[NOISE]. In this video we're going to see an interview with Dr. Adam Aviv. He's an Assistant Professor at the U.S. Naval Academy and an expert on usable security. He's also the author of the paper we read about smudge attacks. So I purchased a Google phone when they first came out, and I was very excited about it. And I put on this password pattern. And it was awesome. And then I started waking up every morning and I would look at my phone, and I would see my password on the phone, just embedded there in the smudge. And I asked myself, well, I wonder how, common this is, and under what conditions this would occur. And so I went to my adviser Matt Blaze, and he was like that's awesome, let's do it. And so, we bought a bunch of photographic equipment and we took a bunch of measurements, and we found that yes, these things seem to be quite common. And, using really off-the-shelf software, like anything that can, affect the levels of the contrast. Images where you didn't even think a smudge was present, they actually have quite a, like you said, amount of information in it. Even partial information could actually be used to recreate, your password. Yeah, so, I think the, the most surprising thing to me was that, at the time, we had two types of technology for touch screens. And if you think way back for the G1 phone it was a plastic face versus the the new Nexus models which you know, are the phones that eventually took over were a glass face. One of the things we wondered was whether or not there'd be a difference between the two form factors. And what we found actually is that the glass face was actually worse. So we're moving into a domain where glass would actually cause more more problems, than than the plastic surface that was prominent at the time. Yeah, so that's, so that's a really interesting question. Fact it's a question I get asked often so, if you look at the literature in the graphical password, gesture-based password space, they always make this promise that they are going to be more secure that text-based passwords. They're going to offer more diversity in the choices that you can make. And that they're going to be more memorable because people are visual things, and they can remember a image or a gesture or some sort of factor better than they can remember some random text, right? And this is going to be the future, and this is going to be great. And if you look at what actually happens in the technology, you find that no, these things don't seem to be used at all which was really surprising for me when I was looking into this. That we actually did build a gesture-based system on Android, and the result of that is that they didn't think it all the way through in terms of actually how we're going to use that, use that gesture on the device itself. And like figuring out how that interaction actually can degrade the security of the system, was totally not really considered at the time of the implementation. And if you go even further down and actually study these things, maybe the gesture-based password on Android is not a good example, but it turns out to be actually less diverse than, a lot of other systems that could have been put in place, which means there are smaller sets of patterns available. And even further research now suggests, with my own and some of my colleagues, that people often choose terribly on these things, as terribly, maybe even more terribly, than they do in text-based passwords. So, it's a really interesting challenge, because on the one hand, history is right. The researchers are right that gesture-based, all gesture-based passwords and graphical passwords are much better for humans, for memorability, for a lot of other things, but if you look at it actually how it gets implemented it doesn't always work out the way we planned. So if you actually look at what's happening, out there right now we, we're not doing this, and it's, I think it's more interesting to ask why we haven't done this while we clearly can. For example using our fingerprint biometric on iPhones is a really common thing on our laptops to unlock the current device that's in front of us. But these things are not necessarily used for remote authentication. One of the reasons is that most of the biometric stuff that we have out there is not actually that secure. People have shown that you can use play dough to thwart fingerprint analysis, and actually doing this as a secure way of logging into sites just doesn't seem to have latched on in the community of people who have developed this. And I think that's probably because they haven't seen it as a actual secure choice the way we want to. That hasn't stopped researchers like myself to think about these problems and what would that mean, and we try to think about it all the time. But in the end if we can't come up with something that's actually strongly secure based on devices that are commodity that are available for everyone to use, it's never going to really catch on.
