Volume 7, Issue 3
http://www.coe.berkeley.edu/labnotes/0607
Commuter-friendly version


A Marketplace for Peer-to-Peer Charity
by David Pescovitz

After hurricane Katrina ravaged the Gulf Coast, Berkeley Engineering graduate students Ephrat Bitton and Anand Kulkarni watched with the rest of the world as logistical snafus, bureaucratic red tape and communication breakdowns prevented charitable aid from quickly reaching the storm’s victims. There was a disconnect between those who had something to offer and those who needed it.

Since then, the two students in the Department of Industrial Engineering and Operations Research have spent their free time developing a Web application to help ensure that such a disconnect would never happen again. Their system automatically pairs donors with those in need, creating a "marketplace of charity" while putting a human face on the process of giving.

"During Hurricane Katrina, there were a lot of people eager to donate, but they didn't feel they had a good way to do that," says Bitton, a graduate student in Professor Ken Goldberg's laboratory. "On the other hand, we noticed an interesting trend where people would offer donations of goods or services on Web forums and community sites like craigslist."

As Web access was restored to Katrina victims, these bottom-up, peer-to-peer approaches to match goods and services with victims worked surprisingly well, even though the websites were not designed for this purpose. But what if they were?

That's the idea behind iCare, Bitton and Kulkarni's Web application that acts as a virtual middleman to link donors and victims. After a disaster, victims can log onto the website and report their specific needs, and those requests will be connected to donors—companies or individuals—who are offering that particular kind of aid. The researchers also hope that this demand-response approach will reduce the wasteful excess of some goods and the shortage of others.

"The next step is to get the supplies from point A to point B as quickly and inexpensively as possible," says Kulkarni, a student of IEOR department chair Ilan Adler’s.

To attack the logistics problem, iCare leverages wasted space on commercial trucks constantly traversing the nation, dropping off goods and returning partially or entirely empty. Once it is fully operational, iCare will access existing databases of shippers willing to donate their empty space. The researchers' algorithm then calculates the best route, often involving a relay race of multiple trucks, to get the donated goods to the victim using just the surplus space.

While iCare is designed from the bottom up to support ad hoc person-to-person aid, Kulkarni adds that they have "no intention of trying to supplant formal disaster response organizations." Far from it. In fact, the researchers are building data mining and visualization features into the system that they hope will make it an appealing tool for aid organizations. The Red Cross, for example, might assess a disaster area and input that data, neighborhood by neighborhood, into iCare. The software could then identify donor groups and commercial organizations best able to provide supplies and equipment they need at the time.

"iCare also provides a real-time map of the geographical distribution of need and urgency," Bitton says. "Many organizations don't have a good idea of that, but we can aggregate that data and display it in a meaningful way that they may find valuable."

Bitton and Kulkarni plan to spend their summer bringing iCare online for an early fall launch. For these two researchers, the system is a natural application of what they've learned studying mathematical modeling and information systems.
 
"We wanted to apply some of our skills to a pressing social problem," Kulkarni says. "There's a massive desire on the part of the public to help after disasters, and they just need a good way to transform that desire into something tangible for the victims."

Find this article at:
http://www.coe.berkeley.edu/labnotes/0607/icare.html

New Spin-Off
by Paul Spinrad

Nano-scale polymer fibers—the thinner, the better—can potentially enable the manufacture of more effective chemical sensors, biochips, protective clothing and other innovations. But the standard technique for producing these fibers, electrospinning, produces a chaotic tangle rather than controllable patterns. A variant of the technique, near-field electrospinning, offers full control over the path of deposited nanofibers, allowing them to realize their engineering potential. "It could open the field up, taking it in completely different directions," says the technique's pioneer, Berkeley mechanical engineering professor Liwei Lin.

Polymer fibers such as polyester are traditionally made by ejecting a polymer-solvent solution from a small hole (the spinneret) and letting it solidify into a thread as the solvent evaporates. The process of electrospinning, first patented in 1934, makes these threads even thinner by using a polymer solution that's charged, then squirting it across a high-voltage gap. The liquid flows from a needle onto a grounded collection plate 10–50 cm below. Applying 10,000–30,000 volts across the needle and the collector accelerates and stretches the stream. But, as the stream approaches the plate, it whips around unpredictably, depositing the delicate fiber into a fine mess.

Lin and his student Chieh Chang revolutionized the technique by bringing the spinneret much closer to the collection plate, to between 500 microns and one millimeter above its surface, to steer the deposited fiber into a controllable shape. Their first success came by using a more highly concentrated polymer solution and, instead of squirting it from a syringe, using a 25-nanometer sharp tungsten point, which they dipped into the solution like an old-fashioned pen.

"The liquid leaves the tungsten tip and forms a stream that follows a fluid dynamics shape known as a Taylor cone," Lin explains, "That close, it's harder to get a thin fiber, so we use a lower voltage, usually around 600 volts."

Last year, Lin and Chang demonstrated the technique by writing a microscopic, cursive "Cal" in 100-nanometer thin polymer fiber. "This was very hard to do, because we did it manually," Lin recalls. "You're moving the stage, not the pen, so you have to write upside down and backwards. And the solution continues flowing whether the collector is moving or not, so if you slow down or stop, it bunches or beads up rather than making a steady line."

One limitation of this technique is that it can produce fibers no longer than several centimeters in length. Just like a fountain pen that runs out of ink, the fiber ends when the last of the liquid polymer has been drawn off of the tip. But for many applications, short fibers are all that's needed.

Lin and Chang have since tried near-field electrospinning with the more traditional hollow needle. Using a needle with an inside diameter of 70 microns, they can produce fibers ranging from between 500 nanometers and 3 microns, thicker than the ones produced using the tungsten tip, but with unlimited length.

For improved control, they now also put the collector onto a precision X-Y stage, a piece of lab and chip-fab equipment that makes small, controlled movements along two perpendicular axes. Lin's group is also testing ways of dynamically varying the voltage to even out the fiber's thickness as the speed of the collector changes.

Polymer nanofibers have several known applications and others yet to be discovered. With polymers formulated to change their electrical resistance in the presence of certain chemicals, nanofibers can be made to act as ultra-sensitive chemical sensors. String several strands together in a handheld device, and you've got a portable sniffer that detects a range of airborne compounds—explosives, for example.

If you make nanofibers out of polyethylene oxide, coat them with glass, then etch away the original fiber, you create nano-pipe fluidic connectors. These can move fluids around to different detection areas in "lab-on-a-chip" biochips for medical diagnosis.

Traditional, disordered electrospun fibers have already been used as "bio-scaffolding" for cell tissue cultures, and nanofiber scaffolding that's precisely arranged could potentially direct cell growth to produce blood vessels, bone, cartilage and other tissues following predetermined designs.

Other possible applications include filtration media, drug delivery, solar cells and even cosmetics.

"There are a lot of people working on electrospinning now; it's recently become a hot topic," observes Lin. "Part of this comes from nanotech's popularity, but it's also because electrospinning is very easy to do. For our work, all we need is a power supply and a needle."

Find this article at:
http://www.coe.berkeley.edu/labnotes/0607/spinoff.html

The Man Who Made E-mail Go
by Patti Meagher

It was 1980. Personal computers hadn’t been invented yet, and Soda Hall hadn’t been built. The Internet was barely 10 years old and still in its nascent form known as Arpanet, the networking system developed by the U.S. Department of Defense. Access to this powerful tool was exclusive in those days, limited to academic researchers working on defense projects.

Eric Allman (B.S.’77 EECS, M.S.’80 CS) was pursuing his master’s degree and working on Berkeley’s INGRES project, one of the world’s first and most influential relational database management systems. On their Cory Hall computer, INGRES staff like Allman had access to the powerful Arpanet network, a rare luxury coveted by the computer scientists in Evans Hall.

“There were sometimes fights,” Allman says. “A lot of faculty and graduate students wanted accounts on our machine, but that was impossible.” He was the only one in his group who knew anything about e-mail.

“At one point I figured, OK, I can write some software that glues this one software to this other software, forwards mail from Arpanet to their primary machine so they don’t have to switch to this machine, and then forwards it back out again. It was a quick hack, but it worked.”

From that simple problem-solving exercise evolved one of the first implementations of SMTP (Simple Mail Transfer Protocol), the Internet protocol used to deliver mail. Allman perfected his quick hack, working on his own time and distributing the program to Berkeley’s Computer Systems Research Group (CSRG), which had a contract with the Department of Defense for developing an operating system that would facilitate collaboration among researchers.

The product, better known as sendmail, celebrated its 25th anniversary last fall. Sendmail became an important element of the Berkeley Software Distribution—the open-source operating system developed by the CSRG in the early 1970s—and now delivers more than 70 percent of the world’s e-mail. But if he had known then what he knows now about spam, Allman says, he might never have tackled e-mail.

“It bugs me that people are using this system without realizing that they could destroy it,” he explains. “I’d like to think better of the human race.” The spam filters that are indispensable today could not have been built into early versions, Allman says, because the emphasis back then was on sharing, not security.

Now chief science officer of Sendmail, Inc., Allman focuses on developing authentication and encryption tools like DKIM (DomainKeys Identified Mail) and milter (mail filter) to better protect our electronic mail from interlopers. Sendmail’s primary commercial product, Mailstream Manager, provides security and a host of other mail management functions to a majority of Fortune 1000 companies in 33 countries. Allman founded the Emeryville company in 1998, after e-mail became wildly popular and too “mission critical” for him to handle alone; until then, he had been distributing and supporting the software for free.

What Allman likes best about e-mail, he says, is its archival value, but he thinks both ordinary people and experts overuse it. When a decision needs to be made, for example, and e-mails go back and forth for days, he explains, “I want to say, ‘Guys, get up and walk into that person’s office and deal with it!’”

Allman, an El Cerrito native, was 14 years old when he first got his hands on a computer, an IBM 1401, and one of the first things he did was to recode the operating system. A self-described “social outcast” who was gay but still in the closet, Allman says computers allowed him to escape from the world.

“When I came to Berkeley in 1973, it was a truly exceptional time,” he remembers. “I learned from Ken Thompson [B.S.’65, M.S.’66 EECS], one of the original authors of UNIX, who took a sabbatical from Bell Labs.” He also rubbed shoulders with some of the most famous names in the business, including fellow alums Bill Joy (M.S.’79 EECS) and Eric Schmidt (M.S.’79, Ph.D.’82 EECS) and retired Berkeley professor Robert Fabry. While Allman’s name may not be as famous as these, he prefers it that way.

“It requires a different ego to write this kind of software,” he says. “A lot of people want to code video games; they want you to know it’s their software. But with a mail transfer agent, you want it to be invisible. The only time people even know it exists is when it’s broken. And you never want it to be broken.”

Find this article at:
http://www.coe.berkeley.edu/labnotes/0607/allman.html