Hallo Du!
Bevor du loslegst den Talk zu transkribieren, sieh dir bitte noch einmal unseren Style Guide an: https://wiki.c3subtitles.de/de:styleguide. Solltest du Fragen haben, dann kannst du uns gerne direkt fragen oder unter https://webirc.hackint.org/#irc://hackint.org/#subtitles oder https://rocket.events.ccc.de/channel/subtitles erreichen.
Bitte vergiss nicht deinen Fortschritt im Fortschrittsbalken auf der Seite des Talks einzutragen.
Vielen Dank für dein Engagement!

Hey you!
Prior to transcribing, please look at your style guide: https://wiki.c3subtitles.de/en:styleguide. If you have some questions you can either ask us personally or write us at https://webirc.hackint.org/#irc://hackint.org/#subtitles or https://rocket.events.ccc.de/channel/subtitles .
Please don't forget to mark your progress in the progress bar at the talk's website.
Thank you very much for your commitment!




======================================================================





Hi, I hope you guys are having a good morning. I know it's early, and so yeah, I've already had my three cups of coffee, so almost caffeinated. All right. So today, as you may or may not already know, the topic of the talk is glass hacks. So this is fun and frightening uses of both Google's new wearable computer as well as other always on devices. So let's let's get started. There's a few kind of key concepts and overarching ideas that you know that that keep coming up when, when, both when I work on these types of problems, as well as when I think about this from a higher, higher level. And and those three things are. One is that we're moving into a world where we're able to collect as an individual massive amounts of data and you'll start to see throughout the talk that this is becoming increasingly true. So just the amount of, let's say, megabytes or gigabytes per day that one person can on their own collect has just, you know, gone up a lot. The second one is that alongside this increase the ability to collect data, we have an increased ability as individuals, not just large corporations or organizations or institutions, but as individuals to process this data and to do interesting things with it. And finally, the question is, well, what does this mean from a privacy perspective? What does this mean from from the perspective of of how we go and conduct our daily lives even? OK, so I'm going to talk a little bit about the history. I'll keep this brief about where we're competing in general. So there is this guy, Edward Thorp, who he wrote a paper in '98 claiming that he was the first one to invent a wearable computer. And what he was talking about was a small device that fit inside of a shoe and allowed him to do roulette calculations. So he is also pretty famous for coming up with blackjack card counting methods. And, you know, he was actually working with Claude Shannon on this one, if you know of information theory. And so it's kind of interesting. This is nin
eteen fifty five. This is a while ago, you know, so that if you look at the Wikipedia article, edit history, which is always fun when you're doing research, you know, of course, this was changed and and it's actually now now now solidified that Steve Martin is in fact the inventor because somebody edited it on Wikipedia to make sure that the working definition made made solidified him as the inventor of the wearable computer. And so the basic idea, the argument, I guess they were having was that one specific purpose versus one is a general purpose computer. But. Academic squabbling. No surprise this guy, Steve Mann, how many people have heard of Steve Man? OK, good. So Steve Mann deserves his own slide. So even if there is some question as to whether or not he invented the first wearable computer, it would be hard to say that he hasn't actually done everything else already. So this man is is the godfather to some degree of wearable computing and being creepy in general. I mean, like, I unfortunately, I can't defend him in that respect. He also did invent HDR, which this is an HDR image. So he he's done some really amazing things and has kind of lived his life as one big experiment to what happens when you decide to log everything you see. What happens when you decide to wear weird looking things on your body, as will come to very soon? So, Steve Man, wearable computer pioneer, I would say, anyways, that's about it for history. Now let's kind of let's go on a little bit to to the present. OK, so if I can get a show of hands, how many people here have have like seen these in person? This is a pair of Google Glass. OK, so some you know, I come from Silicon Valley, so you know, this is it. Unfortunately, I see too many of these. And but yeah, so here's a pair. And yeah, look, you know, I can put it on my face. Wow, OK, I'm going to put this over here for a second. We'll get back to that. But let's talk about data collection. So today, what kind of data do we collect thr
oughout our, you know, throughout our lives? Well, you know, you've got a cell phone. Maybe you've got an internet browser. You know what's being logged about you, your messages, photos that you upload to Facebook. Photos that Google automatically uploads to Google Plus and is viewed by Google employees around the world. You know, location data gets gets logged in general. So when you're using Google Maps, your location data gets it. As with any other service, for whatever reason, they need your location data. I don't always understand like what, for example, unless like Angry Birds wants my location data for. But so, you know, and then basically our data sent up to a server, it's aggregated in a central spot and it's like, have you left things done? But this is, you know, you don't already know all this. This is just this is obvious. So I posit that soon we'll be logging everything that every second of video audio your location, IMU data coming off of your various wearable devices will be logged. Now, whether that's in the central repository or in a distributed set of databases, you know that you control personally is yet to be seen. But we will see you and then a boon to our God. And so but I also think that so right now, instead of instead of this heavy lifting done server, we'll start to see a little bit more distributed kind of storage of this data, as well as processing of the data. So it being sent out to either multiple devices, the processing being done right on your device. This is kind of for the data collection where I see things going well, OK, what does Google Glass? I took, I took it out. That's an Android phone on your face. This is what Google Glasses. Google Glass is what runs Android 4.0. It's got the Ti OMAP platform 44 30, which is the same if you guys are hardware hackers out there. Panda bought A1. A2 has this sort of the the Samsung Galaxy S2. It's a very popular platform, and it's kind of maybe not a surprise that that Google is using it for
 Glass. And so Sarich, who many of you probably know from Cydia, has a really great write up if you haven't seen it already about exploiting Google Glass. So it does come with an OEM unlock bootloader. However, you know you can. You can get around that by just using a security vulnerability, which sort demonstrates in the links below. And once again, these are all on the slides on the internet, so you can go look at those links later. And by the way, also, this picture is from a really fantastic teardown by a friend of mine star Simpson and her friend Scott Borg, and they've got really fantastic care about it. Teardown of Google Glass and I would recommend you take a look at that. Just search cat wig on the internet. OK, so let's talk about the performance of Glass. Google Glass has got a like I said to Matt, 44 30. That's got a dual core arm Cortex-A9 processor. These are kind of some basic numbers that you'll see with the both the A9, as well as the SGX five SEER five series that does the the graphics processing unit on on Google Glass. So not that much processing power. And so but yeah, I think that over time, we'll see, of course, this stuff all going up. I mean, even the modern, some of the modern imagined GPUs get more like 60 or 70 giga flops, which is pretty impressive for mobile drive you. And so, you know, this will come into play later in the talk with regards to like what we're doing, what we're doing with this data. OK, the second one is battery life. You know, it's got a horribly small five hundred eighty seven million power battery. You know, this is gosh, like a third of what your normal cell phone is. It records video for a whole 60 Minutes. And you know, you can also have it like take pictures regularly if you if you install some software. I'll talk about later on it. But you get about eight hours of video. So once again, this pictures from the tear down, it's it's really fantastic. Turnout usability. OK, so when you wear Google Glass in public, no
t only are you shaming yourself, but you have the option to shame yourself even more by talking to it so you can wear it and say, you know, OK, glass embarrass me and embarrass me. And you know, they use voice recognition. It's very fancy you can take a picture. This is parody and I media all over the world, but it is actually really embarrassing to use the voice commands. And honestly, from a real user experience perspective, it all joking aside, it is. It's not an ideal interface for if you're sitting on a bus and you're like, OK, you know, send an email or like, send me pictures so you don't want to be talking to your your face computer. It's it's bizarre. Also touching it. You have to touch it on the side. You know, the best the best user interface experience here is the top. There's a button that's the best. You just tap it once for to take a picture that's actually really nice. I like that. Let's let's keep it there without talking to it. But anyways, that's just OK. That's an overview. Glass. That's boring. OK, let's talk about the fun stuff. OK, so as you may or may not know, Google Glass comes with their own proprietary launcher called Glass Dot app or Glass app or something like this. And you know, you can scroll through cards and do these these interesting things. But you know, when you install a new app, unlike on a traditional Android platform, it would show up in your launcher right now so you can't install. You're not supposed to install apps on Google Glass that way. And so in order to do this in a way that's actually convenient, you need to either write your own launcher or download one off the internet. I made one called Railgun. I'll make all the source available later. You can. If you can check back on on on the either the GitHub or the the sidebar, I'll have links to all this code. I'll tell you which ones have code and which ones don't, but you'll be able to load this onto your glass or your Android Android phones. This is basically just a way 
for for once we've installed these interesting APKs to be able to access them. OK, so this is a good one. Google Glass, as you can see here, does not have a hardware recording indicator light. Let's get a round of applause to Google for that. Wow. Oh my gosh. What were they thinking? OK, so that's just wonderful, you know? But but you ask, Well, how do you know when you're taking pictures or how do you let others know that you're recording them? Google says, Oh well, the screen turns on when, when, when you're taking a picture. And so this is enforced in software. And as we all know, that's totally not circumvented. So one of the one of the apps that I'll post code for later is basically this it. It it. Disables the video screen. Now I will give it to Google that you do, it is what I've currently found. You know, this implementation I didn't need to have, I didn't need to have root access on the device, which is not hard, obviously to get. But I did need root access to to actually turn off the display while taking pictures. But, you know, so like basically when someone's wearing Google Glass, you actually can't tell if they're recording you or not. Now the only thing that really limits somebody from recording you with Google Glass is really just the battery life, you know, like so. So Google has actually given you some kind of some kind of a limit there by by making the battery life so poor. So. So that's nice. OK. Another one is facial recognition. So a lot of this, you know, what I re talking about is what's possible today and then kind of saying, OK, well, what's also going to become possible very soon in the near future? And facial recognition is one of these things. That's a, in my opinion, really still a Two-Way application right now. So while it is useful in some context, especially if you have, let's say, very large amounts of pictures, let's say a wedding photo album and you need to, let's say, find a, you know, a dozen pictures of me if Steven in this in t
his wedding photo album, I'm not married, by the way, but let's say I attended someone's wedding and you know, so that's useful. But you know, the kind of thing where you hear in the media like, you know, recognize people on the street. This is not it's just not possible with current technology, with the current data sets that we have access to in terms of, just like widespread mass facial recognition. However, I think that it will become possible very soon, and it is something that that that I've I've done some work with on glass. And you know, is I think it's something that is actually, to some degree, going to be very useful because you can imagine, you know, wouldn't it be great if if if you know, if you're somebody with Alzheimer's or you're blind and your wearable device could whisper in your ear the the name of the person that you're talking to or remind you things about them, or if you're a pharmaceutical sales rep or reminding, you know, then what the this poor psychiatrist wife's favorite flowers are or whatever, you know? OK, so let me let me show you a little bit of data because this is always good to go by. Example here. OK, so first, let me show you just. Well, with regard to this, this idea here, so this I'm not posting good for I have not implemented, I don't plan on implementing this idea, but it's also this is one of the frightening things. And I think one of the things that is that is potentially very creepy application. So you can imagine an application that, you know, as you're walking around or driving around. It just uses, you know, you know, ANP, an automated automated number plate recognition to pull license plates out of the pictures of what your, you know, out of what you're seeing, pass them out, which is not. It's kind of a solved problem like this is a lot easier than facial recognition because it's basically a combination of a d skewing transform and then digit and character recognition. So it's kind of like twenty six plus 10 classes,
 right? So it's pretty straightforward to classify these license plates. And and then, you know, uploading that number or license plate to a database with GPS location. So like, why would anybody do this? I have no idea why anyone would do this, but is this an example of when you have these, when you start to have always on wearable computers and are collecting massive amounts of data, these, you know, some some random person could it could create an application like this. It would be pretty straightforward. And let me show you an example of the kind of data you might expect to see. So you will that you'll be able to see very clearly that. Let's see here. OK. Oh, f 11. OK, so this is just me taking a picture every few seconds. You know, I'm revealing when I revealing, OK, I'm reading some article, you know, these are the things that when you do these applications because I've been, you know, I've been logging, I've been I've been kind of doing these experiments and you know, you realize, Oh God, I just logged everything I did on my phone by looking down. And, you know, I just, you know, revealed my password by typing it out on my computer, which is, which is, which is which is why I think, you know, I said earlier and I think that a lot of this data is going to move from centralized repositories to distributed repositories because I think that the Google style will upload every photo you take doesn't work when you're automatically logging things. And so but OK, let's see here now. You think, Oh, well, this is not very good resolution. Oh, OK, let's see here. OK? Can anybody read that license plate number? I can. Eight one nine nine two eight one. OK. This is somewhere in on the one on, one in in in in Silicon Valley. So this is this is an idea of like what malicious people could do. I'm not malicious, but I am, I guess, curious. And so, you know, we were just driving along. Oh, there's another one. OK, let's see here. I'm sorry. I feel bad. Actually, I don't. I don'
t feel bad for these people, but I do say that, you know, look at this. I mean, it's just the sheer amount of data that one can collect is frightening. And I think that this is something that that that, you know, I think it's interesting and we just don't have a framework for to, you know, to kind of talk about or deal with this from any kind of, you know, legal perspective because once again, these are all people on public, in public, on public roads. It's not clear that, you know there are some things with regards to driver's privacy. There's some there's some obscure driving policy privacy laws in the United States, but they don't actually apply in this case. So, you know, there's just those those are some of the. OK. Those are some of the things. OK, so let's talk about this law stuff. And and this is actually I think that some of the fun stuff for me. OK. Show of hands. Point of view with an Android development before. Hopefully a lot. Yeah. Good. Good. Good. Good. Good. OK. So you guys like this part? So let's talk about the code that makes this possible. So you're seeing, you know, I had to set up where it was taking a picture every few seconds so that those are those are larger ones. Let me show you. Let me show you a down sample, you know, because some downsampled stuff of of an entire day. OK. So you can see that. Yeah, OK. So this is just this is just me from the beginning of my day and just starting to take pictures of everything. Now, once again, I've downsized these images. Let me see if I can at least fullscreen that. Yeah. Well, fair, by the way, is also the best image viewer. First of all, let's be clear here, like how much faster than I have now? Anyway, so yeah, you can see the look. This is just a picture stream of my entire day. This this last this this this specific stream lasts from, I think, 11:00 a.m. to 11 p.m. or 12 a.m. And you're wondering, I'll tell you how that's possible and a little bit. But you know, this is basically the kind of da
ta that I'm collecting every single day. And so you can really kind of get an idea. And I know I'm going really fast here, but let me slow down some nice views, actually, you know, but the the idea is like, we're just collecting so much. We're capable of collecting so much data and know and a little bit. I'll talk about how we're capable of processing all this data. And you know, what does this mean for privacy? What does it mean when you know, I hold up my phone because I'm sending a text? Let's see what I what am I doing? I have downsides. It looks like we're getting directions somewhere. You know, there's there's a bunch of revealing information and we don't we don't even have the, you know, I think there may be credit cards in there. There's a bunch of horrible things like when you know you're buying something, Oh, look, you accidentally captured your credit card. OK, so let's go back to how this is possible, OK? Full screen. OK. So so there's a bunch of Android developers here. You're probably familiar with the Wake Lock API, so these are all the different if you go in the code and you can see this is the there's an enum somewhere and this is why these are the different options. There's partial week lock screen and wake like yada yada yada. There's even one proximity screen off work like which you may not know about because it's undocumented, which is like when you put your face next to an Android phone and it turns off the screen, that's undocumented. So what else here? Yeah, so basically, and this is this is the API example. OK. You can get pretty far with this and and, you know, with Google Glass, what you can do is using this, using this, this API, you can you can turn the screen off and you can continue for some time to continue to take images, right? So this is going back to that secret photo taker, but you'll realize after a short period of time that it gets killed and you may be wondering. But I said, Awake Lock, what's going on here? So there's there's
 this. There's there's a whole there's a whole slew of of of talk and discussion around the Android, the Android roguelike like system that occurred back like a few years ago. Let's see if I have it. Yeah. So, you know, people from Android, we're trying to get this power management system mainline in the kernel and people from the Linux kernel like, Get that out of here, get it. Don't bring that here. And you know, there is this huge controversy about like, how are we going to do? How are we going to manage our power on Android? How can we bring this back into the life kernel? There's there's these things called opportunistic. There's this concept of opportunistic suspend. Anyways, how you actually go about, you know, keeping the device running. But the screen off is by rooting it and then using this, this lower level power management API that I discussed through the JNI. Basically, you ever here wake down choir. OK, so what that does that calls the native active suspend blocker and you can trace the bottom links here, trace throughout the Android code base. And then that that that calls the native acquires is Suspend Blocker, which is the JNI call to acquire Awake Lock, which itself writes to a file descriptor. And this power away clock, which was that aforementioned controversy on, you know, with the Linux kernel and mainlining some stuff from Android, there is a whole hubbub about you can read if you if you're sick, you can read about the technical background of the Android Suspend blockers. And man, this poor guy, there's some poor guy from Sousse. I forgot Raymond. Anyways, I'll go to it later, but this poor guy had a like, Oh, OK, this poor guy had to like, fight for this one, and it probably shouldn't have been mainline anyways. So to quote Douglas McElroy, I don't know if he has ever seen this piece where Dawkins proposes Webb, which is this beautiful, literate programing thing, and he uses it to implement some kind of word count dictionary thing. And then D
ouglas McElroy does a very well, well, well articulated critique of the program and then writes a shell script at the very end that that does the exact same thing as Darken, his program, which was like a page or two of code. And the quote was the following Shell script was written on the spot and worked on the first try, which is that shell script you see down here, which does, you know, basically everything that this is doing once you have root access, of course. And so you basically, you know, right to this file descriptor way clock. It then allows you to continue to run your camera, logging software or whatever you want in the background. OK, so that's the some low level stuff that hopefully you found interesting. So the videos that you were seen before were basically they were they were recorded with an Android device. They were not recorded with a glass. I had recorded some similar videos and similar images friends using glass. But as I mentioned earlier, the power limitations are just too great. I mean, at the very best, you can get about an hour and a half doing that kind of one picture every four seconds. And so something else was required. And so one of the things that I hacked together was something that somehow I managed to make something that looks even doorkeeper than glass. It's called, it's called lambda hat. And so this is the device that I was actually logging that with. It is an Android phone that is that is strapped to a hat, but it has a much larger battery, $400 amp hour battery to be exact, and it lasts all day long. I mean, I can record for 16 hours straight, taking the picture every few seconds. And so that's kind of what I'm talking about and I mean the always on wearable computers. So this is always on, always logging, logging every single thing, all of the things. And it's frightening, to be honest. I've looked at some of this data when I first started and just thought to myself, My God, what have I done? This is a horrible idea. And I thi
nk in some ways it is a horrible idea so that, you know, this is this is basically what what the lamb that looks like when it's not as geeky. I mean, you can put the board behind the the the brand. This is one of the early prototypes you can put the board behind the top here. This is not this is interesting. I mean, have you guys seen the new Moto X and you know, it is always on and always listening to you? I find that creepy personally, and audio data is actually far, far more sensitive than video data. And within the United States, we have we have a legal framework that that represents that in terms of, for example, California is a two party consent state, meaning that lambda hat I couldn't turn the audio recording functions on without telling every single person I talked to. Hey, by the way, I'm recording this conversation. Hey, by the way, I'm recording this conversation, however, with video, it's not the same level of protection because it's not the same level of invasion. But, you know, so this is always recording. And, you know, I don't know. I have to see what Google actually does with the data. I don't know if they upload it, but you know, this is this is in my opinion, where things are going is always on. Not just press the button, it's always on, always listening. And Google has also stated this. So you know, we're going to talk about quickly processing the the data that we get. So there's a few different methods that have been used in the past. You guys have all probably heard of neural networks, and there's been a recent rehash of of all of this neural network stuff and basically taking a lot of the same neural networks from the 80s and saying, Oh, well, they're bigger and they're deep now. So basically, this is what the trend known as deep learning is, and you basically have these hierarchical neural networks that are there are stacked on top of each other that that are used to represent data. So at the very bottom. Lovell, you'll have like an edge att
ackers that if you ask that you feed it, structured data of let's see it from YouTube, for example, which is what it was done in in the well known cat detector paper, I don't know how many people have have have heard of the cat detector cat face detector paper. OK. So, yeah, so so. Google and Stanford collaborated on basically taking a bunch of data from YouTube, training a deep neural network to to to to recognize it to process this data. And, you know, they trained it with like, you know, Google's like just throw through all the resources at a 16000 core on a thousand machines for three days. You know, however, much of that cost. And it doesn't matter because they make one point four billion dollars every month in ad revenue profit. So, you know, basically they achieved state of the art results, and this is in 2011 what you probably haven't heard of, though. And you know, hopefully I would assume less than people have heard of. The cat detector thing is that some guys from University of Toronto actually reproduce and beat these results on two GPUs. Now, the train time was longer, but two GPUs as well within the financial financial resource, you know, well within the financial means of probably most people in this room. I'm sure you can borrow your friend's GPUs if you needed to. But and this is what I mean when the ability to process this data and the ability to to go through all of this data and tease out interesting things, whether that's faces, whether that's license plate, et cetera, you know, is increasing it and everybody is able to do that now. Not just large institutions and large organizations. And so, you know, to Google and Facebook, it's all about selling ads. And so what did Google do when these guys beat them out with the the GPU implementation while they bought? I mean, they acquired them. You know, so Jeff Hinton, now where is part? Well, so Jeff Indian was actually collaborating closely with one of the people who were at Google at the time. You kn
ow, Jeff, Dean and Andrew in the some answering has since left to go work on Coursera. But you know, they had they had basically their their deep learning. This rich thing called Google Brain and the Google brought on Jeff Hinton script, whose definition is basically, you know, Steve Mann's The Godfather of of wearable computing. Jeff Hinton is certainly the godfather of of neural networks. And so, you know, so what the Facebook Live face looks like? Well, I don't want to be left out of this, so let's hire. Let's hire John McEwan, who's the neural network guy out of NYU who's done a lot of work with them with this deep learning stuff as well. And as has had some really good results on, you know, it has been spearheading this for a really long time. So I don't know if you guys know this, but Mark Zuckerberg personally attended NIPS, which is like the largest AI and machine learning neural network conference and, you know, went to hire everybody else. He he was quoted. You know, he wants to build a theory of mind for every Facebook user. I don't know what that means, to be honest. I'm pretty sure I don't want it, though. It doesn't sound like it's something that I want. You know, and it's always hard to tell whether this is, you know, is this him like, oh, let's do some lofty goal there that will hire a bunch of people so we can keep them away from Google? Or is this actually like some master plan? It's it's never clear. But, you know, I don't know so. And once again, let's be clear, it's all about selling ads like Google makes $15 billion in profit every year from selling click ads like it's a big business, if they can, if they can get an extra few percentage points off of their click through rates and off of their ability to target, like, that's what this is, what it's all about. So let's be clear about what you know, why they're interested in this stuff, and you're thinking, Well, wait, should we be worried? And so this this is going out of the privacy part of thin
gs. You know, we have this new technology that enables us to gather, archive and process data about ourselves and others. So remember, I'm just like logging data like these poor people driving down the one on one like never before. So this is this is this has never happened before, right? OK, so maybe. Now, this is an interesting quote, so instantaneous photographs and newspaper enterprise have invaded the sacred precincts of private and domestic life, and numerous mechanical devices threaten to make good the prediction that what is whispered in the closet shall be proclaimed from the house tops. So this is a quote from a really amazing law review article called The Right to Privacy, which was published in 1890 by Warren and Brandeis. Brandeis went on to become a Supreme Court justice. And this the right to privacy kind of enshrined in in American law. This this concept of a right to privacy. And you know this this this is a really famous. And what was this in response to? So this was in response to instantaneous photography. OK, so around the same time, what had been invented? Why not the Kodak brownie camera? This first made available to everyone the ability to log images easily, relatively inexpensively and capture that moment for eternity. And so this was a whole new thing. It caused a lot of controversy back in the day 1890s. You know, about over 120 years ago. And and so that's the brand. So the question is, are these devices are these always on wearable computers that are logging and watching everything? Yeah, you can probably even throw up on their Facebook and these social networks, although they are they are different in that you voluntarily broadcast data, which you can voluntarily opt out of. In some ways. But you know, with these devices, if I'm walking on the street, I can't opt out of someone looking at me and then recording that, you know, there's just no way for me to do that right now. And you know, other than, you know, probably doing something ba
d. So, you know, are these neo brownies? That's the question that that I posed to everybody in this audience. I don't know the answer to that. I don't know how similar this is to me. There's definitely some parallels. There are other things that are not similar. And it's it's this kind of something to keep in mind. And so, yeah, so that concludes the talk portion, and I'd love to put some portion of time away for four questions. So hopefully we can get those started to. If you guys have. OK, so. We have four mikes. There are two and there are two more. So just line up there if you have questions. You can also use IAC and Twitter. So internet anything? OK, go ahead. OK. Tom, many questions. First one is, are there any countermeasures to be taken against possible violation of your privacy, for example, for example, scramble suits or government orchestrated daily license plate rotation or something? So, you know, it really depends on what you're trying to circumvent. A lot of these, these CMOS based cameras, you know, they have air filters, but I've seen examples of things where if you wear very bright air lights, you can scramble some of, you know, some some kind of off-the-shelf cameras. In terms of license plate stuff, I don't know what can be in the like you said, maybe it rotation or some. Yeah, it's really it's hard. Basically, each one of these solutions need to be needs to become a problem by problem basis to solve that specific problem. You know, some people have proposed these like ridiculous like LED lights that you wear in your face. I think that's I don't think that's the solution. Anything that someone's not going to wear isn't a solution. I don't have all the solutions for you. So but there are there exist some out there. I think some of the other things like the The Antivirale or Jones like make up stuff is a great marketing thing. I'm not sure if it's like realistic or useful. I mean, honestly, the state of the art of face recognition detection is such
 that like if you, you know, wear a hat and like, look down, you're going to be doing pretty well, just like you would see, you know, stopping someone from doing face recognition if they're a person standing up there. And if I was, you know, walking on the street going like this, it'd be a little bit hard to to to see me. So just kind of the same things that you do anyways, I guess. I don't know. I mean, what's the next question, Ethan? Number one, please? Um, since you pushed me a little bit at the beginning, do you think the uh, recording lady will uh to any better since, uh, three or four days ago? Um, a paper comes out about, uh, MacBooks, which, uh, film with you can read, right? So the ladies don't go on. If you, uh, turn on the camera. Hmm. Yeah. So that's that's that's a good point. I think there is a difference, too. I mean, right, it's always about like the difficulty of circumventing something, right? So I mean, obviously, I could disorder the LED right? And and so if I must go in and disorder the already, you know that that would be a good solution to not keeping the light on. Now, obviously, the worry is that, you know, at least with the MacBook thing, in my opinion, it's probably that this is done remotely. But yeah, so I mean. So to answer your question. No, no, the girls are that passionate. I don't know, you know, honestly, if it's as easy as rooting the device and like, that's all you need to do versus install fear, I still think that's probably an issue compared to it was to make the claim. Oh no, it's fine. The display comes on like a software thing is definitely very different than like some kind of semi obscure firmware. Firmware changes? Yes. Internet. OK, go ahead. Yeah, next question is, is Sean, end up on right now? Are you recording? No, I'm not. I'm not. I am not. So that's a good question. And the answer is no. This this this one is actually just a this is just a shell right now. It doesn't have a battery on. Number four, please. So assu
ming Google Glass, did you think that small battery due to weight concerns, how much does your lambda headway? The lambda hat weighs more because it has a bigger battery, but that weight is distributed evenly throughout my head, so it's actually far more comfortable and feels kind of just like a normal hot. Three. You mentioned the always on recording LSD and stuff like that. Do you think that it is a solution or can be part of the solution to have the vendors who build this stuff implement things like always on indicators for recording when on the same time, we usually don't want that the devices that we own behave outside our own control. For example, if I have a phone and then I think I should be the one to decide when the TV is on or off that I am on my laptop. The is only that I can't trust the software on it, but I think they're the the cure would be kind of difficult if you say that vendors should ensure this stuff. OK, so to be clear about, you know what my personal stances versus what, you know, an ideal solution is, I actually kind of agree. And you know, I did bash Google, but I I personally think that at the end of the day, if someone's going to circumvent it, they're going to like and they're going to do something malicious with it, they're going to be able to circumvent it. And like, if there's a hardware already, they'll be able to disorder and like, make it totally private. And I think that probably that is your right as the device owner. I personally, there's the only thing I hate more than people. Probably surreptitiously recording is people trying to tell me what I can and cannot do. And so, you know, if I bash Google, Google, I like Google, but I hate authority figures that tell us what we can and cannot do. And I think that's what Google has tried to be with with some of their devices. So I don't know if the answer to your question is kind of ranty. Sorry, but but yeah, no, I personally didn't. The number one raised that you have control over yo
ur device, even if that's secondary to the privacy of others. OK, internet, go ahead. The Internet would like to know, Oh, if you are selling your lambda arts or your plans to have a Kickstarter or something. So I'm actually after I, after I come back, I won't be flying out to to China after after this, and we're going to be working with some manufacturers there. So coming soon. We personally, I don't want to avoid the kind of like vapor where Kickstarter, which I'm sure a lot of you have invested in Kickstarter is that never actually ended up doing anything. And so we're going to actually try and make a make something before taking other people's money. No one. So as you've said, Google Glass looks a bit dorky and the usability is still awkward. Do you think it will gain traction in the mainstream in the next couple of years? Or maybe it is a novelty item, which is a problem that will solve itself? Hmm, that's that's a hard question. You know, in terms of like, I'm not a prophet, but I have some ideas about it. I have thought about it for a while. So one, if you guys remember back in the day, PDAs, those took off for a while. You know, there is like the The Palm Pilot five and you know, I had when I was a kid, I was really cool with my PDA and those took off well before smartphones hit the mass market. You know, there was the trio, right? The trio did everything an iPhone does. It just did it in a way that was kind of not right. And it wasn't until probably, you know, I for an Android hit the market that that smartphones took off. So I would probably argue that Google Glass is maybe more in the PDA stage than the iPhone stage. Although Google, right, wants it to be like the iPhone. I don't think it is. So novelty. OK. Internet and the next question from the internet is what about something like Tupperware lifo for storage so you can control your log data? So it was the first of all, I heard Tahoe laughs. Yes. Is that what you said? Yes. OK. Yeah. So that's by every
one else. Tahoe Lisa 30 File System is a distributed file system that's actually developed by a guy as you go who frequents the icebridge. My local hackerspace got a rep. And so I think that's interesting and I also think so. That's a really good statement. I think that the other interesting thing is what if not only was the storage distributed in a secure manner? What if the processing could also be distributed in a secure manner so you can get kind of like supercomputer like processing power over a bunch of different devices? I think that would be interesting, too. So, yes. Is anyone else in the room who wants to ask a question? Maybe you can't stand up and we will probably bring you a mic. OK, internet. OK? We still have a bunch of questions from the internet. The next one is. Due to such a device will one day be a central part of our life, and if you know if the glass life project that are completely open, so you have control of your data and if you don't know of any other such project, do you know what will prevent this project from existing? So. Well, I mean, I would like to say, you know, I like to say in general, like, for example, lambda hat is going to be open, but, you know, I don't know any other companies that have a super open mentality with regards to their wearable devices. Yes, I personally think that a wearable computer is going to be a central part of our lives in the future, just like your mobile devices. And then even mates, you know, supplant your mobile device or be your mobile device, I'm not exactly sure. But I do think that then an always on wearable device or always on computer that you keep on your body is always is going to be a central feature of your life in the future. OK, just continue. OK. Next question is, what are your thoughts on narrative clip previously? My motto assuming you know what? Yeah, yeah, OK. So I think it's interesting. It's microcontroller based. So what that means is that it's not really programable as easily as a 
system on a chip. I would I would have preferred the device to be system on a chip. Obviously, you get memory or not memory, but battery life improvements by making it a microcontroller based system. Interesting. I've never actually touched one before. I've looked at it. I think that, you know, the reality is that that's it's a life logging device, whereas I think that the really interesting things are the, you know, more general purpose computers. And so I think the the it to be interesting. I think you need to be able to access data and do it. Yep, just continue. Um, yeah, I cannot really make sense of this question, so I will just read them. Well, if you can make sense and sort of screwed up. So the similarities between bronies and Google Glass are quite obvious. Google Glass goes even further with processing that are automatically. What do you see the difference since the bronies were only capable of taking two pictures locally with drones? The I think that's the Oh brownies. Oh, brownies, OK, OK. I heard brownies, which is like male fans of the I you know, my Little Pony thing, OK? I was like, What? OK, brownies? Yeah, OK. So um yeah. So I mean, like I said, I mean, technology improves and yeah, sure. Is this like, is this like a brownie that broadcasts the internet? Yes. Yes it is. But that's, you know, we've had that for a while. So I still think that they're similar in in the debate that they're that they're bringing it up. And so, yeah, OK. OK. All right. I think we're probably no, no, we have ten more minutes left. I don't know why everybody's leaving. Number three, please, I have a question. You earlier said that Google just doesn't want to store all the pictures on their server, so they can there will be stored on our computers. And can you explain that again? Or more details about that, because I think Google is still interested in gathering all the information that Google can get. Oh, oh, I'm definitely not suggesting that Google's going to want to kee
p these things on your your your own device. I mean, I'm thinking that the users would probably want to end up keeping their life, you know, streams on their own device. Google is going to still probably go the Google Plus auto upload route. So, no, I wasn't suggesting that Google is going to want to promote doing distributed storage of data. They're about centralizing and exerting monopoly power over data. OK, thank you number for this, um, I would like to come back to the to the brownie aspect and parallel because at the time Kodak didn't want it to sell cameras, but the process and they had said You pushed a button, we do the rest. And you talked about the Google Glass as a capture device, not as a process one and not also as the restitution one. And what what are you interested in the institutions of the image? For example, when we see a face it, it can also distort the reality and present us an image that that isn't the reality we see, but the process one or anything else. So. What was the exact question? Yeah, no, the question was, um, you just took the question, you just have to your point of view on this aspect of the Google Glass. Yeah. OK, well, we don't see reality, but gotcha. Gotcha. Processes by Google. Yeah. So, yeah, so that's the augmented reality side of things, right? Google Glass is not an augmented reality device. It's a device that allows you to do push notifications, basically smaller updates you can't overlay on reality. And so that's not what Google Glass is. The battery life would be even worse if if, if it was the field of view on the display isn't large enough. And so it's not. Google Glass is not an augmented reality device, but yeah, so there are definitely some things that you can do to do processing on on glass. But it's definitely that's not its core principles. It's not. It's not to do augmented reality. And I think we're probably a little bit farther off from that augmented reality dream that you know we all love to see. Number thr
ee, OK, I'm with The Pirate Party of the Netherlands, I'm very interested in the legislative side of things regarding privacy, and I'm really interested to know from a guy like you, what would you propose for? What kind of legislation would you propose to limit privacy violations by these kinds of devices? Do you think legislative processes will even work to, you know, limit privacy violations? Yeah. So I think that the the the the one thing that that worries me more than the rogue person on the street surreptitiously taking photos and doing creepy things is draconian government regulation and policy just because guys like I can't even imagine that they would even be able to come up with some kind of something that's reasonable and implemented in a reasonable way and not have it be influenced by moneyed powers. I just I personally, you know, within the United States and, you know, we don't have a Pirate Party that's an active choice in the United States, unfortunately. But you know, I personally from the United States, I would worry about a legislative solution. Just because I've seen a lot of those and they always seem to cause nasty side effects. Yeah. If you were to propose a legislative solution just to just propose to, someone would ask you, literally, what should I put in this law? What would you say to limits of privacy violations? Hmm. Well, so there's two things, right, because there's the there's both the right to not be photographed, but there's also the right to photograph. So. So basically the you know, the fundamental question is, are you asking me, where do I personally draw the line? And you know, and then and then basically saying, you know, what does my opinion have to have to do with then? Like why? Why should that be projected out onto society? You know, I guess you're saying asking me to make that legislative decision. Honestly, that's that's. It's a very complicated thing to answer, and I I don't feel comfortable coming up with like a legislati
ve solution on the spot. No, I just wanted to know what a guy like you will hatch Android devices might say. OK, so you know, if I'm just going to choose one? Let me. I'll just choose if people are going to do something like that like this. It's relatively easy to record people on the street and you know, some people will choose to do that. And if you know, I would say, let's not make it illegal to record people on the street. So I would say I would go more towards the photographer's rights than the person walking along on the street because there are there exist countermeasures you can. For example, you know where a you know where some kind of mask or a hoodie or a hat? You know, there are things that you can do versus, you know, there's I guess there's you know, there's etiquette as well as from the photographer's perspective, which we have today anyways. I just in general would say, let's kind of let's kind of try and put this into the format that we currently have, which is let's have good etiquette and good, you know, things that are not like legally enforced. I think we already have pretty good solutions to these problems. OK, thank you. Yeah. OK, we have time for one last question from the internet. OK. Just a note. First is apparently one of the people in there. See us worked with Steve Mann and is rolling in idea. Steve Man was developing this. And when you say the Google Glass are probably not the right device for this kind of thing, but you could have a real life outlook like, you know, looking billboards in the streets, replacing them with nice painting or some of the. That's a nice idea, I think. And the actual question is completely unrelated. What is the state of the bootloader? You said there was a vulnerability that allows in looking at, uh, or does that work? Oh, yeah, OK. Yeah. So what I was referencing was Cedric has a really wonderful write up on using a security vulnerability to to route your Google Glass. You can. So it comes OEM unlock the bo
otloader comes OEM unlock so you can just do fastboot OEM, unlock your fastboot unlock and it'll unlock the device. However, as Sark points out in his article, If you decide to lock that device later, it doesn't. The flag doesn't get set to lock to get set to relax, so you know you could be violating your warranty. Not to mention, it's a bad idea to keep like an unlocked bootloader, you know, because someone could just like plug something in re flash operating system. From a security perspective, it's a bad idea. But, you know, so there's both the the the the Google sanctioned route, which is fancy fastboot OEM unlock and there's the Vault security vulnerability route, which is just using a security vulnerability to gain access it. And so I don't know the answers, the question, but that's the state of the bootloader. OK, so we're out of time, please. Thanks, Stephen. Thank you, guys. Thank you.