Tech Talk iconTech Talk

A Canticle For DARPATech

I’ll never eat Pentagon m&ms again. A DARPA spokesperson has confirmed that there will be no more DARPATech, the Defense Advanced Research Projects Agency’s bi-annual (but occasionally annual, and at other times occurring only every three years) conference, at which the latest and greatest “mad science” technologies go prime time. A quick eulogy is in order.

DARPATech 2007 featured a particular bumper crop: Robot arms, unmanned autonomous robot Humvees, all-seeing blimps, an autonomous insect-like robot called "Little Dog," a robo-beast of burden called Big Dog, and a deputy director’s bikini-clad demonstration of a core-temperature regulating glove. There was even a bracing dose of reality.

Yes, DARPATech was probably a PR boondoggle meant to remind news outfits that the Defense Department isn’t just about killing people. But is that so wrong? Of all the Defense agencies, DARPA is probably the best-run. DARPA program managers have four-year contracts, and they never get the chance to become career bureaucrats. After their term is up, they are told to skedaddle no matter the status of their project. The agency is low on bureaucracy and high on ideas. And the ideas are life-changing.

However, the Obama administration is likely avoiding highly visible celebrations of war. That might be an unfair description of DARPATech, but how else would you characterize 3,000 defense contractors hanging out at a convention so elaborate and shiny that it makes a trip across the street to Disneyland (literally) seem boring?

Most likely, the biggest reason is money. When Danger Room blogged the 2007 convention, reporter Sharon Weinberger observed that the best kept secret at DARPAtech was "how much it costs."

According to Yudhijit Bhattacharjee at ScienceInsider, the FY2011 budget for DARPA is $2.9 billion. Though the agency lost $100 million from 2010, they shifted $200 million to basic research (bringing that amount to $2 billion). To get to that number, DARPA said that it had to chop some "low priority weapons development programs." Also: shiny conventions across the street from Disneyland.

The DARPA spokesperson told me that the agency has been pursuing “different arrangements.” In January, for example, they hosted the DARPA Industry Summit in Washington, DC “to discuss key globalization issues,” and he says that DARPA expects to hold similar meetings in the future.

Full disclosure: I am an unabashed DARPA fangirl. For me, this is very sad news indeed.

(Blue) Brain and Beauty

A couple weeks ago, I was invited to see the new Rolex Learning Center at the Ecole Polytechnique Fédérale de Lausanne.  Architecture critics are fawning over the $65 million building.  A writer in the Guardian yesterday compared it to "some filmic version of the afterlife."  But to me what's even more remarkable is what's going on nearby at the school:   they're building a brain.

The Ecole Polytechnique Federale de Lausanne is a grim campus of sparsely-windowed gray buildings thirty minutes north of Geneva, Switzerland.  When I was last there a couple years ago, I met Marcus Bartschi, a wry, 43-year-old computer engineer, who sat at a desk in a dim office with a rusty saw hanging inexplicably on a bare concrete wall.  He had a scruffy black beard and dark circles under his eyes, brought on by his two-week-old son.  “My boy has only physical needs now,” Bartschi told me, resignedly, “there’s not much for me to do.”  So he tended to his other baby:  a four ton supercomputer named Blue Gene.  Developed by IBM, Blue Gene is one of the world’s most powerful high-performance machines, and it was being used here for a suitably super, some say impossible, mission:   to simulate a brain.   Bartschi had the vulnerably human task of keeping it alive.  “Hellooooo,” he cooed, as the laptop monitoring Blue Gene fires up.  

The project, dubbed Blue Brain, is the dream of Dr. Henry Markram, the neuroscientist in charge of the EPFL’s Brain and Mind Institute.  Several years ago, Markram approached IBM with an ambitious plan:  to accelerate the normal path of research by building a three-dimensional model of a mammalian brain, as he said, “in silico.”   Skeptics such as MIT’s Marvin Minsky are already questioning what they call won’t rule it out completely.  “We do not yet know if consciousness will emerge in these artificial brains,” he has said, “and we will consider the ethics of this if this happens.”  

Bartschi, a diehard Hitchhiker’s Guide fan, was more skeptical about his Blue Gene baby, “this is no HAL,’ he said  Blue Gene, however, was robust (with a processing speed of 22.8 tera-flops per second, the eight fastest supercomputer in the world) and affordable ($2 million per rack) enough to give Markram his crack.  The goal was to first simulate a single neocortical column, then take another ten or so to replicate an entire brain.  While Markram’s team tends to the data-crunching software, Bartschi had the less glamorous job of caring for Blue Brain’s pillar:  Blue Gene.  “It’s a lot of pressure,” he said, with a sigh, “I want it to work 100% of the time.” 

After the fanfare of the announcement, Bartschi oversaw the installation of the four refrigerator-size racks, which resembled the black monoliths from 2001:  A Space Odyssey.   After two weeks of assemblage, he triumphantly booted up Blue Gene – only to see it repeatedly, and mysteriously, shut down.  For two weeks, Bartschi searched for an explanation, only to discover that a solid steel beam underneath the raised floor was obstructing 50% of the airflow required to keep it cool.  It took another two weeks for Bartschi’s team of six to schlep Blue Gene down the hall.  The next month, a crucial fan broke, and there was no replacement on site.  The part was ordered from Rochester, New York, but, due to a mix-up in shipping, was sent low priority – holding up Blue Gene for more than a week.  Despite such tantrums, Bartschi had developed a soft spot for the machine.  “The whole setup is complex," he said,  “In this sense, it is a living thing.”

Enhanced Imagination Drives Brain-Computer Interface

It's been clear since brain computer interfaces were developed, that customizing these devices would require learning both on the part of the machine and the human. New research in the Proceedings of the the Academy of the Sciences gives evidence that humans quickly adapt to BCIs.

A team of neurologists and computer scientists at the University of Washington recruited epilepsy patients awaiting surgery and recorded their brain activity with electrocorticography (electrodes attached to the surface of the brain) before and after they manipulated a simple BCI. You can find the full article here, to the right of the press release.

First of all, here's what they did. They recorded during three circumstances: when patients imagined moving their hand, when they actually moved it, and when they moved a computer cursor by manipulating a BCI. The activity during the imagined task mapped roughly onto the recordings from the actual movement, but were less powerful. When the patients hooked up to the BCI, the pattern was again similar, but the signal much stronger than both the other recordings.

The press release pitched this as evidence that BCIs are a "workout" for the brain. I don't completely buy this. The brain isn't a muscle and more activity doesn't necessarily mean it's operating at a higher level. What it does indicate (to me), and what I find far more interesting, is that people can quickly change their brain activity to accommodate BCIs. It also shows how important visual feedback is to people who are manipulating these devices. Experiments like this seem like a good way to maximize the level of feedback a user is getting and to test out different ways of delivering it.

It's also substantial proof that the brain activity produced when we imagine a movement or task can effectively drive BCIs. Every group that's developing BCIs right now is doing it slightly differently. So far, there is no clear consensus on which brain signals should be used.

This is the first paper I've seen that focused fully on what brain activity looks like when it's manipulating a BCI. The output of the setup was a cursor moving on a screen. The experiment is a good indication that BCIs have become well enough understood that we can use them in experiments as tools to once again study the brain itself.

That being said, there are also some really interesting things to be learned from this article about the brain in general and the difference between imagining and actuating movement. Here are a couple points that may get you to read it and some questions you can respond to if you do.

1. During both tasks, high frequency signals increase while low frequency signals decrease. Does this mean that part of attending to a task is muting some of the competing activity?

2. Of these two, it is the signal that decreases which map similarly in both imagery and movement. Does this mean you could further localize an area that controls movement imagery in the high frequency signals?

AT&T Wireless Adds Windows Phone 7 Series

Microsoft is a distant third when it comes to mobile phone operating systems, but the company has a legendary marketing ability to hang tough and emerge on top. So the long-awaited announcement of Windows Mobile 7 (or “Windows Phone 7 Series,” as Microsoft now styles it) was picked over by the trade press like a leftover turkey carcass on the Saturday after Thanksgiving.

It didn’t take long for the vultures to notice that Microsoft named AT&T its “premier partner” in the United States (Deutsche Telekom AG, Orange, SFR, Sprint, Telecom Italia, Telefónica, Telstra, T-Mobile USA, Verizon Wireless and Vodafone are others around the world)—nor for them to start circling the two partners with questions.

As PC World's Tony Bradley pointed out, no one really knows what a premier partnership consists of. Does it matter? Let the griping begin. AT&T already can’t handle the volume of data that iPhones are generating, and beginning in April, iPad data will only swamp it further. And now AT&T is going to take on a new data-heavy collection of smartphones from Microsoft?

Speaking as a a founding iPhone user from its initial release in June 2007, I can say that each and every complaint about AT&T is borne out by my own experience. Neither the quality of the phone nor the data connection seems related to the number of bars, and calls are dropped repeatedly and at random—it’s rare for me to complete any but the shortest conversations without redialing. And that's as true of my shiny 3GS as it was of the original phone.

I can’t count the number of people who’ve told me they’re getting an iPhone—as soon as it becomes available on their current carrier, which is inevitably Verizon.

In the 10 years or so (pre-iPhone) that Verizon was my carrier, I never experienced a dropped call that seemed random. Sure, there were occasional dead spots—the New York State Thruway a few minutes north of the Harriman toll plaza, for example—but calls were (forgive the term) reliably dropped there, and never in other frequently-travelled places. And sure, once in a while I couldn’t make a call, but whenever the carrier opened a connection, it stayed up for the duration of the call, no matter how long.

Verizon seems to have such confidence in its network that it is now allowing Skype calls on a large fraction of its smartphones.

That is, the carrier is letting subscribers use the 3G data network to make voice-over-IP calls, which cuts down on the number of minutes a user needs and loads the data network with those calls instead. That’s right: fewer billable minutes, more unbilled data. Other than a better user experience for its customers, there’s absolutely nothing good about this from Verizon’s point of view.

True, the carrier is making the best of some upcoming rule changes at the U.S. Federal Communications Commission, but it says a lot that Verizon is just going ahead with the change early, and doing so in the most open and user-friendly way, while AT&T fights the rule change tooth and nail.

I don’t miss Verizon’s chaotic billing practices (the ones that are so bad they’ve inspired a Website called Verizonpathetic.com), but I do miss being able to call Customer Service late at night and on Sunday, yet another way in which AT&T's user experience falls short. Most of all, I miss Verizon’s rock-solid network.
 

Electric Power Plant Explosion Reveals History's Biggest Lesson

Who was it that said, “We learn from history that we learn nothing from history”? Proof of their wisdom came on 7 February, three days after the U.S. Chemical Safety Board held a public hearing to present its initial findings regarding the cause of the natural gas explosion that rocked a ConAgra Food plant in Garner, N.C., the previous June. (The ConAgra plant is the world’s sole producer of Slim Jim, the beef jerky snack with the dimensions of two or three cigarettes put together end to end.)

Just after 11 a.m., a huge explosion and fire were set off at the Kleen Power Plant in Middletown, Conn., when workers there attempted to purge a natural gas pipeline by venting the inert gases used to test whether there are any leaks in the system. Where did they vent the gas? Indoors. The last stage of the test involves flushing out the inert gases with natural gas. Workers apparently relied on their senses of smell to tell them when the inert gas had been fully expelled. The thinking was that the same additive that allows you to smell the gas emitted from your stovetop when there is no flame to consume it would be a good enough indicator that the purging was complete. But it’s now apparent that well before the smell of gas told the workers to shut off the gas or shut the release valve, it had built up to an explosive concentration and was set off by an as yet undetermined ignition source.

The five people who were killed and the 12 others who were injured by the blast and subsequent blaze at the Kleen Power Plant needn’t have suffered those fates. At the Chemical Safety Board hearing in Raleigh, N.C., a few days earlier, experts highlighted several lapses in judgement that together set the stage for the ConAgra explosion that killed three people, injured 71 others, and caused a worldwide Slim Jim shortage. They were basically the same lapses that would doom the workers at the Connecticut power plant.

First of all, the experts noted, natural gas should be vented outdoors, where there is less of a possibility that it might collect and explode. Why it took an august panel of the keenest minds to figure that one out remains a mystery.

The board also recommended that fire safety personnel be present whenever flammable gas lines are being vented and that everyone participating in the procedure is trained in how to do the job properly. Having people on site who know what they’re doing? What a novel idea!

Another important suggestion was that widely available electronic sensors similar to the carbon monoxide detectors in residences be used during this type of operation. What’s known but often overlooked is that relying on the chemical added to gas to make us able to sniff it out is a deadly gamble. Our olfactory sense is set up to respond to changes. After you’ve been in a room with a scent for a while, your brain basically relegates it to the background, a phenomenon known as odor fade. Bottom line: lives were lost and millions of dollars of damage was done for want of a US $40 gadget that anyone can buy on Amazon.com.

Though blowing a hole through the side of a billion-dollar electric power plant is nothing to sneeze at, the Middletown explosion didn’t have as great an immediate impact on commerce as the Great Slim Jim Famine of 2009. The Kleen Energy plant, whose construction was said to be 95 percent complete, was being built as part of a decade-long effort to improve southwest Connecticut’s access to electric power. That region—where peak energy demand grew 27 percent between 1999 and 2004—had few local generating facilities and was served by an overmatched patchwork of 115 kilovolt transmission lines. Those lines were the area’s only links to the ISO-New England power pool that had 345-kilovolt cables covering the rest of the state, plus Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont.

When the power plant was approved by the Connecticut Siting Council, it was envisioned as an additional power source that would feed the two 345 kV lines stretching deep into the state’s southwest corner that were also on the drawing board. It would, its backers argued, prevent repeats of the multiple summer days when municipalities such as New Haven, Norwalk, and Stamford suffered brownouts (wherein a reduction of voltage dims the lights) and rolling blackouts (where power is interrupted for a brief period to keep the system from completely crashing).

But a funny thing happened on the way to the Kleen Energy plant’s completion. The global economic downturn that began in 2008 upset power marketers’ projections about demand growth. With businesses going under, others scaling back production, and fewer people moving to the area as the boom went bust, peak demand moved back from the brink. The generating facility was suddenly no longer sorely needed as of yesterday. It could wait well beyond tomorrow.

“Any delay in bringing the Kleen Power facility online will have absolutely no effect on our ability to supply electric power to our customers, nor will it affect the price they pay for electricity,” says Mitch Gross, a spokesman for Connecticut Light & Power, southwestern Connecticut’s main utility. “We have a capacity contract [with Kleen Energy’s operators] ensuring that the power would be there if needed, but we’ve already purchased all the power we’ll need for 2010 and most of what we’ll need to meet demand in 2011.”

ISO-New England notes that surplus capacity is available—and not just in Connecticut. “Generating facilities across the country—especially coal-fired ones in the Midwest—are being idled because of insufficient demand,” says Erin O’Brien, a spokesperson for the power pool. O’Brien notes that the demand growth in southwestern Connecticut prior to 2008 would have made the Kleen Power plant’s use as a base load facility a foregone conclusion. Now? “It will run regularly or at intervals, depending on market and system reliability needs,” says O’Brien.

All the more reason why the loss of life on that Sunday morning—where the likely proximate cause was cut corners—was so tragic.

 

Photo: Douglas Healey/Getty Images

Toyota's Troubles Put EMI Back Into The Spotlight

It’s the floor mats. No, it’s sticky pedals. It’s definitely not the electronics.

That’s what Toyota says about the sudden acceleration problem that led to the recent raft of recalls. But some drivers, newspaper reporters, and engineering professors have their doubts, indicating that interference from electronic devices, built into the car, carried into the car, and even outside the car could be causing the problem.

Toyota has said it proved its point by having an outside firm test the electronics on various car models. The firm reportedly fooled around with engine power and various controls within the car seeing if some odd combination created a surge, but could not generate the problem; Toyota’s press releases about the test didn’t say a word about EMI.

Ahhh, does this bring back memories. Back in the ‘90s, a growing reliance of new aircraft on electronic controls was on a collision course with a growing tendency of passengers to bring multiple personal electronic devices on board. A theoretical potential for interference certainly existed—that’s why air passengers are asked to turn off all electronic devices during takeoff and landing, for those are the most unforgiving parts of the journey. And lots of folks thought—and still think—this restriction is silly.

I reported on this with my colleague Linda Geppert back in 1996, for the article “Do Portable Electronics Endanger Flight,” (IEEE Spectrum, September, 1996). And our conclusion was that yes, electronic devices can cause trouble.

However, that trouble is hard to create on cue. The culprit devices perhaps had to be a little out of spec, generating just a little more RF interference than it should—maybe one that slipped out of the factory that way, or one you dropped and thought had escaped damage. (In personal experience, my husband once had to take a new laptop back because turning it on took out the picture on the TV in the next room.) The position of the device mattered—move it around in the plane and the effect would come and go.

Here’s an example of one report we pulled from a database kept by NASA’s aviation safety reporting system, quoted from the article:

In March, 1993, a large passenger aircraft was at cruise altitude just outside the Dallas-Fort Worth International Airport when the No. 1 compass suddenly precessed 10 degrees to the right. The first flight attendant was asked to check whether any passengers were operating electronic devices. She said that a passenger in seat X had just turned on his laptop computer.

The report continues, “I asked that the passenger turn off his laptop computer for a period of 10 minutes, which he did. I then asked that the passenger turn on his computer once again. The No. 1 compass immediately precessed 8 degrees to the right. The computer was then turned off for a 30-minute period during which the No. 1 compass operation was verified as normal.”

So yes, RF emissions from electronic devices can cause weird things to happen. Is it happening in the Toyota case? Toyota says no. And, to date, no real evidence says otherwise.

But if I were Toyota, I sure would be checking it out. Because it does look suspicious.

Which brings me to Apple co-founder Steve Wozniak. The Woz has been complaining that his Prius occasionally surges randomly when it’s supposed to be holding steady under cruise control. My first reaction when I heard this was that he was setting up good story to give to the next cop who tickets him for speeding.

And then I remembered: the Woz is a gadget frea'. He inevitably is carrying multiple smart phones (never just one) and countless other consumer electronic gadgets. His friends love to surprise him with gizmos they’ve picked up in other countries that are not yet available in the U.S. And whenever I run into him he immediately shows me his current favorite gadget.  There is no way he turns off all these devices every time he gets in his car. He’s a walking RF fog.

So here’s a suggestion for Toyota. Forget those expensive outside research firms testing your electronics. Line up about a hundred cars, hook up everything electronic that you can measure, and then just let Woz walk by. If his presence has absolutely no effect, then I’ll believe that it’s a “mechanical problem.”

The Latest Barbie is a Computer Engineer

Barbie is now a computer engineer. This represents Barbie’s 126th career change. It’s a late-in-life choice; Barbie is nearly 51.  It took her a while, perhaps, to get her degree—in 1992, a talking Barbie said “math class is tough.” Computer engineer beat out surgeon, environmentalist, and architect in an online poll, coming in second only to news anchor among girls, but with enough support from adults that Mattel went ahead with both dolls.

Computer Engineer Barbie could be popular with the Barbie set (which these days skews towards kindergartners and preschoolers; it seems to get younger every year), if only for her accessories. She’s got glasses (Mattel missed a bet, though; the hip geeks might be more likely to wear lens-less 3-D glasses rather than the pink ones Barbie sports), a Bluetooth earpiece, a t-shirt printed with binary code, a smart phone, and a laptop. 

I’d like to say today’s introduction of Computer Engineer Barbie honors National Engineers Week which kicks off Sunday, but more likely it was timed to coincide with Toy Fair, also opening this weekend.

Iron Man

After Dan Whaley sold his travel website for $750 million in 2000, he spent his time and money exploring the world. He biked solo across the U.S.  He lived with Tibetan refugees in Nepal.   He drove from California to Argentina.  Now he's on a new mission that combines his passions: saving the world…by geo-engineering it.

Whaley's start-up, Climos, is spearheading a program to remove CO2 from the atmosphere through a process called Ocean Iron Fertilization. The idea is to add iron to iron-limited parts of the ocean in order to facilitate the growth of phytoplankton, which absorbs CO2.  Climos already has the backing of heavy-hitter venture capitalists and scientists from the National Science Foundation to Elon Musk, co-founder of PayPal. Some say it's bad - if not crazy - science.  A Greenpeace ocean specialist calls the plan "dangerous and irresponsible." Others say it must be taken seriously. 

Whaley is the guy at the center of all this, and oceanography is in his genes.  His chief science officer at Climos is his mother, Dr. Margaret Leinen, formerly the assistant Director for Geosciences at the National Science Foundation.   A autodidactic computer programmer, Whaley coded one of the first software systems to detect reversals of the earth's magnetic pole in cryogenically cooled deep-sea sediment samples for the University of Rhode Island's Graduate School of Oceanography. He has also participated in two NSF-funded Joint Global Ocean Flux Study cruises in the Equatorial Pacific where he helped run a piston-coring apparatus to obtain seafloor samples along the Tahiti-Hawaii transect.

Based in San Francisco, Whaley’s now traveling the world meeting with scientists and regulators to pave the way for ocean fertilization.  But the road is already littered with the one other start-up that went down this path.   A couple months before Climos launched, an ocean-seeding start-up called Planktos suspended its project citing “a highly effective disinformation campaign waged by anti-offset crusaders [that] has provoked widespread opposition to plankton restoration in the environmental world.”  While Planktos failed to raise funds, Whaley has already secured $3.5 million.  His success comes in no small part to his family’s background, his Silicon Valley contacts, and his persuasive skills.  “It’s about science first, but it’s also about perception, and people have to really trust that you’re trying to do this in the right way,” he says. “Part of how they think about science has to do with how they feel about the way you’re going about things.”   The first project is aimed for the southern oceans, mostly likely near Argentina, Brazil, or Australia.

Topsy's Twitter Search Now Finds Photos and Text

Like several start-up search engines that offer third party Twitter search, Topsy used to find just the web links shared via Twitter, rather than returning actual content from tweets.

That changed yesterday with Topsy's launch of several new features that offer photo search and regular tweet content search.

From TechCrunch:

Before now, if you ran a search for “Google Buzz”, the site would return links to articles and videos about the new service. Now, it will also surface tweets from influential Twitter users, even if they don’t include a link. That’s important for breaking news when a story may not have already been covered by a publication, or when there’s a tweet that’s important in and of itself (say, Bill Gates’ first tweet). 

The new features will also allow photo search based on the text of tweets that contain photo links, and will let users see what's trending in web links, photos, and tweet content.

Topsy's algorithms rank the relevance of tweets and filter out less influential ones, according to TechCrunch:

The links are ranked by the number of times they’ve been retweeted, and also by the influence of the people who have tweeted them; the site actually keeps track of the number of retweets each user typically gets to establish their overall reputation.

That's a technique that major search engines are still trying to perfect as they roll out real-time search features.

Read IEEE Spectrum's report on the progress with real-time search.

All the Processing, Half the Power

Last August, IEEE Spectrum ran a feature article by researchers from the University of Michigan at Ann Arbor and ARM Holdings in Cambridge, England, who reported their work aimed at marshaling all of a microprocessor’s abilities, leaving hardly any reserve. They discussed sophisticated fault-monitoring techniques that allow chips to operate close to the point at which performance-harming timing errors start to crop up. Computer makers, they said, would soon be able to skirt the razor’s edge of chip reliability by correcting for the occasional error while overclocking a chip to boost processing speed or while running it on much less power in order to gain more energy efficiency.

On 9 February, the Michigan-ARM team stepped forward with a game-changing announcement. The researchers presented a paper at the International Solid State circuits conference (ISScc) reporting that they used fault monitoring and a group of complementary energy-saving techniques--such as shutting off the clock signal in the regions of the chip that aren’t crunching numbers--to maintain the performance and reliability of a 1-GHz chip running on 52 percent less power than it’s rated for. 

Advertisement

Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.

Advertisement
Load More