The rapidly growing opportunity to crowd-source national security.
- By Christopher StubbsChristopher Stubbs is an experimental astrophysicist on the Harvard faculty, with an interest in technical issues that pertain to national security. He is also a distinguished visiting fellow at the Hoover Institution. Sidney Drell is a professor, emeritus, at the SLAC National Accelerator Laboratory and a senior fellow at the Hoover Institution. , Sidney Drell
When Mita bought her new iPhone, she had opted-in to the "citizen-scientist" program. Her smartphone was continually monitoring and storing data from built-in motion sensors, from the GPS receiver, and from the newly developed MEMS Krypton and CO2 gas sensors. Yes, buying the new "green" iPhone had cost a bit more, but Mita wanted to do something to support the new carbon emissions treaty and the president’s vision of a world free of nuclear weapons. Only 10 percent of smartphone buyers had chosen this more expensive option, but that still amounted to hundreds of thousands of verification sensors in Mita’s country alone.
Mita set the phone down on a bench as she rested on her weekend stroll through the park in her small town. The "verification app" she had running in the background was continually assessing the data stream from the smartphone’s sensors, searching for anomalous events. On this particular morning the sensor suite showed a combination of measurements that departed significantly from the historical norm for this time and location. In fact, the accelerometer registered a jolt that was consistent with an earthquake — or perhaps an underground nuclear test. Mita’s smartphone reported this anomaly to the international public verification clearinghouse, and transmitted the buffered time history of measurements, along with time and location tags.
Within a few seconds the clearinghouse had determined that there were 83 other public verification sensing nodes in the area around Mita with the sensitivity to detect such a tremor. The clearinghouse software first validated and then reviewed each of the datasets. A coherent joint analysis of all the accelerometer data from the surrounding area was inconsistent with the seismic signature of an underground nuclear test. So, the clearinghouse designated this as a likely false alarm and automatically sent an email to Mita, thanking her for her ongoing support of the public treaty verification program and summarizing the alert and its resolution.
Verification — the process of ensuring that countries comply with treaties — often poses one of the highest hurdles to international agreements. Particularly when it comes to matters of national security, it is essential to be able to ensure that the other side is not secretly developing new threats by cheating. In the past, the foundation of such efforts has been what are known as "National Technical Means" — that is, spy satellites and other intelligence tools that governments use to keep an eye on the world, in addition to trained spies.
But the digital age is changing that. The increase in data volume, ever-improving connectivity, and the relentless evolution towards ubiquitous sensors in cell phones and other devices affords new opportunities for concerned citizens to participate in solving some of the thorniest health and security issues of our time. In the very near future, anyone with a cell phone will be able to serve as a weapons inspector.
The data gathered by NTM are typically held as highly-classified information, available only to professional intelligence analysts. Often, NTM are complemented by other verification tools, like onsite inspections. So, for example, the United States will verify Russia’s compliance with New START, the treaty that reduces the size of each country’s nuclear arsenal, not only by watching its military bases and such from above, but also through a series of arranged visits to look for behavior that might suggest the Russians are keeping more weapons than they should.
Verifying compliance with some treaties also relies on "Shared Technical Means," or STM — instruments and the data they produce that are shared among participating nations and with verification organizations, such as the International Atomic Energy Agency. For example, the Comprehensive Test Ban Treaty, which bans nuclear explosions, is verified in part by the International Monitoring System, a network of some 250 sensors located around the world. Data from these systems are shared among participating nations, but access is typically controlled by governments, even if the information is unclassified.
The reason that these things are so government-centric is not simply that the data are sensitive, though they often are, but also that the technology required for NTM and STM is incredibly expensive. Still, there has been some public involvement in treaty verification — even with nuclear weapons treaties. Numerous academic and national seismic research institutions help detect and identify low-yield underground nuclear explosions. And the Washington-based Institute for Science and International Security analyzes unclassified data from commercial photo reconnaissance satellites to monitor activities such as nuclear fuel enrichment or missile test preparations.
Soon, however, technological advances will permit advances in what we call "Public Technical Means," or PTM. This includes information that is either generated by or is made openly accessible to the general public, the scientific community, the private sector, and NGOs. Examples include images that are produced by commercial or scientific (as opposed to NTM) satellites and made available to the public; sensors that are attached to the global digital communication network, be it through mobile devices, desktop computers, or stand-alone sensors; and scientific systems, such as seismic networks and remote-sensing satellites.
Perhaps the most exciting aspect of emerging PTM is the prospect of individual concerned citizens becoming empowered to play a role in building a safer, greener world through smartphones and other mobile devices. This has been termed "participatory sensing," "ubiquitous sensing," or "crowd-sourced science" in other applications. The sensors that we describe in our fictional scenario above are not so far from reality. Today’s smartphones are capable of transmitting massive amounts of data. They already have accelerometers and GPS receivers. Samsung’s Galaxy S4 has sensors that measure humidity and air pressure. The ability to measure CO2 levels and the presence of trace gases — even germs and other biological agents — is not far behind. We are on the verge of being able to crowd-source elements of national and international security.
Crowd-sourcing verification is not simply a way to engage and empower interested citizens, it can provide a unique and powerful verification tool. Further reductions to the world’s nuclear arsenals, and perhaps realizing the vision of a world free of such weapons, will place ever-greater importance on verification because the ramifications of cheating will only become more serious. Public means of verification or societal monitoring can be a significant tool for detecting suspicious activity, alerting governments that they should target their national technical means on a specific site, or that international inspectors might want to investigate. What’s more, the PTM approach can allow monitoring of countries that have not signed on to certain international agreements.
Not only is the demand for public technical means likely to increase, but the sophistication of those means is likely to be more agile than those fielded by nations or international organizations. PTM capabilities can evolve and adapt rapidly, without being constrained by the arduous international negotiation needed to revise or upgrade treaty verification measures. Once signed, treaties tend to last a long time. But the associated verification protocols seldom keep up with technical developments. By contrast, the typical replacement time for smartphone users is 18-24 months, so new sensors could be implemented rapidly.
The net ef
fect of the rapidly evolving potential of public and societal monitoring is to make it much more difficult for governments to violate established agreements and get away with such deception for very long. Many more people will know what is going on — around themselves and around the globe.
Already, some of these efforts are being put into place. Stanford’s Quake-Catcher Network (QCN) has developed a sophisticated framework for collecting and combining the data provided by the acceleration sensors in smartphones and laptops. The QCN system allows participants to sign up and download software to their mobile device. If the software detects an acceleration event that meets certain criteria, it transmits the data to a local QCN server. The network’s architecture then uses the signals seen across the sensor network to assess the significance of detected acceleration events.
Clearly, this effort has the potential to contribute directly to the verification of the test ban treaty. In fact, we have independently confirmed that the accelerometer in the iPhone 4 should allow 95 percent confidence in detecting an underground nuclear test with a yield of 1 kiloton from 150 kilometers away.
Public efforts need not be limited to the passive collection of data; those who have a sustained interest in a particular problem can assist with analysis as well. The Galaxy Zoo project was established to recruit minimally trained citizen-scientists to classify the characteristics of galaxies from digital astronomical images. The response was enormous. The Galaxy Zoo project attracted 250,000 online participants from 170 countries. This public wave of amateur astronomers succeeded in classifying over 100 million galaxies, by visual inspection of 100 million images — over 27 terabytes of data. Numerous scientific results have been drawn from the galaxy classifications performed by this volunteer community, demonstrating the tremendous power of crowd-sourcing data analysis, as well as providing valuable lessons for how to carry out image analysis projects in the future.
There are a number of technical hurdles to using Public Technical Means to verify compliance with a treaty. The integrity of the dataset itself must be verified to prevent spoofing and deception; sifting through data collected by a variable and dynamic group of sensors to extract signatures of interest is challenging, to put it mildly; uniform standards must be applied so that different phones don’t produce data in different formats; the level of privacy that participants are afforded must be addressed (e.g., will the system track their location at all times?), but at the very least they must be protected from any retribution, retaliation, or pressure as a consequence of their involvement.
These are not insignificant challenges, but it would be a waste not to take advantage of all this power. The system that monitors compliance with the Comprehensive Test Ban Treaty will ultimately have 337 installations. But tens of millions of people now carry around a sophisticated computer with built-in sensors, a powerful CPU, data storage, GPS geo-location, and wireless connectivity that also happens to be a telephone. The growth of smartphone ownership is projected to continue to increase worldwide well into the future, to say nothing of tablets, laptops, and desktop computers. There is also a steady increase in the number and instrumental capability of nongovernmental satellites observing the planet, producing new streams of data and images. "Ubiquitous sensing" is now an established technical phrase, with associated conferences, journals, and a critical mass of engineers and scientists that consider it an established subfield.
One obvious next step would be to carry out some informative PTM exploratory projects that the public is likely to support and for which the appropriate technology has already been widely adopted. One example might be to harvest all the smartphone sensor data from willing participants in one major urban area, and investigate the sensitivity, clutter, and signal-to-noise properties of the distributed network as applied to, say, seismic sensing. This seems an area where scientists and engineers could work in partnership with government, industry, NGOs, and international agencies to identify both opportunities and needs, and work towards the implementation of new capabilities.
This also seems like a good opportunity to build some prototype smartphone verification apps and (at a more advanced level) to explore the areas of user interfaces and user feedback. The distributed-sensing community is wrestling with the difficult problem of extracting knowledge and understanding from the noisy and cluttered data that are generated by wide arrays of inexpensive sensors. We advocate establishing stronger linkages between the verification community and this rapidly evolving subdiscipline. The recent State Department "Innovation in Arms Control Challenge" is a good step in this direction.
Given the rapid rate of innovation in information technology, and with the ever-evolving capability for both the acquisition and the analysis of large datasets, the technical prospects for crowd-sourced verification are promising. However, there are also diplomatic and political obstacles between stating policy goals and actually implementing the technical framework needed to achieve them. Without political strength at home, courage among global partners, and international organizations with the power and will to legitimize appropriate actions, Public Technical Means and societal monitoring will be of little if any value, no matter how revealing the data they provide.