Some thoughts on Google Glass.

07 July 2014

I feel obligated to make the following disclaimer:

Yes, I am still a privacy advocate. I still teach crypto and train people in using privacy-preserving technologies. I also still don't trust any service that I can't kick because data I produce through them is the product and not the service. That said, Google and Google Glass don't seem to be going away anytime soon. So, here are some of my thoughts on Glass.

If you've been bouncing around the consumer electronics set for a while you've undoubtedly heard of Glass, Google's foray into the red-headed stepchild of computer technology for the last few decades, wearable computing. Glass is an astonishingly small and light device that fits comfortably on the earpiece of a pair of eyeglasses with a mass of just 50 grams (about as much as a quarter cup of sugar). As a bit of trivia, the prototype of Glass developed in 2011 reportedly weighed eight pounds. It's a fully self-contained computing device that incorporates a dual-CPU System On A Chip, 1GB of RAM, 16GB of onboard storage, and a unique heads up display that hovers above the wearer's right eye that looks like a translucent 25" display. It runs a standard build of Android on board, so the user doesn't have to link it to a larger device unless there is no wireless network connectivity (Glass is not cellular-enabled). Android is a general purpose operating system so it can do pretty much anything a larger device running Android can do, including run apps from the Google Play store. The inexorable advance of computing technology has solved many of the user interface problems of early wearable devices; Glass sports natural language voice control, head motion tracking, a capacitative touchpad, and can be remotely controlled with an application running on a mobile device if the user desires. It also offers native integration with many of Google's services, from Gmail to Google Search and Maps, which make it ideal as a navigational aid because the necessary information appears in the heads-up display in realtime.

Unsurprisingly the announcement of Glass has spawned a new and fresh kind of controversy, as only things intimately connected with the global Net can. Glass' forward-facing camera and microphone immediately mark it as a potential privacy concern because people in the immediate area don't immediately know if they're being recorded or not. Some people seem to believe that Glass is always recording everything around it and act accordingly. This has resulted in a number of assaults upon the users (though a certain amount of asshattery was involved in some of the altercations (video) (mirror of the video in case it gets taken down)). It's even resulted in some impressive overreactions - in January of 2014 a movie theater run by AMC summoned a team of agents from Federal Protective Services (a field division of the Department of Homeland Security) to detain and interrogate the individual because he was wearing Glass attached to his prescription spectacles. In some ways this backlash is not dissimilar to some of the problems Steven Mann has encountered with his wearable projects over the years. Something I can't help but find interesting is that the attitudes of some of the most vehement anti-Glass protestors don't seem to involve the same amount of vitrol with regard to being recorded while walking down the street, in stores, in bars and clubs, or while traveling in taxis, buses, or trains. But what do I know? Maybe the perceived risk of retaliation is less when one attacks a person rather than an actual problem. A valid point has been made in saying that the considerable cost of Glass ($1500us) is a public display of conspicuous consumption. For many, a single Glass on someone's head represents a significant multiple of one's monthly income or a significant fraction of one's monthly rent. I don't claim to have any numbers on this but it seems to me as if one needs a significant amount of disposable income to even think about ordering Glass unless someone else is picking up the tab; certainly enough that one doesn't have to worry where the next rent cheque is coming from. On the other hand the retail cost of Glass is comparable to the prices of fairly popular laptop computers. Prices which are willingly met by some percentage of the same people speaking out against Glass, I hasten to add. There isn't much controversy in the tech community over the cost of laptops. Maybe it has something to do with perceived practicality. Others have said that Glass represents a more evolved form of classism, a claim which I cannot in good conscience deny. The tech industry is being seen more and more as a higher, more prestigous caste of society in some of the same ways that the nouveau riche are seen by the working class. This is echoed at lower levels of complexity in interesting and illustrative ways. For example, when Glass was first launched in 2013 prospective customers had to apply for them, and only a fraction were allowed to actually purchase the development kits. A year later the Glass marketing program was expanded into an open beta in a fashion similar to that of Gmail when it first launched in 2004. This drew a dark and solid line between have-access and don't-have-access in the tech field at the time and incited no small number of online battles as only computer geeks can wage.

I can't help but wonder how many people made this comparison but felt it prudent to remain silent for whatever reason.

Some have stated that Glass seems perfectly suited to covert surveillance, deliberate and otherwise. I'm of two minds about this for the following reasons: Glass requires the use of voice control to operate most of its functions. A user having to say "Okay, Glass. Record a video," out loud when people are around isn't very covert. A few people have written code that allows silent control of some of Glass' features (Github link) and more such applications are undoubtedly under development at this time. On the other hand several security vulnerabilities have already been identified in the version of Android that Glass runs, and it was rooted a month after the initial release. So it is conceivable that someone could write malware which compromises Glass through one or more of the most often used vectors of interaction (the camera or microphone) and turn it into a remotely monitored instrument of surveillance in the same way that phones are. This brings up other questions, of course. Can the on-board firmware be trusted, for example? A lack of source code means that it's anybody's guess what the binary image is really doing during execution. We know that Android is exploitable, but what about the drivers for the wifi chipset? For that matter, if Glass is hooked to someone's phone, and the phone in question is pwned, what could it conceivably do to Glass?

On the other side of the rapidly spinning coin, Google Glass is also a potentially valuable tool in other problem spaces. The obvious use that comes to mind is sousveillance, or the surveillance of the systems of power by everybody who has to live within them. When interactions between people and the state go wrong, there is often little that can be done without evidence of wrongdoing, and a subtle-looking wearable camera can help provide that evidence. Glass could also assist in documenting any number of nasty surprises that the world can throw at one, things which most media outlets would have no interest in unless it meant more eyes to show advertising to. Glass' microphone and video camera could also conceivably be used to facilitate realtime translation of languages both spoken and written. Google's speech-to-text API, which both Chrome and Android Voice Control make use of is scarily accurate at turning voices into intelligible text. I wonder if it could be used to convert speech in one language into text, which could then be automatically run through Translate and the results shown to the wearer in the heads-up display fast enough to make normal conversation across languages doable. It wouldn't be a perfect translation but in my experience it's decent enough to hold everyday conversations, though I do make a habit of saying "I'm using Google Translate to help communicate. I apologize in advance for any errors because machine-assisted translation isn't perfect," at the beginning of the conversation.

In effect, this means that Google Glass can be used as a cognitive prosthesis as well as an assistive technology, functionally no different from a cochlear implant. This is already being done, I hasten to add. There is a fairly mature project called OpenShades which is developing assistive Glass apps for the visually impaired which is doing some amazing things with object recognition and environmental mapping. The first of the apps is called Question/Answer, and takes advantage of crowdsoucing to do just-in-time identification of images. The second application, called Memento, uses spatial location and object recognition to speak text notes tagged to physical locations to the wearer. The OpenShades project has opened the code to it's development framework, a variant of JavaScript callled WearScript for rapidly developing applications for Glass. Another possible assistive application for Glass is a cognitive prosthesis to help allay the more frustrating symptoms of prosopagnosia, or the inability to recognize the faces of people which can make social interactions problematic at best. Google has said that they will not permit facial recognition software for Glass but that doesn't mean that we can't write our own and make them available as an option. The way I envision such an application working is something like this:

When first meeting someone new the user of the software would ask permission to take a photograph of the other person's face with Glass, which would be stored locally on the unit or in the phone. That photograph would then be associated with the person's name (spoken and recognized as text) and possibly a contact manager like Contacts. Glass is a Google project and is tightly integrated with their application set, so it would make a certain amount of sense to make use of existing services and APIs rather than roll new ones. It seems possible that other identifying characteristics could be integrated into the identification system to leverage existing non-standard sensory modalities but that might not be a v1.0 feature. Indeed there may not be a great demand for such a thing a few releases in. Later, Glass' forward-facing camera could be used to photograph people within range and facial recognition software used to search the user's database of people to identify them with augmented reality techniques, such as highlighting the outline of the recognized person and displaying their preferred name to the user.

Glass has already been used in the context of internal medicine, namely, as an aid to surgical technique in the operating theatre. A research team at the Wexner Medical Center of Ohio State University is already using Glass as a tool during surgery to bring medical students into the experience during surgical procedures. Glass has been used to consult with specialists offsite during heart surgery, display patients' vitals in the heads-up display, and access patients' medical records in realtime as part of the diagnostic process. Consulation with colleagues has also been successfully accomplished. If Glass was paired with an open-air gesture recognition system like the Leap Motion I think that diagnostic information could be rapidly accessed and manipulated during the procedure in such a way that greatly minimizes the risk of contamination.

I've learned over the years that the propensity toward overreaction of any kind rarely abates but can be pretty reliably expected to escalate as time goes on. I mentioned earlier the movie theatre owner who somehow got a Federal Protective Services field team to drag somebody away and question them for several hours just because he wore a deactivated Glass inside. What other kinds of "What the hell are you thinking?!" situations might arise? It's anybody's guess. The RIAA already has a well-deserved reputation for teeth-gnashing levels of stupidity and mean-spiritedness that even the most hardened misanthrope would shake their head over. ASCAP isn't much better. I'm kind to scared to think about what future backlash might be like to implanted sensory technologies, like prosthetic eyes and fully self-contained cochlear implants. Might it come to pass that someone with visual restorative implants might be sued on charges of pirating a movie or television show simply by watching it? Could someone with auditory cortex implants be sued by the RIAA for merely listening to a concert or a song on somebody's radio without any intent to redistribute what they heard? Or even if they had no way to record the sound to a file on removable storage? Extrapolating a bit from what's happening right now, how unlikely would it seem that people with augmented senses would not be ostracised simply because they have silicon inside them that acts as a camera or a microphone?

If nothing else the open source community seems to think that Glass has enough potential that they've started working on a much less expensive, more flexible and probably more trustworthy implementation of the technology. A couple of months ago Motherboard ran an article on such a project which combines some parts fabbed on a 3D printer, a Raspberry Pi, and a relatively inexpensive pair of video glassses into a fairly slick-looking rig called the Wearable Pi. The project takes a little bit of doing - it's hardly a beginner project - but in the end you'll have a Glass-like head mounted display hanging over one eye, the RasPi tucked into a pocket or belt pouch, and a slim cable connecting the two. To add photographic and video capability I think it would be possible to incorporate a small camera into the headset, something like this one or this one but it would increase the mass of the HUD and the power draw overall. As for the sorts of things you could do with it, the RasPi's a general purpose computing platform with a number of operating systems running on it. If what you're interested in doing doesn't already exist ample resources exist to help you write it yourself. If the RasPi's a little much for you there is a relatively inexpensive project board on the open market designed for just this sort of thing called the MetaWear which allows you to construct wearable devices that can then be slaved to your smartphone via Bluetooth to create a Personal Area Network. Whether or not enough apps exist on the market to do exactly what you'd want is a different matter, however; I haven't investigated them so I can't say one way or the other.So far as I know the Wearable Pi doesn't have voice command yet but there are options that could be explored in the near future.

In conclusion, as someone who really, really dislikes the state of surveillance in general, technologies like Glass give me the willies. I find myself pretty solidly in the "I've got better things to do with $1500us" camp, but that doesn't mean that some good can't come from Glass just because I'm not interested in buying one. Costs always come down over time, alternatives (some that are more trustworthy) are probably going to appear (if these two patents are any indication), and itches will be scratched with new software written. Google Glass and people acting like jerks while wearing Google Glass are two separate things and it would behoove us to ask ourselves what some of the people who've reported assaults are like when they're not wearing a head-mounted computer. Maybe some act like jerks regardless, maybe they were out of sorts at the time (everybody has bad days, after all), and maybe it had nothing to do with their actions but the actions of the people around them. The truth, a wise person once observed, is a three edged sword. If the twenty-first century has taught us nothing else, we would do well to not take the word of one or two articles as gospel but seek out more information, think about it, and come to our own conclusions. It could also said that this is a similar phenomenon to what many old-time computer nerds remember before computers were cool, which was taking one's share of lumps from people who didn't understand us, feared what we did, and believed that beating the snot out of us for being weird was an acceptible thing to do. Times may change but outliers always draw fire. While I don't think that everyone should or could embrace Glass or any of its inevitable clones it does show promise as a useful and beneficial tool in several problem domains, and we should consider the problems these technologies could help solve before breaking out the pitchforks and torches when unchecked surveillance is a bigger problem closer to any of us than a handful of people around the world are.