Future Current

Tad Hogg is a researcher in the Social Computing Laboratory of Hewlett-Packard, focused on harvesting the collective intelligence of groups of people to optimize the interaction between users and information.  At the 2008 Global Catastrophic Risks conference in Mountain View, he presented on the prospects of distributed surveillance with MEMS & nano-scale sensors.

The following transcript of Tad Hogg’s GCR08 presentation “Distributed Surveillance with MEMS or Nano-scale Sensors” has not been approved by the speaker. Video and audio are also available.

Distributed Surveillance with MEMS or Nano-scale Sensors

I’m Tad Hogg.  I’m from Hewlett Packard Labs just a short distance from here.  This presentation will follow quite nicely from the one we just heard, looking at some positive uses for sensors, as well as worrying about their dangers in terms of society and totalitarianism.

The discussion we already heard is that as you get advanced technologies, it allows smaller and smaller groups of people to do worse and worse things potentially, as well as larger consequences for doing good things.  The concern here is that the old defenses that we are used to, both technical, social and institutional may just not be good enough when you have these new technologies coming along.

There are a number of possible responses, and the one I am going to be discussing is ubiquitous sensors of various sorts, basically making the point that you would like to have a variety of types of sensors widely distributed.  That is in essence the technical end of it, but also in terms of institutional and social aspects: multiple owners—many people watching whatever else is going on, including the other people with their sensors and so on, so that you have a diversity of approaches.

When we think about what we would like these sensors to do for us, there are basically two kinds of technical ones.  The first one is accuracy.  You would really like them to focus on what are the relevant events.  Sensitive, so that they can detect small or early versions of whatever events might be of concern.  Then also, a very important social aspect, the information that these sensors deliver to people who will be making decisions is trusted information, rather than more and more spam.  There are both technical and sociological questions here.

What I would like to do in the next fifteen minutes is basically look at three different aspects.  One, how do we go about making these sensors—what is the state-of-the-art for the near-term?  How do we go about deploying these sensors in a variety of ways?  Then, to turn a little to the social and institutional issues, how do we go about using them?

I’m going to be talking about small sensors, which in essence are supplements, not replacements for the kinds of large sensors that we are already used to, in terms of cameras and so on.  I’m going to be distinguishing between two kinds of scales.  In some sense these are both tiny, and sometimes these get mixed together in terms of what’s possible to make now versus what we think we can make in a few years or further down the road.

To think of microsensors (MEMS is the term that this goes with), these are roughly millimeter-scale sensors made out of micrometer-sized parts.  Further out are micrometer sensors, the size of biological cells, whose components are nanoscale parts.

There are a number of examples of micromachines from several years ago.  If you look at the size of a tip of a finger, it gives you an idea of what size the device is.  Unlike what we think of as sensors, in terms of cameras and so on, these things tend to be more toward measuring forces, accelerations, chemistry.  There is a sort of grandiose vision of this scale device called smart dust or motes.  This is an artist’s conception, where you would put these things together into a package that would have some power and communication, being able to do some kind of monitoring in the environment.

Going down by a factor of a hundred or a thousand in scale, there is a lot of current work on nanoscale chemical sensors.  These we are not yet able to put together into these complete packages.  There is quite a variety of approached aiming toward single molecule detection of various sorts, much of it oriented toward biomedical applications.  These would nevertheless be applicable to looking at a wide variety of chemicals.

Now, there are a number of challenges for building and developing these small-scale sensors.  A very major one is, how do they get power?  Will it be passive power that they can somehow pull out of the environment, or active?  In terms of communication, will they communicate with each other?

Instead of talking about those, I’m going to describe a little bit more general kinds of problems in thinking about how we use these sensors, basically because they are living in a very unfamiliar world compared to what we’re used to.

We are visual creatures.  We see and hear things at a distance, and that is what cameras will do also.  These smaller sensors are much more like noses.  They are detecting chemicals or forces, things that are short range.  They can be very precise and pick out very particular kinds of chemicals, but it’s not like they are making an image of what is on the other side of a parking lot.

Another interesting aspect of tiny sensors is that the physics that is relevant for them is sort of what we knew before we went to high school physics and leaned about Newton’s laws of motion.  In this world of sensors it is really Aristotelian physics.  If there are no forces, there is no motion.  Inertia is just not an issue for these kinds of devices, especially when we are thinking about the nanoscale ones the size of cells.  Their world is very different from what we are used to.  Their physical environment, the way they interact with the world, the way they detect things and so on, it is something that we are not used to, being such large creatures.  Percell has this fun little article, which appeared a number of years ago, on what it’s like to be organisms the size of bacteria.  It gives you an idea of what the world would be like from the point of view of these sensor devices.

Now let me turn to some ideas of how might we deploy these sensors in the world to look for a variety of potential threats.  The most obvious one is to deploy them in our environment, then we can talk about some other ones that may be less obvious but could give us a lot more sensitivity if people feel comfortable with them.  That will then lead us to the social end of things.

Sensors in the environment would allow continual, widespread monitoring.  It would provide the opportunity for multiple, redundant networks.  You are not necessarily relying on one central decision maker to collect all this information and decide what it means.  They may for various reasons not be able to interpret things correctly or there might be some disagreements.  If it is just a single source of interpreting, you might have all these widely distributed sensors, but then you go through a bottleneck of interpretation.  They might have reasons intentional or accidental to misrepresent the information.

We are moving more and more to a situation where people carry around a lot of electronic devices.  These could easily have addition sensors.  People could monitor what kind of interesting chemicals you come across.  Provided you could communicate this information around, you then would have a source of widely distributed mobile sensors owned by a number of different people as a way of getting sensors that move around in the environment.  This could also be in a much more widescale version.  Say you put sensors in money, or any other thing that is transacted and moved around.

The idea of these kinds of mobile devices that carry these sensors around is a bit like the way our immune systems work.  You have cells that go around looking for trouble, and the important thing is that they get near everything, just to see where trouble might happen, and not just wait for it in one part of the body.  Something like cell phones or money that distributes all around among different people has the chance to get close to whatever might be of concern.

Now we go one step further when we talk about nanosensors.  These devices are potentially getting small enough that they go inside our bodies.  A primary motivation for developing these things would be for medical diagnostics and therapeutics to help in medical treatments.  A secondary use would then be to have very early detection of things that might be going wrong in a number of different people in very subtle ways—for instance, for early detection of epidemics, either natural or deliberately created.

Now let me turn to some of the social and institutional aspects of these sensors.  Supposing you could make them and supposing you could get them deployed and so on, the kinds of things that you would generally be concerned with are looking for fairly rare events.  It is one thing to have a chemical sensor that can detect single molecules of whatever; it is a very different thing to decide from all this information that many of these sensors might be providing whether they are actually seeing anything that is normal or where you need to do something about it.

This question of finding rare events is an important thing to keep in mind about not just how good the sensors are but how they are going to respond to the information that they are going to collect.  This leads to the problem of false alarms.  You can have the sensors that are looking for some particular toxin. When something triggers the alarm, do you respond by waiting until a number of other sensors in different places respond?  You have to have some kind of policies.  Different people might have different thresholds in which they become concerned and want to do something. Of course, if there are too many false alarms, as we are all familiar with with car alarms all over the place, eventually people ignore it.  Then there is the chance that you will miss the real alarm.

How is it that various groups that collect this information tell the information about that?  How is it that you convey to people what is it that might be a concern?  If you do this too often, people will just tune it out.  There is also fear about how the information of these sensors might be misused in various ways.  In particular, in terms of medical sensors that are put there for your own medical benefit, someone could ask if you would mind sharing the information with who knows whom so we could have a good early detection of epidemics.  That might make very good sense from the point of view of a society as a whole, but the information that these sensors collect is also very intrusive.  Things as simple as “I’ll be embarrassed if people find out where I was last night,” or if I have a particular kind of health condition that’s not yet serious, maybe I’ll lose my health insurance if the insurance companies find out about this.  Toward more serious things, you might end up with legal sanctions that you would want to worry about.

At an individual level it would make sense to say, “I don’t want the information in whatever sensors I have to be distributed.  I want to keep the information to myself.”  That actually raises a public policy issue.  How much do we focus on some of the individual sanctions that these sensors could be used for, realizing that the more we focus on using these sensors to prevent people individually from doing bad things will lead to people being less willing to reveal this information?  We could be blinded as a society to much larger threats, just because people become reluctant to reveal the information or carry the sensors around.

This leads to what The Economist would call a public goods problem.  “I’m kind of worries a little bit about this information getting out, but I’m also worried about these threats.  By the way, if all my neighbors have all this sensor information that they are displaying, I’ll just be protected by them, and I don’t need to share my information.”  If everybody thinks that way, nobody gets any information at all.  This is a classic public goods problem which comes up in many situations that society faces all the time.  There are a variety of economic approaches to try and change those incentives.

Then we come to the issue of, suppose there are a lot of sensors out there producing a variety of information and people are summarizing them in different ways, how can we trust them?  We see what happens with email and spam.  People might want to deliberately spoof these sensors.  One general idea is you get robustness from diversity: diversity from the sensor types, diversity from who owns the sensors, where they are, and so on.  You also have to be aware that networks can be vulnerable to deliberate false rumor spreading.

Even inadvertently, multiple sources could be claiming there’s a problem, but they are actually all just getting the same news story, so to speak, and republishing it over and over again.  One approach to that is various social or economic reputation mechanisms to think of who owns the sensor and how is it that they’ve behaved in the past.  Could we use the social networks that we have online to deal with propagating trust, so I not only get information but I get a justification of why I should believe it, because it has various sources that certify it.  These are sources that I trust in particular and others may not trust.

To summarize, sensors that are widely distributed are one approach to threat reduction, particularly looking at the issue of very small groups being able to use advanced technologies to do large-scale detrimental things to society.  How is it that you can tell ahead of time?  There are challenges both technical and sociological, or institutional and in public policy, with respect to making these devices, deploying them, getting people willing to have them around and so on.

Here are just a couple of places to look for further reading.  If you are interested more in the technological end of it, there was a special issue a couple years ago in Applied Physics looking at nanoelectronics, which involves both the computation and control of these devices as well as what sorts of chemicals they can sense.  Probably more of you might be interested in the social and policy issues surrounding these things.  I imagine most of you have heard of David Brin’s book The Transparent Society, which looked at a number of these issues.  David Friedman has also looked more generally at some of the economic issues involved with dealing with technological advancements.  Then there is also an open source sensing project.  While I leave these here, I’d be happy to take any questions.

This entry was posted on Saturday, January 17th, 2009 at 12:14 am and is filed under IEET events, Nanotechnology. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.