Making sense of e-mail madness.
Not all e-mail is created equal. Some messages may be relevant for years, while others lose meaning within minutes. Yet e-mail in-boxes treat all messages alike, regardless of who wrote them, what they’re about, or when they were sent. Adam Smith has set out to change that with Xobni, software that pulls useful information out of e-mails and contextualizes it according to sender.
Smith’s goal is to help people unlock and harness the social relationships embodied in their in-boxes. The first version of Xobni (inbox spelled backwards) is a plug-in for Microsoft Outlook and works only on Windows computers–but the results are remarkable.
Once it’s installed, Xobni scans every e-mail and extracts information such as a sender’s phone numbers, what time she is most likely to e-mail you, who else she has corresponded with, and what files the two of you have exchanged. It labels all the data with descriptive tags, which it then indexes and analyzes. When you click on a specific e-mail, it displays all the information relevant to that sender in a sidebar that runs down the right side of the Outlook window. The tags also allow Xobni to search all indexed e-mails very rapidly.
Smith and his friend Matt Brezina founded Xobni in San Francisco two years ago and have raised $4.25 million from companies including First Round Capital and Khosla Ventures. The plug-in has received one startling endorsement: Bill Gates used it in a public demo at a Microsoft conference, even though the free download remains in beta. There have been rumors of Microsoft’s offering around $20 million for Xobni, but the young cofounders didn’t bite. Instead, they intend to offer a “Pro” version and à la carte features for sale. Xobni also plans to extend its reach to other e-mail programs, including Web-based services such as Yahoo Mail and Gmail. And the team has already begun building in access to social networks such as LinkedIn.
Other companies have tried to streamline e-mail before. But if Xobni can reach a significant fraction of the world’s 400 million Outlook users, Smith may save people time and annoyance by making e-mail more useful.
Blaise Agüera y Arcas
Building immersive 3-D environments.
Imagine taking hundreds of photos in the Rockies and being able to piece the images into a virtual re-creation of the peaks. With Microsoft’s Photosynth, you can. Created by Blaise Agüera y Arcas, the software uses digital photos to construct 3-D environments called “synths.” Agüera y Arcas created the first version in 2006, drawing on Seadragon–a data navigation technology he’d developed previously–and computer-vision research from Microsoft and the University of Washington.
In August, Agüera y Arcas and his team released a version of Photosynth that allows users to construct their own synths for the first time. The software runs on users’ computers and includes algorithms that let them more easily pivot in 3-D space. It also allows them to post their synths online and discover other synths of the same or similar places. As users add synths of cities, stores, and homes, Agüera y Arcas says, the Photosynth site will be able to “enrich online 3-D mapping, shopping, real estate, and other immersive Web applications that involve real objects and places.”
Simple, flexible Web publishing.
The Internet has made publishing on a global scale almost effortless. That’s the rhetoric, anyway. The truth is more complicated, because the Internet provides only a means of distribution; a would-be publisher still needs a publishing tool. A decade ago, people who wanted such a tool had three choices, all bad: a cheap but inflexible system, a versatile but expensive one, or one written from scratch. What was needed was something in the middle, requiring neither enormous expense nor months of development–not a single application, but a platform for creating custom publishing environments. For tens of thousands of sites and millions of users, that something is Drupal.
Created as an open-source project by Dries Buytaert, Drupal is a free content management framework–a tool for building customized websites quickly and easily, without sacrificing features or stability. Site owners can choose from a list of possible features: they might, say, want to publish articles, offer each user a profile and a blog, or allow users to vote or comment on content. All these features are optional, and most are independent of the others.
With Drupal’s high degree of individualization, users can escape cookie-cutter tools without investing in completely custom-made creations, which can be time-consuming, costly, and hard to maintain. The Howard Dean presidential campaign used Drupal in 2004, and today it’s used by Greenpeace U.K., the humor magazine the Onion, Nike’s Beijing Olympics site, and MTV U.K., among many others.
The diversity of its users has led to many improvements, Buytaert says: “The size, passion, and velocity of the Drupal community makes incredible things happen.” There are tens of thousands of active Drupal installations worldwide. Thousands of developers have contributed to the system’s core, and more than 2,000 plug-ins have been added by outside contributors.
Buytaert began the work that became Drupal in 2000, when he was an undergraduate at the University of Antwerp. He had a news site called Drop.org, and he needed an internal message board to host discussions. After reviewing the existing options for flexible message boards, Buytaert decided he could write a better version from scratch.
The original version of Drupal (its name derives from the Dutch for droplet) worked well enough to attract additional users, who proposed new features. Within a year, Buytaert decided to make the project open source. He released the code in January 2001 as version 1.0.
Since open-source projects tend to attract expert users, they often lack clear user interfaces and readable documentation, making them unfriendly to mere mortals. But Buytaert understood from the beginning how important usability is to the cycle of improvement, adoption, and more improvement that drives the development of open-source software. The core Drupal installation comes with voluminous help files. The central team regularly polls users as well as developers (which is unusual in an open-source project) to decide what to improve next. The process reveals not just features to add, but ones to remove, and ways to make existing features easier to understand. For example, the project’s website has been redesigned to help people new to Drupal figure out how to get up and running.
Buytaert has also founded a company, Acquia, to offer support, service, and custom development for Drupal users, especially businesses. He calls Acquia “my other full-time job” and likens it to Linux distributor Red Hat, which provides custom packaging and support for its version of the open-source operating system.
With Drupal version 7, due later this year, Buytaert hopes to include technologies that will make sites running Drupal part of the Semantic Web, Tim Berners-Lee’s vision for making online data understandable to machines as well as people. If Drupal hosts a website containing a company’s Securities and Exchange Commission profile, for example, other sites could access just the third-quarter revenues, without having to retrieve the whole profile. The goal of sharing data in smaller, better-defined chunks is to make Drupal a key part of the growing ecosystem of websites that share structured data. If this effort succeeds, it will ensure Drupal’s continued relevance to the still-developing Web.
Gaming with the flow.
Jenova Chen has been playing video games for 20 years, and he’s desperate to see something new: right now, he says, most games focus on stimulating players by inciting aggression. “I want to expand what a video game can be,” he says. So as a graduate student in interactive media at the University of Southern California, Chen looked to psychologist Mihály Csikszentmihályi’s theory of “flow,” which identifies a state of focus that people find enjoyable and fulfilling. Chen uses the theory’s principles to design games that offer just enough challenge–not so little that players become bored, not so much that they become anxious.
Chen’s first effort was FlOw, a Web-based “Zen game” in which players control a sea creature that swims, eats, and evolves. After graduating in 2006, Chen cofounded Thatgamecompany to continue his work. The company released a PlayStation 3 version of FlOw in 2007; it has become one of the most downloaded games on the PlayStation Network. The next game, Flower, will be released later this year. By going with the flow, Chen may help video games reach a whole new audience.
Inferring social networks automatically.
Social-networking sites such as Facebook require users to find and confirm connections with other people. But what if your cell phone could automatically identify the people you know, and even sort them into categories?
If that capability arrives, it will be thanks to reality mining, a field that Tanzeem Choudhury pioneered as a PhD student at the MIT Media Lab. Working at Intel after graduation, she created a pager-size sensor pack–loaded with software plus microphones, accelerometers, and other data-gathering devices–to collect and analyze data about human interactions and activity. For instance, by processing verbal utterances, she can identify the most influential people in a social network.
Now an assistant professor of computer science at Dartmouth, Choudhury is conducting experiments with the sensor-laden iPhone. Within a few years, she says, simple versions of her software could be available for cell phones.
Personal updates made simple.
In 2006, Jack Dorsey created Twitter so that he could let friends and family know what he was doing, wherever he–or they–might be. Today more than two million people use it to send out 140-character-or-fewer updates, called “tweets,” through Twitter’s website or by text message over mobile devices. Dorsey’s ethos of simplicity shapes everything about Twitter, from the application itself to the company’s San Francisco offices (see “Home Tweet Home,” July/August 2008). Twitter’s popularity has given rise to an entire ecosystem of applications. Yet Dorsey, cofounder and now CEO of the bemusing microblogging service, is secretive about how Twitter will ever make money; critics say that’s because its executives have no idea. What’s not a secret is that Twitter has had difficulties supporting its growing band of obsessives: in recent months, twitterers have been frequently confronted by error screens bearing messages such as “Twitter is stressing out a bit right now.”
Jason Pontin, TR’s editor in chief, recently chatted with Dorsey about these and other issues using Twitter’s @reply function, which directs a public message to a particular user.
jason_pontin @jack Explain Twitter.
jack @jason_pontin Twitter is a real-time repository of state for people, events, & things. A personal news wire of sorts.
jason_pontin @jack I twitter every day. But whenever I explain it to people who’ve not, they are uncomprehending or angry. Why?
jack @jason_pontin People have to discover value for themselves. Especially w/ something as simple & subtle as Twitter. It’s what you make of it.
jason_pontin @jack Critics say that tweets are trivial. Is that missing the point?
jack @jason_pontin It depends on the context the recipient brings. There’s a universe in the smallest, most “trivial” details of one’s life.
jason_pontin @jack Even people who love Twitter are frustrated by the service. It’s broken far too often to feel reliable.
jack @jason_pontin We love what we’re building & we hate to see it suffer. Our goal is to make it reliable enough to be trusted as a public good.
jason_pontin @jack Twitter also seems to lack basic stuff. I can’t organize my followers intelligently. Or search very well. When will Twitter grow up?
jack @jason_pontin Unfortunately, we’ve neglected the user experience to focus on stability of the foundation. We have designs to put this right.
jason_pontin @jack You recently got $15 million from Spark Capital and Bezos Expeditions. Will you buy some servers and infrastructure with that?
jack @jason_pontin I can’t confirm the number, but I can confirm we’ll make the money work for our users (20 of whom happen to be our investors)!
jason_pontin @jack What’s Twitter’s business model?
jack @jason_pontin We’re building what we love. While we have many ideas for sustainable revenue, Twitter’s will emerge naturally from our work.
jason_pontin @jack Sometimes it sounds like your monetization plan is: let’s get acquired by a communications company.
jack @jason_pontin We’re not focused on answering that question. We’re determined to build a solid platform and service we can take all the way.
Stefanus Du Toit
Programming for parallel processors.
PROBLEM: As the ever-shrinking computer chip begins to run into fundamental physical limits, designers have begun building multiple processor “cores” onto each chip to improve performance. But writing software that can run in parallel on multiple cores is complicated and time consuming, and few programmers have the expertise to do it. As a result, most of the capacity on a multicore chip goes to waste.
Solution: Stefanus Du Toit has created software that makes it easier to translate traditional serial programs into parallel programs. He began its development as a graduate student at the University of Waterloo, in Ontario; in 2004 he cofounded RapidMind, in Waterloo, to commercialize it. The company has raised $10 million and partners with Advanced Micro Devices, Hewlett-Packard, IBM, and others.
With RapidMind’s technology, programmers write software in C++ as usual; then they use a special interface to specify which parts of the program should be parallelized. The platform automatically parcels out those tasks among the cores. It builds code into the final program that manages workload, ensuring that each core is fully utilized and preventing errors such as one task’s stalling while it waits for another to finish. Finally, the platform optimizes the program to run on a particular chip–say, an eight-core chip from Intel. The finished program runs more efficiently; in one example, an image-processing application rewritten with the RapidMind platform ran 10 times as quickly on eight cores as on a single processor.
Deconstructing software to find bugs.
PROBLEM: Programmers, despite their best efforts, make errors, any one of which could cause a system to crash or admit an attacker. Although automated test programs have improved software, major bugs still slip through, costing businesses and governments billions of dollars each year.
SOLUTION: As a graduate student at Stanford, Seth Hallem perfected an improved approach to finding bugs, called static analysis. Where ordinary test software runs a program and hopes to stumble on errors, static analysis breaks it into pieces that perform discrete functions, such as “add the results of lines 42 to 47.” The computer determines what each piece does and then simulates how various functions might interact, looking for problematic combinations.
Previous attempts at static analysis were either too simplistic to find important bugs or too comprehensive to ever finish the job. Hallem developed algorithms to weed out redundant analysis and examine only the most important combinations, allowing millions of lines of code to be examined quickly and effectively. He cofounded Coverity in San Francisco to apply the technology commercially. More than 450 customers, including Raytheon and Yahoo, use Coverity’s tools to vet their software.
Enhancing video search.
The amount of video on the Web is growing at an incredible rate. Effectively searching online video, however, remains difficult. Microsoft researcher Xian-Sheng Hua hopes to crack the problem by teaching computers to recognize objects, scenes, events, and other elements of digital images.
Hua uses machine-learning techniques and annotated videos to train computers to automatically categorize new videos. While this general approach isn’t new, Hua’s system permits multiple labels for each video segment–and relies not only on specified tags applied by experts but also on descriptions written by large numbers of grassroots Internet users. These user-generated tags are gathered by means of online games, “pay for labeling” schemes, analysis of how people search for video, or other methods. Hua applies some automated filters to the labels to ensure their quality.
The system, which runs online, is first trained on videos tagged by experts; it’s then periodically updated and retrained using the grassroots labels. This “online active learning” makes the algorithm more accurate and several times faster than previous systems; applying multiple labels to each video increases the speed further. The technology should aid searches for still images, too. Some of the techniques involved are already being incorporated into Microsoft’s Live Search Video. Ultimately, Hua says, the technology should improve not only online video and image searches but also video surveillance and digital media management.
Making memory at Internet speed.
PROBLEM: At the heart of the Internet are the routers that direct packets of data to their destinations. But by briefly holding each packet in memory while figuring out where to send it, these specialized computers create a bottleneck. The speed of today’s 10-gigabit-per-second links forces router makers to use fast but expensive static random-access memory (SRAM) instead of slower, cheaper digital random-access memory (DRAM). As connection speeds increase, the amount of SRAM needed will become prohibitively expensive, leading to data loss and limiting applications such as voice calls and videoconferencing.
SOLUTION: As a graduate student at Stanford, Sundar Iyer created a technique that lets equipment makers combine SRAM with DRAM to make routers at once faster, more reliable, less expensive, and more energy efficient. In Iyer’s “perfect caching” scheme, each arriving data packet is stored in an SRAM chip. Once every hundred nanoseconds, the cache sends all the packets to the main memory, made from DRAM. Fifty nanoseconds later, another SRAM cache takes only the packets it needs and sends them to their destinations. Iyer founded Nemo Systems to develop the technology in 2003; Cisco bought Nemo in 2005 and is building the system into its next generation of enterprise routers.
Locking microchips to prevent piracy.
PROBLEM: High-tech piracy isn’t limited to illegal downloads and knockoff DVDs: there are growing, multibillion-dollar gray and black markets for the microchips that run everything from video players to high-end weapons. Unscrupulous employees in overseas foundries that produce chips for other companies can divert extra chips, made for pennies, and resell them.
SOLUTION: Farinaz Koushanfar, an assistant professor of electrical and computer engineering, has developed a way to foil hardware pirates using tiny physical variations between circuit elements on a chip–variations produced normally in the chip-manufacturing process. As small as a stray atom or two, the variations cause identical signals traveling to two such elements to arrive a few trillionths of a second apart; each chip contains hundreds of these pairs. For each pair, Koushanfar designates the first signal to arrive as a 0 and the second as a 1, creating an ID code unique to each chip. When a buyer first uses the chip, it transmits its ID to its designer over the Internet. The designer sends back a corresponding “unlock” code that makes the chip usable. Koushanfar has created prototypes of the coded chips, and several chip makers have expressed interest in the technology.
Streamlining human-computer interactions.
When the Nintendo Wii came out, most people saw a fun new way to play video games. Johnny Lee saw a surprisingly good infrared camera that could make innovative computer interfaces affordable. At the 2008 Technology, Entertainment, Design (TED) conference, he drew spontaneous applause when he demonstrated two devices he’d hacked together, which used the $40 Wii remote and some inexpensive hardware to simulate systems that can cost thousands. The audience may not have realized that Lee had spent no more than “a few days” on each. “I have some knack,” he says, “for being able to identify easy projects that have a relatively big impact”–like those at right. Having completed his PhD at Carnegie Mellon, Lee is honing that knack as a researcher in Microsoft’s hardware division. Read why Lee thinks researchers should focus on bringing technologies to all.
Meredith Ringel Morris
Searching websites jointly.
“I’m not really interested in technology for the sake of technology. I’m interested in how it helps people connect and work with other people,” says Meredith Ringel Morris, a computer scientist in the Adaptive Systems and Interaction Group at Microsoft Research. Her tool SearchTogether, shown below, is a plug-in for Internet Explorer that makes it easy for groups to share the work of searching without duplicating each other’s labor. Bookmarked websites appear in a frame beside the main browser window, along with users’ comments and ratings. A chat window at the bottom of the screen lets users discuss results in real time if they’re online simultaneously. Morris says that collaborative search combines the two activities she thinks people are most interested in doing online: communicating and gathering information. She’s also working on a tool that will help groups search collaboratively when sharing one computer, which could be particularly useful in classrooms.
Building household robots.
Housekeeping robots are still the stuff of science fiction, but not for want of hardware: there’s almost no task too precise or delicate for a robot that knows in advance what it’s supposed to do. The problem lies in teaching robots to deal with the unknown. That’s precisely what Andrew Ng, an assistant professor of computer science, set out to do when he founded the Stanford Artificial Intelligence Robot (STAIR) project a few years ago.
Previous robots have had some ability to improvise–many could locate familiar objects in unfamiliar environments, for example. But Ng has gone a step further: STAIR can deduce how to pick up an object it’s never seen before. Using traditional machine-learning techniques, Ng trained STAIR on a database of pictures of objects such as wine glasses, coffee mugs, and pencils, as seen from different perspectives. Each object was correlated with information about the best place to grasp it: the stem of the wine glass, the middle of the pencil. After its training, STAIR could generalize those associations to adapt to new situations–lifting, among other things, a lunch box by its handle and a piece of intricate lab equipment by its metal stem. It was even able to remove dishes from a dishwasher and place them on a drying rack.
The STAIR team has made other advances–its innovative system for robotic depth perception even spawned a side project, software that converts static 2-D photographs into 3-D images. But despite this progress, Ng knows that building a general-purpose household robot is beyond the means of any one lab. So he’s developing an open-source robotics operating system that will let researchers integrate a robot’s sensor systems and functional components in new ways, without having to write code from scratch.
Engineering electric sports cars.
As he pulls away from the headquarters of Tesla Motors in San Carlos, CA, JB Straubel apologizes for the condition of the car. The outside looks fine, a gleaming orange. But inside, instruments dangle from the dashboard. A message scrawled on blue masking tape warns that the passenger’s-side air bag is disabled. A bell chimes mysteriously. The car had been shipped to England and subjected to vibration tests designed to “shake it apart and kill it,” Straubel says. Now it’s an engineering car–one Straubel, the company’s chief technology officer, feels comfortable drilling holes in and bolting prototype hardware to. “It’s pretty much already written off,” he says. “But it’s also the fastest car in our fleet at the moment.”
He punctuates the sentence by hitting the accelerator. Straubel looks remarkably calm as the car surges forward, pressing him into the seat. From a dead stop at the on-ramp, it takes just a few seconds to overtake the vehicles on California’s Highway 101. In sports cars, this kind of acceleration is ordinarily accompanied by rapid-fire shifting, but Straubel never takes his hands off the steering wheel. Powered by batteries and an electric motor, the Tesla Roadster isn’t bound by the limits of old-fashioned gas-burning engines. At its top speed of over 120 miles per hour, it remains in its first and only gear.
Straubel doesn’t come close to 120 miles per hour today. Since the car can accelerate to 60 miles per hour from a stop in just under four seconds, “you get caught up to traffic pretty fast,” he says, easing off the accelerator. “It kind of spoils you.” It’s easy to see why this powerful alternative to gas-guzzling internal-combustion engines (see Hack, “Tesla Roadster”, September/October 2008) has generated such remarkable excitement.
Straubel, more than anyone else, is responsible for the car’s impressive acceleration. The Roadster is the first production model from Tesla, which was founded to mass-produce high-performance electric cars. The car’s carbon-fiber exterior and aluminum frame, which make it visually appealing but keep it light, are based on designs from British automaker Lotus. Straubel and his hand-picked team, however, engineered the car’s brains, muscles, and guts–the electronic controls, electric motor, and battery pack that enable the Roadster to beat many of even the quickest gas-powered cars off the starting line.
Electric cars are best known for their environmental benefits: they produce no harmful emissions, and they’re so efficient that they reduce total carbon emissions even if the electricity used to recharge them comes from power plants that burn fossil fuels. But Straubel’s achievements capitalize on another, less appreciated advantage. Gas engines deliver their peak torque–the key to acceleration–only within a limited range of engine speeds. Keeping the engine in its optimal range requires a convoluted system of gears and clutches, and acceleration is still compromised. Electric motors, however, deliver maximum torque from a standstill up through thousands of revolutions per minute. That makes it possible to use a transmission with just one or two speeds–and it makes electric cars more responsive than gas-powered ones. Yet most electric vehicles haven’t reaped the full benefit of their torque advantage, says Marc Tarpenning, one of Tesla’s founders. That’s because they have typically been underpowered, partly in an effort to make them as inexpensive as possible. Straubel set out to change that.
During his early days at Tesla, the company licensed a number of technologies from AC Propulsion, a small company that had pieced together a prototype electric car with acceleration similar to the Roadster’s. Tesla’s founders decided to use AC Propulsion’s parts to produce their own prototype. But those parts were “ruinously expensive,” Tarpenning says, “and no two were alike.” Straubel has since reëngineered almost every one of them.
It was soon clear that the extreme torque provided by electric motors can be a problem, especially in a high-powered car. Without a well-tuned motor controller, the torque can jerk the driver around, says Andrew Baglino, one of the engineers Straubel hired. What’s more, the complex interplay between the driver’s application of the accelerator, the conditions of the road, and the electronic characteristics of the battery and motor can have unexpected consequences. AC Propulsion’s controller was “a hokey analog system–messy circuitry that was 20 years old,” Straubel says. As he and his team worked to develop a production-ready car, they found that one controller would work well while another would inexplicably fail. “We’d debug it for weeks trying to figure out what the hell was different, and we never could,” Straubel says. The unreliable controllers would sometimes cause the motor to jitter. Worse, at times all power would cut out–once, as the car was hurtling down the highway.
Straubel reasoned that a digital control system would solve these problems. Switching to digital would require starting from scratch, but he was sure the new system would both improve performance and speed development. Yet the decision was made to stick with the analog system, in the hope that its kinks could be worked out.
Undeterred, Straubel put Baglino to work on what appeared to be a side project: designing test equipment that put the company’s motors and batteries through the paces of simulated driving cycles. This equipment was to have digital controls, which Straubel intended to translate into a digital controller for the car.
Meanwhile, the engineers continued to painstakingly debug the analog system. “It felt silly to be solving problems that we knew we were trying to make obsolete,” Straubel says.
After months of working on the digital test equipment, the engineers had learned enough to design a prototype digital controller. It worked, and soon the messy analog system was gone. The jittering and jerking gave way to a digitally controlled, reliably smooth ride–and a car that was, incidentally, far more responsive.
The Roadster’s exceptional motor, too, is a tribute to Straubel’s persistence. Tesla initially used a third-party transmission that included two gears–one to accelerate from a stop and the other to reach high speeds. The system gave the Roadster a top speed of more than 120 miles an hour. However, the shifting system routinely wore out after just a couple of thousand miles. So Straubel found a way to replace it with a single-speed gearbox. Early on, Straubel and his team had redesigned the patterned metal plates and wire coils at the heart of electric motors to improve both efficiency and torque. But the electronics feeding power from the battery to the motor still limited its output. To exploit the added torque, Straubel added higher-performance transistors and retooled the electrical connections between the motor and the gearbox. These changes increased the torque that the motor could deliver at low speeds and allowed the engineers to use a single-speed transmission without sacrificing either acceleration or maximum speed.
But Straubel’s most notable contribution may have been to keep the car from bursting into flames. Tesla’s founders decided from the start to power the car with lightweight lithium-ion batteries of the type used in laptops, and they knew they had their work cut out for them. If lithium-ion cells are pierced, crushed, overcharged, or overheated, they can combust. The challenge was even greater because the individual cells were small: it would take 6,831 of them to give the car a decent range. All those cells would have to be wired together into an ensemble that was durable but allowed the charging and temperature of each cell to be carefully controlled.
This was fine with Straubel, who had been building electric vehicles since before he was old enough to drive and had long wanted to make a laptop-battery-powered car. Under his direction, all those goals were reached. But along the way, the team discovered that in some (extremely rare) cases, manufacturing defects within a cell could cause it to heat up and catch fire without any outside cause. (This problem led to the recall of millions of laptop batteries in 2006.) Using computer models, Straubel found that if any one of the 6,831 cells caught fire, it could set off its neighbors, starting a chain reaction that could destroy the battery pack and turn the car into a smoldering wreck. Tarpenning asked at the time, “So, JB, what’s going to happen to our energy storage system?”
As it turned out, the solution was already at hand, largely because of an argument Straubel had won early in the development of the battery pack. The car’s initial design called for air cooling to control the temperature of the batteries and extend their lifetime. But Straubel quickly realized that that approach wouldn’t provide the necessary control.
“We had a lot of heated discussions about what direction we should go,” Straubel says. But his cool-headed logic, along with some hard figures, won the day. The resulting liquid cooling system–a network of tubes running past almost every cell in the pack–also offered a solution to the problem of the spontaneously combusting cell. With slight improvements, the system was able to evacuate the heat from a flaming cell so quickly that it couldn’t set off its neighbors. As with the digital controller, Straubel had been able to find a solution, even if it meant going against the grain.
Tesla began shipping Roadsters this year; the first four were delivered by June. Richard Chen, a former Google product manager who hopes to have his car by Christmas, mailed in a $100,000 check long before the production car existed, and before the company had even announced a price. His excitement is not unique: the car, which has a base price of $109,000, is back-ordered for at least a year.
Its success may have an impact well beyond Tesla’s bottom line. Bob Lutz, GM’s vice chairman, was quoted in Newsweek as saying that the Roadster was a deciding factor in GM’s decision to return to electric cars after abandoning them several years ago. If a Silicon Valley startup can do it, he reasoned, why can’t GM? What’s more, the Roadster may be changing the image of electric cars and increasing their chances for success. People such as Chen, who got to test-drive the car before finalizing his purchase, are buying it not to save the planet (though the green credentials are a nice side benefit, Chen says) but simply because it’s so much fun to drive.
These days, Straubel is focusing on improving the Roadster and engineering a sedan to open up a new, wider market for the company. And tentative plans are in the works for a small car, such as an electric version of Daimler’s tiny, inexpensive Smart car.
All that means long days for Straubel, and part of what keeps him going is the belief that he’s doing something important: finding a way to deal with the world’s energy woes. But he seems most driven by pure enjoyment. That’s clear enough when he’s behind the wheel of the latest version of the Roadster, whose new electronics can deliver far more power than the first version had. “It’s amazing what a few hundred more amps can do,” he says, laughing, after a burst of acceleration. “It’s fun, huh?”
Putting DIY projects online.
When Eric Wilhelm finished his PhD in mechanical engineering at MIT, he and three friends started Squid Labs, a consultancy based in Emeryville, CA, that finds fixes for clients’ technical problems (how to make solar-collecting concrete, for example). But Squid Labs was also founded as a place where the colleagues could explore their own projects and ideas–funding them through their consulting jobs. In 2005, Wilhelm had an idea for a how-to website where people could share step-by-step visual instructions for original projects. The team spun the site out into its own company, and Instructables was born.
Wilhelm had hit upon the idea at the right time, just as the spirit behind open-source software began permeating other technological fields. Instructables offers its growing community of more than 300,000 registered users an easy way to document do-it-yourself technology projects and share ideas with others.
Building robotic flies.
Robotic flies equipped with cameras, microphones, and other sensors would be a spy’s dream. But researchers have had trouble creating the materials needed to make robots that look and behave like real insects. Robert Wood, an assistant professor of engineering and applied sciences, took on the challenge: he developed a revolutionary fabrication technique that allows engineers to make a range of very tiny parts for any kind of robot.
Wood’s technique bears similarities to origami. To create three-dimensional structures that bend and rotate precisely as needed–not only for flying robots, but also for crawlers and swimmers–Wood builds “fold lines” into layered composites of materials such as polymers or carbon fiber. Last year he used the method to build the world’s smallest robotic insect capable of taking off. It is powered and controlled externally, but he plans to develop an onboard power source and sensors, and to refine the robot’s control systems. Wood’s ultimate goal is a fully autonomous robotic insect.