 The phone lab is a large platform, well, not necessarily huge, but a platform, smartphone platform testbed that we built here at the University of Buffalo. So the problem or sort of an opportunity was that the smartphone platform that runs on your device. So this smartphone, I can install apps on it and installing apps on it can actually be a fantastic way to run smartphone experiments. But there's a lot of software that runs below the app that provides the Android platform itself that provides the sort of Android services that apps use. And that software is impossible for apps to modify. And that software is also, I would argue, very important to the mobile systems community. It does all sorts of things that our community spends time doing research on. For example, computing the device's location, trying to manage energy, figuring out how to use multiple wireless networks, et cetera, et cetera. And so the ability to modify that platform or run experiments on it is something that I think is useful to the mobile systems community. And so we set out to try to provide this. We were aided in this quest by great cooperation from Sprint. So Sprint has been able to give us a discounted price for participants in the project. At this point, we have about 80 students, faculty and staff at the University of Buffalo, including me and my wife, who carry these phone lab smartphones. And this phone lab smartphone has an instrumented platform on it. And it can also have a platform that contains experimental features. So the phone operates as normal except there's a lot of data, depending on what my students are working on and what other researchers are working on that's being collected about what the phone is doing. That data gets offloaded when the phone is plugged in, and then provided to experimenters in order for them to do analysis. To make the project safe, experimenters are required, in many cases, to file IRB requests from their own institution, ensuring that they're going to handle the data safely. That's a process that universities in the United States have that ensures that human subjects review procedures are being followed and the data is going to be handled safely. So anytime we get out anything that's the least bit sensitive, we make sure that people do that so that we know that our participants data is safe. For experimenters, the process of using phone lab involves essentially getting our platform, our phone lab experimental platform, making your changes, and then merging your changes back into the platform so we can deploy your changes to our participants. My student, Jinghao, she has been managing this process for years. She's gotten very good at it. He's excellent at working with experimenters. And so if you want to run a phone lab experiment, the test bed will be open until the middle of February, at which point it will shut down. So if you would like to run an experiment, now's the time. I'm not going to argue that the process of modifying the Android platform is easy because it is not. But there are certain things that you can only find out by modifying the platform. So for example, my student, Scott Hasley, is doing a project where we're trying to automatically detect various types of waiting indicators. Doing that requires grabbing information very low in the device's graphic stack. And there's no way that you could do that as a map. So from a participant perspective, we have 80 participants. Those participants pay us money. They don't pay Sprint directly. We collect money from them. We pay Sprint. And so we have to do billing. We have to do participant support. I have a fantastic undergraduate working for me named Ed Santos, who's in charge of helping participants with billing problems, with device malfunctions, replacing devices that get broken. Essentially, we are this little mini phone company. And so Sprint doesn't even know who our participants are. They don't have their names. They don't have their billing information. We maintain all that. And we're also a point of contact for any service that the device needs. So essentially, again, we're almost operating in our own little telephone company here, maintaining our own software platform, providing support to our participants, and collecting money from them and paying Sprint. So phone line has been a really fun experience. I'm not sure I would build a smartphone platform testbed in the same way again. But I do think that as a research community, and we need to think about how do we maintain access to this code base. There are millions of lines of code that make up the Android platform. And those lines of code do really important things. And if you can't experiment with that part of the system, you're missing out on the ability to understand how to improve billions of smartphones all over the world.