How to Double the Power of Solar Panels

KEVIN BULLIS

Tuesday, October 16, 2012

In an attempt to further drop the cost of solar power, Bandgap Engineering, a startup in Woburn, Massachusetts, is developing a nanowire-based solar cell that could eventually generate twice as much power as conventional solar cells.

That’s a long-term project, but meanwhile the company is about to start selling a simpler version of the technology, using silicon nanowires that can improve the performance and lower the cost of conventional silicon solar cells. Bandgap says its nanowires, which can be built using existing manufacturing tools, boost the power output of solar cells by increasing the amount of light the cells can absorb.

Right now most solar-panel manufacturers aren’t building new factories because the market for their product is glutted. But if market conditions improve and manufacturers do start building, they’ll be able to introduce larger changes to production lines. In that case the Bandgap technology could make it possible to change solar cells more significantly. For example, by increasing light absorption, it could allow manufacturers to use far thinner wafers of silicon, reducing the largest part of a solar cell’s cost. It could also enable manufacturers to use copper wires instead of more expensive silver wires to collect charge from the solar panels.

These changes could lead to solar panels that convert over 20 percent of the energy in sunlight into electricity (compared with about 15 percent for most solar cells now) yet cost only $1 per watt to produce and install, says Richard Chleboski, Bandgap’s CEO. (Solar installations cost a few dollars per watt now, depending on their size and type.) Over the operating lifetime of the system, costs would come to between 6 and 10 cents per kilowatt-hour. That’s still higher than the current cost of natural-gas power in the United States, which is about 4 cents per kilowatt-hour. But it’s low enough to secure solar power a substantial market in many parts of the world where energy costs can be higher, or in certain niche markets in the United States.

Meanwhile, Bandgap is pursuing technology that could someday improve efficiency enough to allow solar power to compete widely with fossil fuels. Double the efficiency of solar cells without greatly increasing manufacturing costs, and you substantially lower the cost per watt of solar panels and halve the cost of installation—currently the biggest expense in solar power—by making it possible to get the same amount of power out of half as many cells.

Both the cells Bandgap is about to introduce and the cells it hopes to produce in the long term are based on the idea of minimizing the energy loss that typically occurs when light passes through a solar cell unabsorbed or when certain wavelengths of light are absorbed but don’t have enough energy to dislodge electrons to create electricity. (That energy is wasted as heat.) In a conventional solar cell, at least two-thirds of the energy in sunlight is wasted—usually much more.

The company’s existing technology makes use of the fact that when light encounters the nanowires, it’s refracted in a way that causes it to bounce around in the solar cell rather than simply moving through it or bouncing off it. That increases its chances of being absorbed (see “Black Silicon Solar Cells to Capture More Light“).

But what Bandgap ultimately wants to do is to change the way light is converted to electricity inside the cell. If the nanowires can be made uniformly enough, and if they can be formed in such a way that their atoms line up along certain planes, the tiny structures could change the electronic properties of silicon. These changes could allow solar cells to generate electricity from low-energy light that normally produces only heat, says Marcie Black, the company’s founder and chief technology officer. It does this in part by providing a way to combine energy from more than one photon of low-energy light.

The technology could take many years to develop. For one thing, it requires very precise control over the properties of each of millions of nanowires. Also, the techniques needed to make the solar cells might not be cheap or reliable enough to produce them on a large scale. But such solar cells could theoretically convert 60 percent of the energy in sunlight into electricity. That will be hard to achieve in practice, so the company is aiming at a more modest 38 percent efficiency, which is still more than twice that of typical silicon solar cells made now.

Researchers are taking several other approaches to producing very high-efficiency solar cells, such as using quantum dots or combining several kinds of materials (see “TR10: Nanocharging Solar” and “New Materials Make Photovoltaics Better“). The nanowire technology could be simpler, however. “In theory, the approach has many potential advantages, but you’ve got to get it to work,” says Andrew Norman, a senior researcher at the National Renewable Energy Laboratory in Golden, Colorado. Bandgap hasn’t yet built solar cells using the approach it hopes to pursue in the long term, but it’s made indirect measurements showing that its nanowires can change the electronic properties of silicon. “This is still in the research phase,” Black says. “We’re being very honest with investors—there’s still a lot of work to do.”

A Grand Experiment to Rein In Climate Change

By 
Published: October 13, 2012 239 Comments
LEGGETT, Calif. — Braced against a steep slope, Robert Hrubes cinched his measuring tape around the trunk of one tree after another, barking out diameters like an auctioneer announcing bids. “Twelve point two!” “Fourteen point one!”

Robert Hrubes measures the circumference of trees to determine how much carbon they can store, one approach to reducing greenhouse gases in the atmosphere.

Articles in this series will examine the state’s new carbon trading law.

A blog about energy and the environment.

Mr. Hrubes, left, and Tom Tuchmann with a map of the Usal Redwood Forest. The foundation that owns the forest is considering selling credits under California’s new “cap and trade” law.

Mr. Hrubes’s task, a far cry from forestry of the past, was to calculate how much carbon could be stored within the tanoak, madrone and redwood trees in that plot. Every year or so, other foresters will return to make sure the trees are still standing and doing their job.

Such audits will be crucial as California embarks on its grand experiment in reining in climate change. On Jan. 1, it will become the first state in the nation to charge industries across the economy for the greenhouse gases they emit. Under the system, known as “cap and trade,” the state will set an overall ceiling on those emissions and assign allowable emission amounts for individual polluters. A portion of these so-called allowances will be allocated to utilities, manufacturers and others; the remainder will be auctioned off.

Over time, the number of allowances issued by the state will be reduced, which should force a reduction in emissions.

To obtain the allowances needed to account for their emissions, companies can buy them at auction or on the carbon market. They can secure offset credits, as they are known, either by buying leftover allowances from emitters that have met their targets or by purchasing them from projects that remove carbon dioxide or other greenhouse gases from the atmosphere, like the woods where Mr. Hrubes was working.

Dozens of verifiers from different fields, from chemists to accountants to foresters, will be the first line of defense in making sure the benefits are real.

Mr. Hrubes said his goal in any audit was to ensure that the forest’s owner was “being conservative whenever a judgment call has to be made” in calculating greenhouse gas reductions.

The outsize goals of California’s new law, known as A.B. 32, are to lower California’s emissions to what they were in 1990 by 2020 — a reduction of roughly 30 percent — and, more broadly, to show that the system works and can be replicated.

The risks for California are enormous. Opponents and supporters alike worry that the program could hurt the state’s fragile economy by driving out refineries, cement makers, glass factories and other businesses. Some are concerned that companies will find a way to outmaneuver the system, causing the state to fall short of its emission reduction targets.

“The worst possible thing to happen is if it fails,” said Robert N. Stavins, a Harvard economist.

Just three years ago, California’s plan was viewed as a trial run for a national carbon market that one day might tie into existing markets in Europe and elsewhere. President Obama’s first budget proposal included a cap-and-trade program to cut national greenhouse gas emissions 14 percent by 2020; the House later passed an energy andclimate bill that incorporated such a program.

But in 2010, political forces backed by the biggest emitters, oil and coal companies, blocked the plan in the Senate. In that year’s midterm elections, conservative Republicans disavowed their party’s role in creating similar programs; they continue to deride it as “cap and tax.”

California air regulators are proud of their record in leading the nation to new auto emissions standards in the 1960s and efficiency standards for appliances in the 1970s. And so the pressure is on the state’s Air Resources Board to get this right.

At first, only four means of carbon reduction will be approved for offset credits: timber management, the destruction of coolant gasescuts in methane emissions from livestock waste and tree planting projects in urban areas. Already, developers of offset projects in more than 20 states are preparing to enter the new market, which for now accepts only credits generated in the United States. Some projects send coolant gases to be destroyed at an incinerator in Arkansas; others, tied to dairies in states like Ohio, Virginia and Wisconsin, will capture methane from livestock waste.

Most of these projects already sell offset credits in other markets like the Regional Greenhouse Gas Initiative, a cap-and-trade program covering utilities in the Northeast.

But offsets can be prone to misuse; some have generated significant private profits while producing questionable environmental benefits. The European Union’s eight-year-old carbon trading market has been tarnished by fake credits and audits that failed to meet minimum standards. California’s offsets have already been challenged in court by environmentalists who argue that offset developers will earn money for actions that they would have taken even if the program did not exist.

“If there is a loss of confidence because there is a sense that people have been cheating and the offsets are not real, that will be a problem,” said Kevin Kennedy, an economist with the World Resources Institute in Washington.

That is why there is such a need for qualified verifiers. This summer, four foresters from around the country gathered in a Los Angeles suburb for a $2,900 test-preparation course to master the new system in advance of a required state test.

All had experience in verification in other carbon trading systems — so much so that they offered their instructors sharp critiques on the 111 pages of rules. One even challenged the algorithms central to the forest benefit calculations.

“If they don’t get the equations right, there could be a real problem,” said Terese Walters, a forester from Oregon. She is hoping that having California credentials will lead to lucrative opportunities. Ms. Walters and Caitlin Sellers, a forester from Florida in the class, both work for Environmental Services of Jacksonville, Fla., one of the country’s largest environmental consulting firms. David Bubser, another student, is a Minnesota forester and a regional manager for the nonprofit Rainforest Alliance.

There are several basic requirements for a forest offset. Credits cannot be granted for preserving trees that were going to be left standing anyway. The change must be long-lasting: trees must be left intact for a century. And owners must hire accredited verifiers to audit their claims.

The offset marketplace is already beginning to hum as companies gear up for California’s rollout.

Independent verifiers can make $800 to $1,200 a day, according to Mr. Bubser. Scientific Certification Systems, Mr. Hrubes’s employer, which verified 4.2 million tons of carbon offsets around the world last year, added two foresters this summer, for a total of six.

Sacramento’s municipal utility recently held a conference call with potential vendors of credits to offset some of the 1.2 million tons of carbon dioxide emitted annually from its gas-fired power plant — possibly by buying 200,000 credits annually.

Utility officials made it clear during the call that the more measurable and reliable the offset, the more valuable it would be. The administrators of California’s program have set a floor price for allowances at $10 per metric ton of emissions during the first auction in November. Once the program gets going, the actual value of allowances will fluctuate as they are traded.

The Redwood Forest Foundation, created to promote sustainable forestry but also to keep timber jobs in Mendocino County, is considering selling offset credits. Its biggest asset is the 50,000-acre Usal Redwood Forest, where Mr. Hrubes was working, which the foundation acquired in 2007 with a $65 million bank loan. The foundation needs to pay down its debt. It reaped $19.5 million selling a conservation easement last year, but the idea of a new revenue source is alluring.

“When you need an economic return, one way is to maximize timber harvest,” said Tom Tuchmann, the group’s acting executive director. “The other way is to look at nontraditional value streams.”

But making strategic decisions about how many trees to harvest and how many to use to lock up carbon is an uncertain business. Other carbon markets have generally not done well by investors, and some brokerages have closed their carbon desks.

“There are so many people who are disappointed,” said Thaddeus Huetteman, the president of Power and Energy Analytic Resources of Atlanta. “What they are really looking for is for California to show we can create a new market of significance in the world’s ninth-largest economy.”

Brain Implant Detects, Responds to Epilepsy

SUSAN YOUNG

Friday, October 12, 2012

Next year, medical researchers will test in patients a one-of-a-kind brain implant that can sense electrical activity in the brain while simultaneously emitting electric pulses, says device developer Medtronic.

Deep-brain stimulators are mainly used to regulate the movement problems associated with Parkinson’s and other diseases, but they are also used in Europe and Canada to treat epilepsy and are being used experimentally to treat severe depression and obsessive-compulsive disorder. But doctors must use trial and error to determine the best parameters for the electrical stimulation programmed into each patient’s chip.

The smarter brain stimulator is an improved version of Medtronic’s existing deep-brain stimulator device, which has already been implanted in more than 80,000 people around the world. Medtronic has added an extra chip so that it can detect electrical activity and respond automatically to changes in the brain.

“If you are in the brain already, you might as well take advantage of the fact that you can listen in,” says Lothar Krinke, who manages the Deep Brain Stimulation division at Medtronic. This means the device could respond automatically when a patient’s symptoms grow stronger, or could turn itself off when the patient is asleep. “We really only want to deliver the electricity when it is needed,” says Krinke. The company has tested the device in lab animals and says that next year outside teams of researchers will test it in patients with diseases such as Parkinson’s and epilepsy.

Although invasive, these sorts of neural implants are vital for patients who otherwise fail to respond to medication, says Dwayne Godwin, a neuroscientist who studies epilepsy at Wake Forest School of Medicine. “Not every patient responds in the same way to treatment,” he says. “As these devices become better established, we will get a better understanding of which are better for certain types of disorders.”

Other brain implants have the ability to sense electrical activity and stimulate the brain, just not at the same time. For example, NeuroPace, a medical-device startup in Mountain View, California, has developed a brain implant that spends most of its time monitoring the brain for an oncoming seizure (see “Zapping Seizures Away”). When an impending seizure is detected, the device, which is currently in clinical trials, delivers imperceptible pacemaker-like shocks that prevent the disruptive activity from spreading and causing a seizure.

A system that can sense and stimulate at the same time could be useful in patients whose disease symptoms fluctuate over time, as is often the case in Parkinson’s patients, says NeuroPace CEO Frank Fischer. “I think it’s a very interesting research tool to be able to look at applications such as movement disorders, where changes may be naturally occurring and a patient could benefit from different levels of stimulation,” he says.

Krinke says adding sensing capability to the deep-brain stimulator could also help determine whether the implant is still functioning properly when a patient’s symptoms worsen, which could either be due to progression of disease or device failure. “The device can self-diagnose whether it is broken,” says Krinke.

Knowing whether a patient’s disease is worsening is more of a challenge, he says, but as researchers continue to use the device to study brain circuits relevant to disease states, eventually the device might become a diagnostic tool. “The future is that we can measure electrical signals that are related to disease progression,” he says.

Super-Cheap Health Tests

DAVID TALBOT

Monday, October 15, 2012

Diagnostics for All, a nonprofit in Cambridge, Massachusetts, is making a test for liver damage that could cost just pennies. It consists, remarkably, of a stamp-size square of paper with wells that change color when a drop of blood is applied.

The test could provide an enormous benefit in poor countries, where liver damage is widespread as a side effect of drugs administered to HIV and tuberculosis patients. (As many as one-fourth of people taking antiretroviral drugs in the poor world develop liver problems—five times the rate elsewhere.) The liver function tests administered regularly in the developed world require tubes of blood, lab equipment, and electricity. The paper chip from Diagnostics for All needs none of that.

The test uses patterned channels and wells to allow for filtering and multistep reactions; the technology originated in the lab of Harvard chemist George ­Whitesides, who pioneered this method, and was licensed from Harvard (see “Paper Diagnostics“). The paper absorbs sample fluids and uses capillary action to convey them to the test wells imprinted on it. These wells are spotted with chemicals that change color when they react with certain markers in a liquid.

The chip is meant to work simply with little additional equipment, making it suitable for the poorest regions. “This is a world in which there are very few resources—that is to say, almost no money, very few doctors, no electricity in many places, no refrigeration,” Whitesides says. “The conditions are such that it’s very difficult to imagine how you deliver even pretty straightforward health care.”

Five years after the company was formed, Diagnostics for All, which is led by biotech executive Una Ryan and sustained by grants from the Gates Foundation and others, is moving toward a viable product. The first trial of the liver test is in progress on HIV patients at a hospital in Vietnam. Funding, manufacturing, and distribution models are still being worked out, but the company can make between 500 and 1,000 tests per day at its Cambridge facility and hopes to obtain regulatory approvals so that the liver test can reach patients by 2014, says Jason Rolland, who leads engineering efforts as the company’s senior director of research.

DFA (as the company is known) is working on other paper-based diagnostics: an assay that detects antigens for multiple diseases, including malaria and dengue fever; a test for preëclampsia in pregnant women; and even nucleic-acid tests to detect pathogens in blood. The company is also developing tests that farmers could use to check for foodborne toxins. In all cases, results can be interpreted by a clinician or a smartphone app, after which test patches can be incinerated. They are, after all, just paper.

Japan’s Softbank snaps up Sprint in $20 billion deal

By Mari Saito and Tim Kelly and Nicola Leske

Mon Oct 15, 2012 11:57am EDT

(Reuters) – Japanese mobile operator Softbank Corp said it will buy about 70 percent of Sprint Nextel Corp, the third-largest U.S. carrier, for $20.1 billion – the most a Japanese firm has spent on an overseas acquisition.

The deal, announced by Softbank’s billionaire founder and chief Masayoshi Son and Sprint Chief Executive Dan Hesse at a packed news conference in Tokyo on Monday, gives Softbank entry into a U.S. market that is still growing, while Japan’s market is stagnating.

Part of the deal is a direct infusion of billions of dollars into Sprint, giving it the firepower to buy peers and build out its 4G network to compete in a market dominated by AT&T Inc and Verizon Wireless.

Shares in one of those potential targets, Clearwire Corp, surged 35 cents to $2.67 near midday. Sprint owns 48 percent of Clearwire. While Softbank said no action was required, most analysts and investors see a Sprint-Clearwire tie-up as an inevitable consequence of the Softbank deal.

One way or another, analysts have long said the U.S. telecommunications industry needs to consolidate, but few looked to Japan as a catalyst. Some investors and rating agencies worried that Softbank is biting off more than it can chew.

But the 55-year-old Son, a rare risk-taker in Japan’s often cautious business circles, is betting U.S. growth can offer relief from cut-throat competition in Japan’s saturated mobile market. Combined, Softbank and Sprint will have 96 million users.

“It could be safe if you do nothing, and our challenge in the U.S. is not going to be easy at all. We must enter a new market, one with a different culture, and we must start again from zero after all we have built,” Son told the news conference.

“But not taking this challenge will be a bigger risk.”

FIREPOWER

The financing is complex, involving at least three steps with two entities as well as a debt conversion.

Softbank’s newly created U.S. subsidiary New Sprint will buy $3.1 billion in old Sprint convertible bonds to start. After shareholders and regulators approve the proposed deal, Softbank will then buy $4.9 billion in New Sprint shares. The two steps together represent the $8 billion infusion directly into Sprint.

On top of that, 55 percent of existing Sprint shares would be exchanged for $7.30 per share in cash, representing a further $12.1 billion.

The transactions are to be completed by mid-2013, at which point New Sprint will be a publicly traded company and the old Sprint will survive as its subsidiary.

Sprint added 2 cents to $5.75 near midday after surging last week on the first reports of a pending deal. The offer, while a substantial premium, is still less than some observers had hoped. A fund manager at T. Rowe Price, a top-15 Sprint shareholder, told Reuters last week he thought Sprint would be worth $10 a share in 18 months.

Hesse, who will stay on as Sprint CEO, said the Softbank investment would give Sprint opportunities it hadn’t had since he joined the firm in late-2007, and enable the U.S. firm to play a bigger role in future market consolidation.

“This is pro-competitive and pro-consumer in the U.S. because it creates a stronger No. 3 … it competes with the duopoly of AT&T and Verizon. When you look at what Softbank has accomplished in Japan with the No. 3 carrier, it’s something we can learn from,” he said.

Hesse, one of the few corporate CEOs in America to star in his own company’s commercials, also acknowledged the financial challenges Sprint has faced – which the new capital could fix quickly.

“Sprint has been engaged in turnaround since 2008. We have been at a disadvantage due to our debt,” he said on an investor call Monday morning.

But it was not all serious, as the American took some time for a bit of levity with his Asian counterparts. At the press conference, Hesse noted it was the first time he had seen his long-time acquaintance Son with a tie on.

SOFTBANK WEAKENS

Softbank shares tumbled more than 8 percent on Monday before closing down 5.3 percent to their lowest finish in five months. The stock has lost more than one-fifth of its value – or $8.7 billion – since news first surfaced late last week about its interest in Sprint.

On Monday, credit rating agency Moody’s said it was reviewing Softbank’s ratings for a possible downgrade, but some analysts said Son’s gamble might pay off in the end.

“It’s the same (market) reaction as when Softbank said it was going to buy Vodafone a few years ago. Everyone came out and said it was far too expensive,” Fumiyuki Nakanishi, general manager of investment and research at SMBC Friend Securities, said ahead of the announcement.

Softbank bought Vodafone’s Japan unit for $15.5 billion in 2006.

“Son made a company worth 3 trillion yen, and now it will be worth 6 trillion yen. That’s quite impressive, and I think investors will realize he’s making the right decision down the road,” said Nakanishi.

Four banks – Mizuho Financial Group Inc, Sumitomo Mitsui Financial Group, Mitsubishi UFJ Financial GroupM and Deutsche Bank – have approved loans totaling 1.65 trillion yen ($21.1 billion) to Softbank, three sources with direct knowledge of the matter told Reuters.

Sprint, which has lost money in its last 19 quarters, has net debt of about $15 billion, while Softbank has net debt of about $10 billion.

Brokers have warned the deal could leave Softbank with “unacceptably high” gearing – a ratio of debt to shareholder capital.

Standard & Poor’s has warned the deal “may undermine Softbank’s financial risk profile” and pressure its free operating cash flow for the next few years.

Reflecting the concerns, Softbank’s 5-year credit default swap spreads – the cost of protecting its debt against default – widened to 267/327 basis points from around 160 basis points before the deal, and yields on its yen bonds have risen sharply.

NO CLEARWIRE OBLIGATION

Analysts have said the Softbank acquisition of 70 percent of Sprint for $20 billion would imply the No. 3 U.S. wireless company is worth about $28.6 billion, some two-thirds more than its market capitalization at Friday’s close.

Sprint is going through a $7 billion upgrade of one of its networks, while closing its Nextel iDen network, which makes Softbank’s capital especially useful. But the Clearwire question looms large as well.

Macquarie analyst Kevin Smithen, in a note to clients, described Softbank as the “white knight” that could give Clearwire management and investors a successful exit, though he also warned the company may drive a hard bargain in negotiations.

An alliance with Sprint could also give Softbank leverage when dealing with Apple Inc, helping bolster its domestic position against KDDI Corp, which also offers the iPhone in Japan, and market leader NTT Docomo, which has yet to offer the Apple smartphone.

With Sprint in hand, Softbank may also look to acquire smaller U.S. carrier MetroPCS Communications Inc, Japanese media have reported.

Sprint has long had an interest in MetroPCS, which earlier this month agreed to merge with T-Mobile USA, part of Deutsche Telekom AG.

The Sprint deal takes outbound deals by Japanese firms to a record $75 billion this year, according to Thomson Reuters data, underscoring a strong appetite for overseas assets seemingly unaffected by signs of slowing global growth.

This is not the first Japanese foray into telecoms overseas. NTT Docomo racked up big losses after a string of failed investments in names like AT&T Wireless and Taiwan mobile operator KG Telecom in the late 1990s and early 2000s.

Raine Group LLC, a boutique merchant bank focused on the technology, media and telecoms sector, and Mizuho Securities were lead financial advisers to Softbank.

($1 = 78.3550 Japanese yen)

(Writing by Linda Sieg and Ben Berkowitz; Editing by Jeffrey Benkoe)

Google Puts Its Virtual Brain Technology to Work

TOM SIMONITE

Friday, October 5, 2012

This summer Google set a new landmark in the field of artificial intelligence with software that learned how to recognize cats, people, and other things simply by watching YouTube videos (see “Self-Taught Software“). That technology, modeled on how brain cells operate, is now being put to work making Google’s products smarter, with speech recognition being the first service to benefit.

Google’s learning software is based on simulating groups of connected brain cells that communicate and influence one another. When such a neural network, as it’s called, is exposed to data, the relationships between different neurons can change. That causes the network to develop the ability to react in certain ways to incoming data of a particular kind—and the network is said to have learned something.

Neural networks have been used for decades in areas where machine learning is applied, such as chess-playing software or face detection. Google’s engineers have found ways to put more computing power behind the approach than was previously possible, creating neural networks that can learn without human assistance and are robust enough to be used commercially, not just as research demonstrations.

The company’s neural networks decide for themselves which features of data to pay attention to, and which patterns matter, rather than having humans decide that, say, colors and particular shapes are of interest to software trying to identify objects.

Google is now using these neural networks to recognize speech more accurately, a technology increasingly important to Google’s smartphone operating system, Android, as well as the search app it makes available for Apple devices (see “Google’s Answer to Siri Thinks Ahead“). “We got between 20 and 25 percent improvement in terms of words that are wrong,” says Vincent Vanhoucke, a leader of Google’s speech-recognition efforts. “That means that many more people will have a perfect experience without errors.” The neural net is so far only working on U.S. English, and Vanhoucke says similar improvements should be possible when it is introduced for other dialects and languages.

Other Google products will likely improve over time with help from the new learning software. The company’s image search tools, for example, could become better able to understand what’s in a photo without relying on surrounding text. And Google’s self-driving cars (see “Look, No Hands“) and mobile computer built into a pair of glasses (see “You Will Want Google’s Goggles“) could benefit from software better able to make sense of more real-world data.

The new technology grabbed headlines back in June of this year, when Google engineers published results of an experiment that threw 10 million images grabbed from YouTube videos at their simulated brain cells, running 16,000 processors across a thousand computers for 10 days without pause.

“Most people keep their model in a single machine, but we wanted to experiment with very large neural networks,” says Jeff Dean, an engineer helping lead the research at Google. “If you scale up both the size of the model and the amount of data you train it with, you can learn finer distinctions or more complex features.”

The neural networks that come out of that process are more flexible. “These models can typically take a lot more context,” says Dean, giving an example from the world of speech recognition. If, for example, Google’s system thought it heard someone say “I’m going to eat a lychee,” but the last word was slightly muffled, it could confirm its hunch based on past experience of phrases because “lychee” is a fruit and is used in the same context as “apple” or “orange.”

Dean says his team is also testing models that understand both images and text together. “You give it ‘porpoise’ and it gives you pictures of porpoises,” he says. “If you give it a picture of a porpoise, it gives you ‘porpoise’ as a word.”

A next step could be to have the same model learn the sounds of words as well. Being able to relate different forms of data like that could lead to speech recognition that gathers extra clues from video, for example, and it could boost the capabilities of Google’s self-driving cars by helping them understand their surroundings by combining the many streams of data they collect, from laser scans of nearby obstacles to information from the car’s engine.

Google’s work on making neural networks brings us a small step closer to one of the ultimate goals of AI—creating software that can match animal or perhaps even human intelligence, saysYoshua Bengio, a professor at the University of Montreal who works on similar machine-learning techniques. “This is the route toward making more general artificial intelligence—there’s no way you will get an intelligent machine if it can’t take in a large volume of knowledge about the world,” he says.

In fact, the workings of Google’s neural networks operate in similar ways to what neuroscientists know about the visual cortex in mammals, the part of the brain that processes visual information, says Bengio. “It turns out that the feature learning networks being used [by Google] are similar to the methods used by the brain that are able to discover objects that exist.”

However, he is quick to add that even Google’s neural networks are much smaller than the brain, and that they can’t perform many things necessary to intelligence, such as reasoning with information collected from the outside world.

Dean is also careful not to imply that the limited intelligences he’s building are close to matching any biological brain. But he can’t resist pointing out that if you pick the right contest, Google’s neural networks have humans beat.

“We are seeing better than human-level performance in some visual tasks,” he says, giving the example of labeling, where house numbers appear in photos taken by Google’s Street View car, a job that used to be farmed out to many humans.

 “They’re starting to use neural nets to decide whether a patch [in an image] is a house number or not,” says Dean, and they turn out to perform better than humans. It’s a small victory—but one that highlights how far artificial neural nets are behind the ones in your head. “It’s probably that it’s not very exciting, and a computer never gets tired,” says Dean. It takes real intelligence to get bored.

Amazon to spend $1.16 billion to buy Seattle HQ

By Alistair Barr and Bill Rigby

Fri Oct 5, 2012 5:50pm EDT

(Reuters) – Amazon.com Inc said on Friday that it will spend more than $1 billion to buy its corporate headquarters in Seattle from Microsoft Corp co-founder Paul Allen’s investment firm.

The world’s largest Internet retailer plans to buy 11 buildings, comprising 1.8 million square feet of corporate office space, for $1.16 billion. Amazon said it expects to close the deal in the fourth quarter.

Amazon had been leasing the space. But at the end of August, Allen’s Vulcan Real Estate, part of his Vulcan Inc investment vehicle, put the buildings in Seattle’s South Lake Union area up for sale.

Allen, who co-founded Microsoft with school friend Bill Gates, has been the central figure in revitalizing the South Lake Union neighborhood, which was a semi-industrial wasteland until recently, but is now a thriving center for tech firms and features a number of fashionable restaurants.

Amazon already has plans to build new offices nearby to house its growing staff.

(Reporting By Alistair Barr)

Polls: Voters back clean energy, climate policies

3:26PM EST October 2. 2012 – On the eve of the first presidential debate, a flurry of new polls suggest most Americans support clean energy and policies to reduce climate change — topics that have garnered scant attention on the campaign trail.

Nine out of 10 registered voters (92%) said it was “very” or “somewhat” important for the United States to develop and use solar power, according to an online survey of 1,206 adults released Tuesday by the independent polling firm Hart Research Associates. This support spanned the political spectrum, including 84% of Republicans, 95% of independents and 98% of Democrats.

“The consistency is very impressive,” Molly O’Rourke, partner at Hart Research, told reporters during a news conference. She noted similar results when voters were asked how they view solar energy (85% favorably) and federal incentives for the industry (78% supportive). The Solar Energy Industries Association, a trade group, commissioned the survey.

The results, along with those of two other recent polls, come as President Obama and his GOP opponent, Mitt Romney, prepare for their first debate Wednesday. So far, climate change and clean energy have not been major campaign issues, but nine environmental organizations delivered more than 160,000 petitions Friday to the debate’s moderator, Jim Lehrer, urging him to ask about them.

Another new survey found that 7% of likely voters remain undecided about how they’ll vote, and most of them say global warming will be one of several important factors determining that, according to the Yale Project on Climate Change Communication and the George Mason University Center for Climate Change.

The undecideds are much more similar to likely Obama voters than likely Romney ones on climate change and energy-related attitudes and policy preferences, the survey found. For example, 80% say global warming is happening, compared with 86% for Obama backers and 45% for Romney supporters. The survey of 1,061 American adults, taken Aug. 31 to Sept. 12, was released Sept. 24.

Similarly, undecided voters in eight swing states — Florida, Michigan, Nevada, New Mexico, Ohio, Pennsylvania, Virginia and Wisconsin — favor presidential and congressional candidates who support clean air and clean energy policies over those who don’t, according to surveys of 22,412 likely voters released last week by Public Policy Polling.

By a roughly 2-to-1 ration (54% vs. 27%), these voters side with Obama’s view that the Environmental Protection Agency needs to set standards to lower carbon pollution, rather than Romney’s position that such limits would be bad for business and thus shouldn’t be imposed, according to the survey sponsored by the Natural Resources Defense Fund Action Fund, an environmental group.

The poll, taken Sept. 14 to Sept. 20, found that 50% of likely votes in these states would cast their ballot for Obama and 44% for Romney; 6% remain undecided. It also found that the undecideds favor congressional candidates who support “standards to reduce toxic mercury pollution from power plants” over those who oppose them (59% to 23%).

The presidential candidates have largely avoided the controversial topic of climate change. Yet it did come up briefly at the recent political conventions.

Romney, who as a presidential candidates has expressed doubts about the causes of climate change and has called for a broad expansion of fossil fuel drilling, mocked Obama’s 2008 promise to slow the rise of the oceans and heal the planet. “My promise…is to help you and your family,” Romney said, drawing applause.

Obama, who’s embraced an “all of the above” energy strategy that calls for renewable energy, nuclear power and limited oil and gas drilling, countered at the Democratic convention in September. “Climate change is not a hoax,” he told delegates, also eliciting cheers. “More droughts and floods and wildfires are not a joke. They are a threat to our children’s future.”

U.S. small-business borrowing rises in August

(Reuters) – Lending to small U.S. businesses rose in August for a second straight month, a report showed on Monday, a sign the economy may have been regaining its footing even before the U.S. central bank’s latest round of stimulus.

The Thomson Reuters/PayNet Small Business Lending Index, which measures the overall volume of financing to small U.S. companies, rose to 109.9 from July’s upwardly revised 106.7, PayNet said. It was the highest level this year.

PayNet had initially reported the July figure as 103.8.

Borrowing was up 10 percent from a year earlier.

“We’ve got a winning streak going,” PayNet founder Bill Phelan said.

“These business owners are looking out three, to six, to nine months, and they are seeing some positive profit-producing opportunities: there must be something these business owners are seeing.”

The Federal Reserve last month unleashed a new round of bond-buying to lower borrowing costs and spur businesses to spend and, eventually, to hire.

PayNet’s lending index typically correlates to economic growth a quarter or two in the future.

Separate PayNet data showed companies were under less financial stress. Accounts overdue by 30 days fell to a new record low – 1.16 percent of the total – from 1.18 percent the previous month.

Longer-term delinquency rates also eased. Accounts behind 90 days or more, or in severe delinquency, dipped to 0.25 percent from 0.27 percent.

Accounts behind 180 days or more, which are considered in default and unlikely to be paid, fell to 0.32 percent from 0.34 percent.

PayNet collects real-time loan information, such as originations and delinquencies, from more than 250 leading U.S. lenders.

(Reporting by Ann Saphir; Editing by Dale Hudson)

Want a job? Look to the energy field

First of five reports this week on the job outlook in key industries.

8:49AM EST October 1. 2012 – In 2009, Andrea Conaway was an X-ray technician whose career plan was showing a few fractures.

Hospitals were consolidating around Pittsburgh, so she went searching for oil, in a sense. She went back to school, got an associate’s degree in computer electronics, and in August, began a job as an associate systems analyst at EQT, a Pittsburgh-based natural gas driller.

“I figured this industry was the future,” says Conaway, who says she boosted her income to almost $50,000 a year from $38,000. “It’s growing like crazy around here. My friends were getting raises and great benefits, and I knew I wouldn’t be.”

GRAPHIC: Where the jobs are

Of all the places that America’s new jobs are, the emerging energy business, directly or indirectly, might be responsible for more of them than almost anything else.

Since 2002, the exploration of natural gas deposits embedded in shale, followed by oil drilling that began in earnest late in the decade, has created more than 1 million jobs, says Moody’s Analytics economist Chris Lafakis. That’s out of 2.7 million the whole country created.

“It’s really huge,” Lafakis says. “And the jobs pay very well.”

Jobs directly in the oil and gas extraction business pay an average of just under $150,000 a year, Lafakis says — almost exactly three times the national average.

Big jobs in small towns

Just counting positions directly in the energy industry, the shale boom has accounted for as many as 33,000 new U.S. jobs this year, according to Bright Labs, a San Francisco start-up whose website provides job-hunt data and tips.

More than 3,500 are in metropolitan Houston, Bright says. But the job expansion stretches through cities of all sizes. Oklahoma City’s 400 jobs are near the top of the list, Bright says. Denver, Pittsburgh, and Williston, N.D. — all near newly exploitable oil and gas deposits — are also seeing big changes from shale for shale-related jobs.

The most plentiful jobs — at least, of those that end up being advertised online — seem to be in engineering. Six of the top eight most-filled new jobs in the industry are for some kind of engineer, says Bright senior data scientist Jacob Bollinger.

Broad-based boom

But the shale job surge is more broadbased than those numbers alone would suggest, thanks to the energy industry’s complicated infrastructure and supply chain, researchers say. While oil and gas deposits are concentrated in a handful of places, more than 30 states saw oil and gas support employment, including suppliers and service companies that work with energy companies, rise at least 50% in the past decade, Moody’s says.

Because new rigs have to be built, and oil and gas have to be moved to market either via newly built pipelines or by truck and rail, new jobs abound at both drilling companies and their suppliers, says Tom Tunstall, research director for the Institute for Economic Development at the University of Texas San Antonio. It has closely studied the development of the Eagle Ford shale field that spans more than a dozen counties in South Texas.

In Pennsylvania, where officials say shale added 18,000 new energy industry jobs between 2008 and last year, another 5,000 jobs were added for freight trucking, and 500 more were created to build roads, according to a state-sponsored study this summer.

One company that’s gotten a big boost is the Union Pacific railroad, which now gets about 2% of its business from shale-based companies, hauling everything from tank cars of oil to sand used in the process of extracting gas and oil from deep underground, CEO John Koraleski says.

Rising U.S. energy production also is creating manufacturing jobs for everything from steel for piping to rail cars. Among the biggest: a $650 million, 350-job plant by French tubing manufacturer Vallourec to make products for shale drilling in Youngstown, Ohio, and a planned natural gas cracking plant outside Pittsburgh that could add 10,000 construction jobs and hundreds of permanent positions at Shell, if it goes forward.

“Every week, I see a billion-dollar investment,” Navigant Consulting energy economist Julie Carey says.

Teachers and vets may apply

Pioneer Natural Resources, an Irving, Texas-based driller, has hired 400 people in the South Texas area in the last two years, says Joey Hall, vice president in charge of Pioneer’s South Texas Asset Team.

About three-quarters of those are blue-collar workers in the fields, he said. But the company has retained at least another 1,000 outside contractors to build rigs, drive trucks and do other work.

On top of that, the company added scientists and engineers back at headquarters to support both Eagle Ford and other operations, he said. In all, the company added 850 staffers last year and another 500 so far this year — to a team of about 4,000.

“These are good, decent, honorable jobs,” Hall said. “People can and do leave satisfied and provide for their families.”

The picture is similar at ConocoPhillips, the nation’s biggest independent exploration company. Conoco added more than 500 jobs in Texas and North Dakota related to shale exploration in the last three years, spokeswoman Davy Kong said.

More than half were professional staff such as engineers, and the rest were field staff, such as mechanics and construction workers.

About 15% of the 54,000 new jobs expected in the Eagle Ford shale by 2021 will require a college degree, and a little more than 10% will require direct experience in the energy business, Tunstall said.

Community colleges and business groups are organizing training programs for those who need them. A federally funded $15 million program called ShaleNet plans to train pipeline operators and oil-field technicians nationwide, who are expected to follow work to new places as new fields open, said Laura Fisher, senior vice president for the Allegheny Conference, a business group in Pittsburgh.

At Pioneer, field hands and supervisors include everyone from military veterans to former schoolteachers, Hall said. Pipeline technician James Laake took about a $20,000 raise from his old job as a prison guard; the 24-year-old high school graduate, who has a welding certificate, said he needed only brief on-the job training.

“The transferable skills for veterans are leadership, logistics and working with large equipment,” Hall said. “Electronics are electronics, and a truck is a truck.”

Contractors and spinoffs

Because jobs related to shale are in more places, more job categories and more industries than you might suspect, figuring out which jobs are really caused by the shale boom is a bit tricky.

Only about a quarter of employment growth driven by the shale boom has happened at energy companies, Moody’s found. Another 308,000 have been “indirectly” created by the boom — those include the ones at transportation companies or equipment firms, Lafakis said. More than half a million are “induced” jobs — created to serve the needs of the other half-million people, once oil and gas brings more wealth to their towns.

The boom is causing everyone from law firms to banks to add white-collar staffers, as well — and to open everything from hotels to the Red Dog Ice House, a big- screen sports bar coming to Carrizo Springs, Texas.

At Huntington Bancshares in Columbus, Ohio, executives give shale drilling in Ohio, Pennsylvania and West Virginia much of the credit for the bank’s move to No. 3 among Small Business Administration lenders, from No. 15 three years ago, says CEO Stephen Steinour. The company has added 500 bankers, from loan officers to private bankers serving landowners whose petroleum-rich property turned out to be more valuable than they ever thought.

“It’s not all about shale, but shale is what’s driving the confidence to keep investing,” says Jim Dunlap, Huntington’s director of regional and commercial banking.

Hard to predict

Because the industry is still new, predicting its future growth is difficult, Lafakis said. Moody’s is projecting much less growth in the next decade than in the last one, but the firm’s analysts think it’s more likely than not those estimates will prove conservative, he added.

The big question mark is the price of oil and natural gas. Hiring for gas fields has slowed as drillers have pulled back in response to lower gas prices. Companies are moving money to developing shale oil instead.

Most agree that large-scale oil hiring has at least another year or two to run, before the infrastructure of the new fields is put into place, Hall says. Operating rigs will require less labor, and job growth will moderate, he adds.

Then keeping shale’s job boom going will depend on how many manufacturers in chemicals, autos and other energy-intensive fields will boost hiring by growing in the U.S., rather than abroad, thanks to energy oil prices lower than the world average, says George Schink, energy economist at Navigant Economics.

“We’ve done most of our hiring,” Pioneer’s Hall says. “But we’ll do some for another couple of years.”