave” of slowed-down cars propagate backwards through traffic like a wave through water. Physicists have long produced eerily accurate computer models that replicate this phenomenon precisely. But because it’s pretty hard to commandeer an entire highway for the purposes of research, no one has ever replicated the phenomenon in a real-world experiment.

Until now! The Japanese team got a cluster of vehicles to drive in a circle. As the New Scientist reports, here’s what happened:

They asked drivers to cruise steadily at 30 kilometres per hour, and at first the traffic moved freely. But small fluctuations soon appeared in distances between cars, breaking down the free flow, until finally a cluster of several vehicles was forced to stop completely for a moment.

That cluster spread backwards through the traffic like a shockwave. Every time a vehicle at the front of the cluster was able to escape at up to 40 km/h, another vehicle joined the back of the jam.

The shockwave jam travelled backwards through the ring of vehicles at roughly 20 km/h, which is the same as the speed of the shockwave jams observed on roads in real life, says lead researcher Yuki Sugiyama, a physicist in the department of complex systems at Nagoya University.

“Although the emerging jam in our experiment is small, its behaviour is not different from large ones on highways,” he told New Scientist.

Check out the video of the experiment. Towards the end, the shockwave becomes deliciously mobile — you can really see it moving backwards.

This also puts me in mind of William Beatty, the electrical engineer who — while stuck in traffic in 1998 — figured out a way to hack traffic jams and erase them. Basically, when he was stuck in a jam, he’d slow down until he had a really large amount of space between him and the car in front of him. Then he moved forward in at very slow, uniform speed, so that he no longer stopped and started. Sure enough, the wave stopped at him: Everyone behind him began driving at a uniform 35 mph. “By driving at the average speed of the traffic around me, my car had been ‘eating’ the traffic waves,” he wrote. The only problem, of course, is that he himself was stuck traveling at the average speed of the wave in front of him, which — at 35 mph — is pretty pokey.

(Thanks to Greg Sewell for this one!)

Can you fix things that break in your household? Probably not. Our schools systematically stream “smart” people away from working with their hands, and I think that’s a huge problem for the US, on pretty much every level — commercially, globally, intellectually and spiritually, really.

My latest column in Wired magazine is on the stands now, and it tackles this precise problem. There’s a copy of it on the Wired web site, and one archived below — but of course you should also immediately drop whatever you’re doing and buy a physical copy of Wired, then fill out the subcription card too, heh.

By the way, if you like this column then you’ll love Matthew Crawford’s essay “Shop Class as Soulcraft”, which appeared in The Atlantic Monthly in 2006. I discovered it while doing my research, and wound up interviewing Crawford for my piece. He’s working on an entire book about the demise of America’s prowess with tools, and judging by how superb his essay was, his book will rock with hurricane force, I suspect.

How DIYers Just Might Revive American Innovation

by Clive Thompson

What a mess. I’m sitting on the floor of my apartment, surrounded by electronic parts, a cigar box, a soldering gun, and stray bits of wire. I’m trying to build my own steampunk-style clock — hacking a couple of volt meter dials to display hours and minutes. It’ll look awesome when it’s done.

If it ever gets done — I keep botching the soldering. A well-soldered joint is supposed to look like a small, shiny volcano. My attempts look like mashed insects, and they crack when I try to assemble the device.

Why am I so inept? I used to do projects like this all the time when I was a kid. But in high school, I was carefully diverted from shop class when the administration decided I was college-bound. I stopped working with my hands and have barely touched a tool since.

» MORE...

I’m coming to this one late, but apparently some Italian scientists have spent $6 million euros building a robot that makes coffee.

You laugh. (Well, I did, anyway.) But part of the challenge of producing robots that help people, of course, is mastering some of the unexpectedly complex motions of human limbs — and making coffee is precisely this sort of unexpectedly complex task, so it’s actually kind of a cool thing to try and achieve. As the scientists note in this designboom story:

“The problem of using two hands together, the way humans do when the pick up a heavy plant pot, is a particularly sticky one. At present robots can use a single arm with reasonable accuracy and flexibility. But until now they have fallen short of the technological complexity and artificial intelligence needed for a two-handed approach.

“We want to develop a system of two-handed manipulation, equipped with sensors that make
the robot conscious of its surroundings and the people in its working space’, Siciliano said.

Truthfully, having watched a video of the robot in action, ay yi yi would I not want that thing slinging volcanic McDonald’s-hot coffee anywhere near me. The robot’s motions are quite elegant, in their own way, but the training is still pretty spastic.

Sometimes I wonder whether robotics money ought to be spent less on making robots that replicate human activity, and more on robots that enhance human activity — i.e. that do things of which we’re completely not capable. Like blasting holes in walls, or picking up that car that’s illegally parked in front of your house and crumpling it into a ball. On that note, I was particularly charmed to read yesterday on Boing Boing about the guy who created a remote-controlled vigilante robot to chase drug dealers away from his neighborhood by spraying them with water.

(Thanks to Yishay Mor for this one!)

Last week, Wired News published my latest video-game column — and this one’s about the peculiar relationships we strike up with AI characters inside games.

It’s online free at the Wired site, and a copy is archived below!

Going Gunning With My Imaginary Friends
by Clive Thompson

Can a machine think?

That’s the question that mathematician Alan Turing posed in 1950, when he posited his famous Turing Test. He argued that artificial intelligence could be thought of as intelligent if it passes a social test — if it can fool a human into believing it’s real.

Alas, critics agree that no machine has passed the Turing Test. We’re never fooled by chatbots for very long, as the annual Loebner Prize contest proves. The thing is, we humans are awfully good at decoding social cues and detecting humanness; we can instantly tell when a preprogrammed “conversation tree” is repeating itself. That’s why many philosophers say machines will never pass the Turing Test.

» MORE...