This CHM conversation (with NY Times
moderator John Markoff asking the questions) was more about the challenges faced by Ms Arati Prabhakar, PhD
then it was about DARPA. It would've been very appropriate for a Women in Engineering
meeting. However, there were several important topics related to Ms Prabhaker's two terms of employment at DARPA, which we've attempted to capture in this event summary article. Note the addendum on Silicon Valley looking to recreate its past via companies establishing innovation labs.
Brief Backgrounder of Arati Prabhakar:
Growing up as an Indo-American in Lubbock, TX was quite challenging, but not nearly as difficult as being a woman in the EE curriculum at Texas Tech. After obtaining a BSEE from Texas Tech, MSEE and PhD in Applied Physics from Cal Tech, Arati Prabhakar started working on Silicon Gallium Arsenide (GaAs)
projects at DARPA in 1986. Ms Prabhakar later started the Microelectronics Office
there during her first tour of employment. It's now called the Micro Systems Technology Office (MTO).
After leaving DARPA in 1993, she became the Director at NIST
and then worked for a number of Silicon Valley firms, before returning to DARPA in July 2012 to become its 12th Director.
GaAs Projects at DARPA- then and now:
- GaAs was of interest to DARPA in the mid 1990's, because of its "radiation hardened"properties.
- Bell Labs tried to build a 16K bit memory out of GaAs material.
- GaAs was said to have higher electron mobility than traditional semiconductors, but that didn't turn into a competitive advantage.
- GaAS did blossom in RF devices used in the microwave world. It is also used in advanced radar systems (see below).
Advances in traditional MOS VLSI/ULSI technology precluded GaAs being used for many promising potential applications (e.g. high speed communications). That's because it was more expensive than MOS, used more circuit board space, and consumed more power.
What has DARPA Done and What Are They Doing Now?
- GaAS arrays are used today to build advanced radar systems for military aircraft and ships.
- The GaAs power amp in cell phones (which communicates with cell towers) traces back to GaAs research done decades ago at DARPA.
- GaAs will also be used in the electronics within cell towers (Arati did not say how).
- Gallium Nitride technology has come into full fruition and will be used in the next generation of military radar.
Arati Prabhakar noted the following points during her conversation with NY Times
John Markoff and selected questions from the audience that followed:
- We are at the end of Moore's law and semiconductor companies are well aware of that.
- The semiconductor industry has become totally globalized, especially manufacturing (Note: outside of Intel, most ICs are made in China, Taiwan or South Korea).
- Geopolitical threats have refocused a lot of U.S. government research to counter-terrorism.
- Most powerful technologies developed at DARPA disrupt and change the way the military works. That may cause intransigence in accepting new technologies progressed by DARPA. "Here comes DARPA again..." some military people might say.
- Over several decades, DARPA has built a compelling track record which has earned them more respect and credibility from the U.S. military.
- It's not easy for DARPA to shift from stealth technology development or to move to precision guided weapons or infra-red night vision military systems.
- DARPA's Robotics Challenge is a contest that had its first trials last December. The winner will be decided in the finals which will take place "in about another year (2015)."
- Such a "DARPA Challenge" is great way to interact and build technology for a greater community.
- Ground robotics is an incredibly difficult challenge. Bandwidth was being squeezed on and off to simulate an environment where there was no communications. As a result, the robot couldn't count on consistent communications (with the host computer).
- DARPA invested $$$$ in Boston Dynamics which Google has now bought. DARPA was very pleased with that. "It's a very promising sign," Arati said.
- There's a long history of DARPA making investment in technologies to show what's possible. As the technology matures, private capital gets involved in the next stage.
- Robotics will take massive private investments and real markets to become commercially viable. It has tremendous potential to help the military on unmanned missions.
- A robot developed by a Japanese start-up company named Schaft (now owned by Google) won the DARPA Robot Rescue challenge last year. Along with seven of the other top-scorers, Schaft can now apply for more DARPA funds to compete in next year's finals.
- DARPA's budget is only 2% or 3% of all U.S. government R&D. [That's amazingly less than what one might expect from the organization that created the ARPANET- the precursor to the Internet].
- What is R&D? At DARPA, it's building proto-types and basic research into new areas that will open up technology opportunities.
- How fragile are research technologies and industrial policies? It all depends on the follow on use in the military and commercial worlds.
- "The brain is a new (research) area for DARPA," said moderator John Markoff. (Arati said below that DARPA has been investing in brain technologies for some time- so not really "a new area").
- "Biology is intersecting with physical science and information technology," said Arati. It has enormous potential to be the foundation for a whole new set of powerful technologies. DARPA's interest is in turning biology into technologies.
- DARPA has been investing in neural technologies and brain function research for some time. One application is to get prosthetics to U.S. veterans that have lost their limbs.
- The human brain controls how our limbs move. Understanding neural signaling that leads to motor control is a DARPA goal. Three videos were shown to illustrate several DARPA neuroscience initiatives. The first showed a prosthetic arm controlled by a human's brain. The videos can be seen along with the entire program on the CHM YouTube channel. See link below.
- DARPA is trying to figure out how the U.S. can be the most productive user of new technologies and what areas of technology the U.S. should be thinking about developing.
- DARPA has invested in biological technologies for the last 20 years, starting with biological defense systems.
- DARPA has created a Biological Technologies Office (BTO) to foster, demonstrate, and transition breakthrough fundamental research, discoveries, and computer science for national security.
- Amazing things are happening in neuroscience and neuro-technology as well as in synthetic biology. DARPA wants to turn those into scalable engineering practices (both in terms of time and production cost) so that they can be commercialized.
- The ability to sequence and synthesize DNA is on an aggressive cost curve (although not a DARPA story) and that's an ingredient in the synthetic biology world.
- Arati is inspired by the progress she's seen in lots of engineering disciplines that together have potential to unleash important new trends in biology.
- But she's chastened about how little we understand in biological involved complexity, especially when compared to IT and semiconductors.
- Biology technologies are highly adaptable, but intractable when compared to transistors or lines of code.
- Biology related research examples at the BTO include: synthetic biology, brain work, fighting infectious diseases.
- Arati said that DARPA's two main jobs were:
1] Core mission is to pursue advanced technologies for use in the U.S.
2] Obligation to engage and raise technology issues so they get into the public eye, and that a broader community can then decide how society uses the new technologies.
CHM YouTube Channel Video:
- DARPA has a very substantial cyber-security portfolio. DARPA is not responsible for operational security in the U.S.
- "Patch and pray" is what we have (for operational national security in the U.S.) and we're trying to do that ever faster. DARPA wants to figure out technologies that give us a future with respect to cyber security.
- DARPA's focus is on the related security technologies that can be used for society to have a secure life and have the foundations for security in and away from home.
- There are many attack vectors, because of the vastness of our information environment. Therefore, there can't be a single silver bullet security solution. Rather, a layered set of many different security technologies are needed, which could be combined to thwart various security threats.
- DARPA is working to scale formal methods to build "meaningful size Operating Systems (OS's) that can be proven to be correct for specified security properties."
- DAPRA recently flew a drone in the courtyard of the Pentagon with an OS that has some properties that are unhackable now. "This could be the beginnings of a something that's a big dream," Arati said.
- Current state of GPS: It's cheap and easy to gauge your position once the satellite is up in the sky. Military is addicted to it. But new position, navigation and timing systems are also needed.
- Future geo-location timing and navigation systems will be based on layered systems and "atom physics." The latter involves cooling atoms by shining lasers at them so that their atomic properties can be tapped.
- The challenge is how to get these new geo-positioning and timing technologies from a huge room with researchers, to a shoe box size unit in a submarine. A key objective is to minimize or eliminate drift in timing or position.
- Quantum components in the way we smell have been researched at DARPA , but quantum computing "is not an active area" (as many thought). Arati was not sure if anything is going on in quantum communications at DARPA.
- Completely breaking complexity of massive platform military system is a DARPA goal. "They are un-Godly expensive, complex, and tightly coupled." The result might be visible in next generation fighter planes, in drones and what a soldier carries on his body.
- DARPA has created the Mining and Understanding Software Enclaves (MUSE) program to reduce software complexity.
- MUSE seeks to make significant advances in the way software is built, debugged, verified, maintained and understood. Central to its approach is the creation of a community infrastructure built around a large, diverse and evolving corpus of software drawn from the hundreds of billions of lines of open source code available today. Automating assembly of code to do higher level functions is a goal.
- Today, 2/3 of R&D investment is from private enterprises, rather than the 50/50 private/public sector split of several years ago.
- Private R&D investment is growing faster than GDP [Note that doesn't take much with GDP growing at <2% for last 5 years since recession "ended" in June 2009]
- Role of government is shifting when it comes to R&D investments. The federal government is no longer the prime source of research, as it was for years and decades. We now have incredibly innovative industries, Arati said (this author strongly disagrees).
- Our ecosystem (the U.S. government, universities, and industry) is healthy. We have adapted to change after change and she is optimistic that will continue in the future.
- How universities pursue research and what's going to drive the educational system are key questions U.S. must address.
- "Universities have always been a way that enables DARPA to get great things done," Arati said to wrap up the program.
The video of this CHM event can be viewed here.
Comments from Richard Weiss, DARPA Director of Public Affairs
(received July 4th via email):
I am surprised at the number of errors in your blog, especially since much of the relevant information is on DARPA’s website.
I don't believe Arati would have said that one of DARPA's two primary missions is "Obligation to engage and raise technology issues so they get into the public eye, e.g. how U.S. society uses the new technologies."
DARPA is not a policy shop – it is a tech projects shop. So while DARPA does take seriously so-called ELSI obligations when a tech development raises societal questions, that is not a “primary mission” of DARPA.
It is incorrect (perhaps just sloppy sentence construction) to say that “Substantial cyber-security portfolio at DARPA is subject to “patch and pray.”” Certainly our portfolio is not “subject to” patch and pray. Our portfolio aims to improve upon the nation’s current reliance on “patch and pray” when it comes to cyber security.
Author's Response: T
he post has been updated to reflect some of Mr. Weiss' critiques (after carefully listening to the archived event webcast to verify accuracy of same). Clarifications were also added (e.g. who said what) to avoid misunderstandings.
It's my strong opinion that a reporter's job is to accurately "report" what was said at an event. An event summary is not a research paper where the company's website is to be checked for accuracy or completeness or to validate/confirm what was said during the program.
Finally, you can check the archived event webcast to verify that Arati did say: "(It's DARPA's) Obligation to engage and raise technology issues so they get into the public eye....."
Addendum: Excerpts from NY Times article-
Silicon Valley Tries to Remake the Idea Machine (Bell Labs)
This superb article chronicles the decline of research amongst established silicon valley companies which have bought start-ups rather than invest in their own research labs (like the long gone AT&T Bell Labs or Xerox PARC). But a resurgence is underway, starting with Google X.
A few excerpts from the article:
"The federal government now spends $126 billion a year on R. and D., according to the National Science Foundation. (It’s pocket change compared with the $267 billion that the private sector spends.) Asian economies now account for 34 percent of global spending; America’s share is 30 percent."
"Most of the insurgent tech companies, with their razor focus on advancing the Internet, were too preoccupied to set up their own innovation labs. They didn’t have much of an incentive either. Start-ups became so cheap to create — founders can just rent space in the cloud from Amazon instead of buying servers and buildings to house them — that it became easier and more efficient for big companies to simply buy new ideas rather than coming up with the framework for inventing them. Some of Google’s largest businesses, like Android and Maps, were acquired. “M. and A. is the new R. and D.” became a popular catchphrase."
"But in the past few years, the thinking has changed, and tech companies have begun looking to the past for answers. In 2010, Google opened Google X, where it is building driverless cars, Internet-connected glasses, balloons that deliver the Internet and other things straight out of science fiction. Microsoft Research just announced the opening of a skunk-works group called Special Projects. Even Bell Labs announced this month that it is trying to return to its original mission by finding far-out ways to solve real-world problems."
"Instead of focusing on basic science research, “we’re tackling projects that advance science and solve significant problems,” says Regina Dugan, the former director of the Defense Advanced Research Projects Agency (Darpa), who now runs a small group inside Google called Advanced Technology and Projects. “What this means is you’re not compromising this idea of doing really important and interesting science and this sense of it really mattering.” To put a finer point on it, Astro Teller, who oversees Google X, told me: “We are not a research center. We think of ourselves as a moonshot factory, and the reasons for using that phrase is the word ‘moonshot’ reminds us to be audacious, and the word ‘factory’ reminds us we have to industrialize it in the end.”"