Quantum Leaps and Windows Updates
juin 5, 2019
Take a fascinating dive into Microsoft’s approach to quantum computing. Then, get an inside look at recent improvements to the Windows Update experience. Machine learning and increased transparency are helping the Windows team track user feedback and deliver our best update experience ever.
Windows Insider Podcast Episode 20
JASON HOWARD: Welcome to the Windows Insider podcast, where leaders at Microsoft and Windows Insiders discuss the latest in tech trends, careers, and innovation. I'm your host, Jason Howard, and you're listening to Episode 20: Quantum Leaps and Windows Updates.
If you're a longtime listener of the Windows Insider Podcast, you'll recall that we did a brief bit on quantum computing in Season 1. From medical breakthroughs to finance and even to cryptography, quantum computing stands to transform the way we solve our world's most complex problems. Get ready to get entangled once again with some truly mind-bending concepts with our special guest, Janet Schneider, here from Microsoft's quantum computing team.
Then, for the second half of our episode, we chat with Microsoft's Chris Morrissey about big changes to the Windows Update experience. Chris also gives us a peek into how the Windows team is evolving the way it gathers data and feedback, with the goal of continually improving Windows and putting users first.
Alright, let's get onto the show!
JASON HOWARD: Welcome to the show, Janet. Can you please introduce yourself to our audience?
JANET SCHNEIDER: Thank you very much. I'm Janet Schneider. I'm a Principal Software Engineer in the Cryogenic Computing Group at Microsoft.
JASON HOWARD: I have to tell you, I am super excited to dig into talking about this space that you're working on, you know, the whole realm of quantum and cryogenic computing. It might sound a bit like science fiction but they're very real, even here today. So to get the listeners oriented a little bit, can you tell us what exactly is cryogenic computing?
JANET SCHNEIDER: Cryogenic computing is computing at very, very cold temperatures. So if you think about, just to kind of give you a sense of the scale here, like, the room that we're in right now is about 280 degrees kelvin. And if you want to do computing for quantum, you have to be in, like, the millikelvin realm. So to get there, deep space is, like, three kelvin. The dark side of the moon is 26 Kelvin. Liquid nitrogen is 77 Kelvin. So, it's at those very, very low temperatures where it's a different type of computing than regular CMOS at room temperature.
JASON HOWARD: That's kind of funny because like zero degrees Celsius is pretty cold.
JANET SCHNEIDER: Right.
JASON HOWARD: Zero degrees Fahrenheit, yeah, it's cold. Zero degrees kelvin, nobody can exist.
JANET SCHNEIDER: There's no -- there's no energy whatsoever. Right.
JASON HOWARD: Yeah. So, quantum computing, right?
JANET SCHNEIDER: Yes, yes.
JASON HOWARD: We did an episode last year where we kind of delved into the broad concept of it, right?
JANET SCHNEIDER: Right.
JASON HOWARD: But we didn't get super deep. We talked about the field in general, kind of where the field was headed.
JANET SCHNEIDER: Right.
JASON HOWARD: But, like, getting kind of behind the curtains a little bit, what exactly is quantum computing? Like, how is it different than what we think of as computing today?
JANET SCHNEIDER: Right. So computing today is all digital-based, meaning it's all based on ones and zeroes. We encode information and we process the information as ones and zeroes. Quantum computing marries really three disciplines: information technology, computer science, and quantum physics, right?
So, you want to use quantum physics to encode the information and to do the computation. And you want to use quantum physics, quantum mechanical behavior to actually be able to run algorithms to solve problems.
JASON HOWARD: So, that's a far cry from, you know, silicon and whatnot processing a pattern of ones and zeroes.
JANET SCHNEIDER: Right, right.
JASON HOWARD: Far shot from that.
JANET SCHNEIDER: Yes, yes.
JASON HOWARD: So knowing the differences between, you know, looking at ones and zeros and staring at a stream and looking at the outcomes, right?
JANET SCHNEIDER: Right.
JASON HOWARD: How is quantum computing going to change the world as we know it?
JANET SCHNEIDER: So, quantum computing allows you to do computations that you can't do today. It allows you to model systems that we can't model today.
So, one of the classic problems they look at for quantum computing is, you know, factoring large numbers for cryptography, for instance. If we want to factor a 260-bit integer using Shor's algorithm, you'd need to model two to the 260 bits of information, which is more than there are atoms in the universe.
So, it's not possible to solve that problem in the digital realm. You have to actually model it using quantum mechanical systems that allow you to, by their very nature, model the states and the possibilities.
JASON HOWARD: So even if we took every computer that exists on the planet today, and obviously there's a ton of them --
JANET SCHNEIDER: Right.
JASON HOWARD: -- and we put them all towards working on one unified goal of solving that problem, we wouldn't get an answer.
JANET SCHNEIDER: That's right. You'd have to take all the matter in the universe towards solving that problem, and you wouldn't be able to solve it.
JASON HOWARD: I'm pausing for a moment because my brain trying to wrap around that is just -- I'm having a bit of a difficulty here. (Laughter.)
So obviously, knowing that there are some hard problems in the world to solve, what are some of the potential applications that once we kind of get over the hump and, like, quantum computing becomes a more prolific thing, like, what are some of the real-world applications that will be exciting that we're going to go out and try to tackle?
JANET SCHNEIDER: Right. It's not going to be the typical problems we solve today. It's going to be really in three realms, like material science and chemistry where you want to model an actual physical system and all of the states that are involved, like all of the energy state of an electron when you're forming a molecule, for instance, or how materials interact with each other.
Another one is machine learning. Looks like it has some promise. And a lot of the research they've done today shows that you can get significant advancements if you have a lot of quantum RAM, which means we'd have to scale way farther than we are today, like, with thousands and thousands of qubits to model all data in a superposition, right, and do machine learning on it.
JASON HOWARD: Wow. So you mentioned the term qubit a moment ago. So as you get to breaking down the building blocks of what actually represents quantum computing, can you walk us through the key terminology that will help us wrap our minds around it a little bit, like what exactly is a qubit?
JANET SCHNEIDER: Right. Okay, so I'm going to start with what's a bit, and that's a physical entity that can represent a one or a zero, right?
JASON HOWARD: All right.
JANET SCHNEIDER: And that's what we compute with today. And that's, you know, could be a lightbulb that you turn on and off. It could be a lever of some kind, right? In the computers that we have sitting here in front of us, it's the presence or absence of a voltage on a very, very tiny little bit of wire.
A qubit encodes more information, so it represents -- it's a logical concept. And what it is, it's representing a quantum mechanical system that has some quantum mechanical properties, like it can be in a superposition, meaning that it can have two possibilities at once, right? And you don't know what it is until you measure it.
And this is like the analogy in quantum mechanics is photons, right? So, you could have a photon that's in a superposition and you don't know until you measure it whether or not it's oriented in a vertical polarity like a dolphin, or horizontal like a snake, right? You just don't know what it is yet. That's important because that's what allows us to model all these possibilities.
The second is you have to be able to manipulate the probability of a particular outcome. So that's where we can apply gates to a qubit that say, "Okay, I want to rotate the possibilities leaning towards one way or the other."
But the really interesting piece comes in where you want to be able to entangle them. So from quantum mechanics, we know that when you have entangled particles, their fates are tied together. If you measure one, the states collapse for the other one and it only has one outcome possible. So, you want to be able to model that in some physical system called a qubit.
JASON HOWARD: That's fascinating. So, it seems like this is completely different than, like, say, flipping a coin, right, because if you flip a coin, you got heads or tails, right?
JANET SCHNEIDER: Right.
JASON HOWARD: And if you happen to take a picture of the coin midair while it's flipping, that's not going to affect the actual outcome of what the coin's going to do, right? You can catch a picture and it may be heads or tails when you took the picture, but when it lands, taking the photo didn't affect the outcome of what the coin was actually doing.
JANET SCHNEIDER: Right.
JASON HOWARD: But when you get into the quantum realm, if you take a measurement, you've totally collapsed what was actually going to happen at the end. You've affected what the potential result was going to be.
JANET SCHNEIDER: Measurement changes the system, yes. And that's the key challenge with building a scalable qubit.
JASON HOWARD: So, a bunch of big tech companies and even nation-states are working on this particular field at the moment.
When looking at Microsoft in particular, like, what is Microsoft's strategy in this particular space, and how are we doing it differently than other companies?
JANET SCHNEIDER: The biggest thing that we're doing differently is that we are investing in a topological qubit. So a lot of competitors are looking at superconducting qubits, ion traps. People have built qubits out of large optical systems that are as big as this room which have high fidelity, but, you know, that doesn't scale very well. So, we're trying to build something that can scale and have high fidelity.
So, the topological piece of it means that it, you know, a qubit is not an electron. It's not a particular particle. It's a system that has quantum mechanical behavior. So our system is topologically protected.
If you can imagine, you have a table with, like, a rubber band sitting in it, and that rubber band has to always touch the table. You can move that rubber band. Like the universe can observe the rubber band and move it around, but the actual state of what it is, like the shape of it, stays the same.
JASON HOWARD: Okay.
JANET SCHNEIDER: And so, that's what protects it, because the biggest challenge with qubits is you set up this great superposition and you want to do a bunch of operations on it, and then you want to be the one measuring it. But if anything else in the universe observes that state, you've broken the spell. It's called decoherence, and you've nullified your computation. So this system of topological qubits protects the qubit itself from observation.
JASON HOWARD: (Laughter.) It's just, like, I chuckle, right, not because this isn't a serious topic but because I find it utterly fascinating. And of course, over the series of doing podcast and webcasts, and things for the Insiders that are listening, like, there's always something that I learn. And I have come to find out in talking through this topic with you, and with others along the way, just how much I don't know.
JANET SCHNEIDER: (Laughter.) Yeah.
JASON HOWARD: And I guess that's my natural response is, you know, I just chuckle when I realize how much I don't know because it's mind-blowing and fascinating all at the same time.
JANET SCHNEIDER: There's a couple other aspects, too.
JASON HOWARD: Okay.
JANET SCHNEIDER: So, it's not just the topological qubit. You know, Microsoft is investing in this full stack scalable system, and part of that means if you have a bunch of qubits and you want them as cold as possible, so you want as little energy in the system as possible so that you can very carefully control the state.
So when you have a whole bunch of qubits, you know, running at millikelvin, you want to be able to control, read out, and do error correction on those qubits. You want to have something running at a temperature that's not so far away, because otherwise, you have to pipe all those signals up to room temperature, which is challenging, just logistically. And amplifying the signals up to room temperature is also challenging, aside from just having all the wires sticking out of the system.
So, that's what I'm working on right now is that cryogenic system that could be the qubit control, read out, and error correction.
JASON HOWARD: Wow. It sounds, at least from what you just said, that in the process of trying to read the data, is it possible to accidentally change the results while you're trying to get the measurement of what the results were?
JANET SCHNEIDER: It's more that you want to be able to have, like, real-time control of the qubit.
JASON HOWARD: Okay.
JANET SCHNEIDER: You don't want to have it travel too far and have to amplify and change the signal. You want it to be in the same temperature domain.
JASON HOWARD: Gotcha. Okay.
JANET SCHNEIDER: And so the third thing that's really setting Microsoft apart from the rest is the language and developer kit that they've created. So, Microsoft has the Microsoft Quantum Development Kit that includes the Q# language, which is built to scale, the compiler and the libraries, and a series of samples called the Katas, which are designed to teach people about these concepts.
Because we really want to grow the ecosystem of people who understand what quantum computing is and they start developing algorithms, and they can run them on a simulator, and they can analyze the algorithms for how they perform so that they'll know when we have the scalable computer, we have, you know, the ecosystem built out and the partners so that we can, you know, really make use of the quantum computer.
JASON HOWARD: So knowing how difficult a space quantum computing is, right, there's obviously kind of I don't want to say a finite list, but there's a list of things that make this a really tough space to work in, right?
JANET SCHNEIDER: Right.
JASON HOWARD: So, what are some of the key challenges being faced by people who are developing technologies, currently?
JANET SCHNEIDER: I think as far as the qubits go, it's how do you create the long coherence time and high fidelity. That's, like, a key issue that people are working on. Being able to talk to them at scale, because when you're dealing with cryogenic systems, like, you know, they're colder than the dark side of the moon. And so, you know it's not a very hospitable environment to debug issues when you're dealing with a cryogenic system.
And then, we just really want to engage people to learn more about quantum computing. We want people to be developing algorithms to understand the space, because, you know, there's years of development and computer science, you know, looking at, okay, I can ignore what the physical system is. I know it's ones and zeroes. I can, you know, develop these algorithms for solving tough problems. We need to be investing in solving the tough problems in the quantum realm.
JASON HOWARD: So looking at your personal journey, right?
JANET SCHNEIDER: Yes, yeah.
JASON HOWARD: How did you get into the entire, like this quantum space? And more specifically, how did you get on Microsoft's quantum team?
JANET SCHNEIDER: Yeah, so, I've been following quantum computing for a while, just because it's interesting. You read about the different advances in quantum mechanics and the things that people are discovering every day with solid state physics. It's just fascinating to read about.
And so, I knew that we were investing in quantum computing. And them, I saw about a year and a half ago, we released the Microsoft Quantum Development Kit. And I asked my mentor, "Hey, can you give me an introduction?" And I just came knocking and I met people and I just started asking question after question. And they're very, very brilliant people, very kind, great teachers.
And so, I spent a lot of time with the team. And then, when there was an opportunity that came up, you know, I applied and interviewed, and -- and here I am.
JASON HOWARD: Wow. So, your particular background.
JANET SCHNEIDER: Yes.
JASON HOWARD: What did you do before joining the team? Like, what is your degree in? Like, what did you study?
JANET SCHNEIDER: Yeah. My degree is in computer science and math.
JASON HOWARD: Okay.
JANET SCHNEIDER: And I've worked in low-level systems. But as far as quantum mechanics, and quantum physics, and quantum computing, I had no background. So, I just started reading. Like, I grabbed this book, Quantum Physics: What Everyone Needs to Know, you know, and Quantum Computing: A Gentle Introduction. I just tried to read up on it.
And the -- the website that Microsoft has, has a lot of resources. You know, if you go to Microsoft.com/quantum, we have the development kit, we have the blogs, we have all kinds of documentation that teach you about the key concepts. So --
JASON HOWARD: It's been quite a journey, then.
JANET SCHNEIDER: Yes, it has.
JASON HOWARD: So now that you are in this space, what are you working on personally? Can you give us a peek into what's happening on your side of the lab, so to speak? Like, obviously, no top-secret stuff, right?
JANET SCHNEIDER: Right.
JASON HOWARD: You know, we're not trying to give away key competitive advantages or whatnot, but what is it that you're working on?
JANET SCHNEIDER: So right now, I am working on this cryogenic system that will be used to control qubits. And so, most of that is pretty top secret about what we're developing there.
But I'm actually switching roles very shortly. In another week, I'm going to be joining the Quantum Software team, and I will be working with a team that brought you Q#, the compiler. They're investing in the runtime for scaling out qubits. So, it's all very exciting.
JASON HOWARD: So, it sounds like a shift from the environment in which it's being manipulated into the software that you're using to try to engage within this field.
JANET SCHNEIDER: Right, yes.
JASON HOWARD: That's a pretty big change.
JANET SCHNEIDER: Yes, it is.
JASON HOWARD: Are you excited?
JANET SCHNEIDER: I'm very excited.
JASON HOWARD: Yeah. So obviously, you've done some incredible work relating to the Windows Insider program, right?
JANET SCHNEIDER: Yeah.
JASON HOWARD: Kind of stepping -- you know, stepping away from the quantum piece just for a moment here. So you've mentored entrepreneurs through the Insiders4Good program that we have. So in relation to the Insider program, what has your experience been like? Like, what are your -- the mentees that you had, you know, worked with previously, what are they up to now?
JANET SCHNEIDER: The Windows Insiders4Good East Africa Fellowship was an amazing experience. I got to meet some incredible people.
So, I have a few updates from my -- from my mentees I'm very excited about. So, Irving Amukasa, who brought you the SophieBot, which is the chatbot, where people can ask questions about sexual health, you know, he's done a lot of amazing work with AI.
And recently, he was a contributor to a paper that we released from Microsoft called AI for Africa.
And I sent these links ahead of time so we can include them in the podcast.
He also did a side project to detect Kenyan sign language to help out a friend of his. And he has a Medium post on that, so you should check that out.
JASON HOWARD: Wow.
JANET SCHNEIDER: He's done a lot of really great work, yeah. Super proud of him.
And Rosine Mwiseneza, she is still working on SmAgri, which is a company that will -- they want to automate a lot of the farming that they do.
JASON HOWARD: Okay.
JANET SCHNEIDER: And so, she has an article that she posted on why smart irrigation is key to increasing production of food. So, she's continuing on that effort and we're super proud of her.
And then, we have Ange Uwambajimana. She wanted to solve the problem of providing low-cost IV pumps that would detect, you know, when they're empty so that you could increase patient safety in hospitals in Rwanda.
So, she has been shopping this around to hospitals in Rwanda. So, you know, I think she's doing really well there. She's continuing.
JASON HOWARD: It sounds like an entire group of bright upstarts that, you know, started off with an idea and have spun each idea into something that's much bigger than what they came up with originally. And I mean, obviously, there's no stopping the future for them.
JANET SCHNEIDER: Yes, that is true.
JASON HOWARD: That's amazing.
JANET SCHNEIDER: And then, there's Alfred Ongere, and he has started a developer community in Kenya called AI Kenya. And I included a link for one of their Meetups in December where they talk about all things AI. And recently, at the Nairobi World AI show, they were voted the best developer community. So, we're all very proud of them for that recognition.
JASON HOWARD: Wow. So as we come close to wrapping up here, I have another question that I want to ask you, of course.
JANET SCHNEIDER: Okay.
JASON HOWARD: There are listeners out there, some who may never have heard of quantum before, some of whom may have gotten bit by the bug, kind of like I have, right, where I'm trying to read up and learn more. And you know, I've got a book and I'm going to order the other book that you mentioned earlier, which we'll include links so everybody can go out and find them, if they're interested in.
So for any listeners out there that are fascinated and want to learn more about quantum, what resources are out there, right? You mentioned Microsoft.com/quantum earlier. Is there any place that you would recommend they start or check out first?
JANET SCHNEIDER: I think checking out the Katas is the best place to start, because that is designed to teach people about quantum computing. And it will introduce the concepts along the way as you work through the Katas. You know, how do I create a qubit? How do I measure it? How do it put it in superposition? How do I entangle it? How do I manipulate the possibilities for that qubit?
And then, it moves on into known algorithms that have been shown to have a known speed-up in the quantum realm, like Grover's algorithm. So, you can actually go through some real algorithms that have been developed.
The other resource is -- that I included was brilliant.org has an online class on quantum computing based on the Microsoft Quantum Development Kit. So, that's another great resource to work though if people want to learn more.
JASON HOWARD: All right.
JANET SCHNEIDER: Yeah.
JASON HOWARD: So as I mentioned earlier, we will of course include all these links at the trailing end of the podcast, but I will also ensure that, on our website, which is where we host the podcast, which is aka.ms/wippodcasts, make sure and include all these links so folks can go there and have a place to go reference this as well.
JANET SCHNEIDER: Great.
JASON HOWARD: So, I'm going to ask you one more question.
JANET SCHNEIDER: Okay.
JASON HOWARD: What is the most interesting thing you've learned along the way, which I guess that's kind of not a fair question, given how much fascinating stuff there is to learn in this space, but has there been anything that's really just been kind of a -- just a wow moment for you?
JANET SCHNEIDER: Yeah, that's an interesting question. I like that it's this big, unsolved problem, and every single aspect of it is interesting. I don't think I can pick a favorite interesting problem. It's all interesting. You're trying to, you know, build something that you can manipulate quantum mechanically, right? That's interesting.
You're trying to represent everything mathematically with these different states. That's interesting, all these notations and algorithms. All the algorithms that you're trying to write are interesting.
The language development side of it is interesting. How do you actually write a new programming paradigm to represent a new type of computation?
Like, every last bit of it is. Quantum computing is interesting. That's my answer. It's like in Hidden Figures when one of the characters walks in and they've got that lunar module, you know, mounted in the room and she's just standing there. Like, that scene gave me chills because that's kind of what it's like, like you're standing there like, "Oh my God, that's a spacecraft." You know, it's -- it's kind of like that.
JASON HOWARD: That's really the magnitude of what all of this is going to represent.
JANET SCHNEIDER: I think so, yeah. Yeah.
JASON HOWARD: You're trying to do another moon landing.
JANET SCHNEIDER: Exactly, yes.
JASON HOWARD: Just on a really, really tiny scale.
JANET SCHNEIDER: Very, very cold, yeah. Colder than the moon landing. Yeah.
JASON HOWARD: (Laughter.) Well, I have to say, like obviously for me, and hopefully for everybody listening, a completely fascinating conversation. I obviously have a ton more homework to do, so hopefully you and I get to stay in touch. And I'll try not to randomize you with too many crazy questions along the way.
JANET SCHNEIDER: Definitely. Definitely. It's been fun chatting.
JASON HOWARD: Awesome. Thank you again for your time. Thank you for being here in the studio with us.
JANET SCHNEIDER: Thank you.
JASON HOWARD: Next up, we dive into a conversation about improving Windows Update experience and how Microsoft is making new investments in measuring and improving the quality of Windows. Our guest in the studio today is Chris Morrissey. Chris, welcome to the show.
CHRIS MORRISSEY: Thank you so much.
JASON HOWARD: Can you please introduce yourself briefly to our audience and share a few words about what you do here at Microsoft?
CHRIS MORRISSEY: Absolutely. Well, I'm Chris Morrissey. I lead the COSINE Servicing Delivery team, and what I lead over there is the communications aspect of the team. So, what we do is we're responsible for the communication for rolling out all the Windows updates, both the monthly ones and the feature updates.
JASON HOWARD: Wow. And so CSD is kind of what we lovingly refer to you as. So, any Insiders who have seen us mention CSD on Twitter or on the answers forum, this is the group that we're referencing here.
So, getting to the update process for Windows, can you walk us through Microsoft's approach to the Windows 10 Update experience, namely where we've been and where we're headed?
CHRIS MORRISSEY: Sure. So, in the past with Windows 10, our update experience was basically we would offer that starting with, as you know, the Windows Insider Program. And then we'd shift to our what we call phased or controlled launch program. So, we'd start off with, say, our seekers, those that want to seek the Windows Update, and then we'd start rolling out the launch to more people as we felt the experience would be good for them.
What we've done recently is introduced machine learning into that model. And what that's done is it allows us to look at more variables and better understand which device is going to have a better experience so we can make sure that our users have the best update experience possible. And then we can keep tweaking the product to make sure it's just fit for purpose here.
Lately, though, we've even taken another approach, which is to add more control for users. And you'll see in our new release, the May 2019 update, we are offering our users way more control on when they want to update and take that update.
JASON HOWARD: So, you mentioned this, the whole seeker concept. So, looking backwards in history, if we go to, say, XP or Vista or even 7, possibly even Windows 8, was the seeker experience part of this process at that point or was this something that was introduced during the Windows 10 timeframe?
CHRIS MORRISSEY: This is more a Windows 10 timeframe paradigm, that there was a little bit of that before, but we've really made it more official, so it's actually in the UI. The user can go there, seek the update, and go get it.
And we've seen some good experience with that, because, therefore, when we haven't been ready to roll out the update to the full universe of the install base, those more advanced users, let's say, that know what they're doing, that they want to go try the new bits, to see the new features, to see the new functionality, they're able to do so.
JASON HOWARD: So, we have millions of Windows Insiders and other users giving us feedback, and of course exciting things are happening in the feedback realm that will be super impactful in the product.
Mike Fortin and his team, which you are part of, has recently posted a series of blogs that are very transparent and talk about all the quality variables that our engineers are looking at, mainly in regards to how feedback is sorted and prioritized. Can you talk a little bit about that?
CHRIS MORRISSEY: Absolutely. User feedback is very critical for us. It's one of the key signals that we use besides telemetry and some other things and our own self-hosting and the Windows Insider program.
What we've been doing with user feedback is trying to sort through it to make it more usable and get it to the right engineers as fast as we can. And as Mike discussed in that series of blogs, which started last November and runs up very currently to the beginning of April, is all the different elements of programs that we use, from drivers and how we launch those and our driver ship room process, how we look at app compat.
One of the key things that we're looking at feedback, though, is when users give us feedback, and you're a user, I'm a user, let's say, how you describe an issue may be different than I describe it. So, how do we group those together so the engineers can get something they can act upon that has utility to them?
And, secondly, a big one is how do we find those very severe issues but are very small in volume. And we refer to those, the scorpion in the haystack. (Laughter.) We want to make sure we find those, because those could be quite egregious.
So, we've taken a new model here. We keep improving our feedback program. And you might have seen Mike announce back in the fall, we now have a program so a user, when they give feedback, can also say, "Hey, this is super severe for me," so we put a severity rating on it.
Secondly, we're using machine learning to identify those. With all the hundreds of thousands of bits of feedback we get per week, we have to find a way to sort through those, categorize those, and get them to the right engineering teams. And that's what we've been focusing on with this latest release. So we're pretty excited to see how that's going to work out, and our own testing shows it's going to be good.
JASON HOWARD: That is definitely one of the interesting things about the Insider Program, both whether it's, you know, the beginning, middle, or end of any development cycle, is trying to find the feedback that, as you mentioned, is very severe in impact, but doesn't necessarily have a lot of upvotes or it isn't something that a lot of people have hit because they either haven't worked through the particular scenario that's going to trigger the particular bug or experience, right?
Or just generally, when teams are doing their prioritization work, right, it just happens to be somewhere in that long tail, as opposed to being this, you know, blinking red light that's, like, hey, look at me over here, there's a lot of people hitting this issue.
CHRIS MORRISSEY: Exactly, that is one of the biggest challenges we face.
JASON HOWARD: So, really, what's shiny and new about some of this expanded process, right? Like, obviously, there's kind of a multi-faceted approach of the way that COSINE as an organization and, obviously, Microsoft as a whole is trying to approach what they're doing and how it relates to customers and using these products. Can you talk a little bit about that?
CHRIS MORRISSEY: Absolutely. One of the things we're trying to do is bring AI and ML more into the process. As I mentioned, we're using that to find these severe but yet small-volume issues.
We're also doing that with the rollout process. And it's not just how we roll out devices looking at the different attributes, you know, the functionality, what type of processor does it have, the drivers on the device, et cetera, so we can kind of tag and target those devices.
It gives us feedback to understand what combinations of specs and configurations cause issues, and then we're able to do that. So, in other words, our machine learning model, we've been using it for the last three or four releases, is getting more sophisticated and it's learning and it's getting smarter. And with that, we're then able to understand trends and then the engineers can actually work on those as well to root cause them and solve them.
So, it's number one, our modeling is getting more sophisticated and our ability, then, to do the engineering around that to mitigate ahead of time is getting more sophisticated. So, it's really interesting stuff. And we never had the tools to do this before, and now we've got those tools.
JASON HOWARD: It sounds fun, but it sounds like it's a bit ironic that you're using computers to tell you which computers are going to be the best computers on the next build.
CHRIS MORRISSEY: Yeah, as long as they're not operating us, we're going to be okay. (Laughter.)
JASON HOWARD: So, as we're talking here, it sounds like there's two main puzzle pieces to the whole concept of quality, the listening and data-gathering aspect of it, and then obviously on the back side, once you have all this data, there's a need to take action on it. Which, of course, involves analyzing, prioritizing, so on and so forth, with the feedback that's coming in. Can you highlight some of the changes that we've made in these processes?
CHRIS MORRISSEY: Absolutely. As we get feedback in, as like I said, we're sorting through all that, but even on the outbound aspect, what we notice is customers want to have a nice smooth update experience. I mean, they don't want to lose their device for, you know, 10, 15, 20 minutes, maybe an hour while it's updating, they want to have this great experience. They also don't want to have incompatibility problems going forward.
So, we do a lot of work with our ecosystem partners, for instance, ISVs. There are certain classes of software, anti-malware, for instance, which are very sophisticated. We have to work with those partners to make sure they're approaching things the correct way, keeping their software updated along with ours, so that we have this great together type of experience.
Same thing with drivers with our different partners that are making hardware, things that are related that require a driver. That's a big piece.
And then the last piece I'd say that is for the user themselves is being more transparent so they can understand where we are in the process and what's going on.
I'll give you one example there. If we find working with a given class of partner, let's just say it has to do with a certain piece of hardware and a driver, we know that driver hasn't been updated and there could be incompatibility issues that would provide a bad experience or, you know, or even worse, we will, then, what we call "pause" the update. In the past, we called it a block, but let's just say we're pausing the update to protect them so they can't update at that time until we can go remediate that situation. Could be action we take, could be action the partner takes, could be something we do together.
And we're trying to now be more transparent so the user says, "Ah, I was trying to update, the update's not available to me right now, and here's why. I understand that, yes, I have that type of peripheral. Ah, I have that driver. Yeah, I don't want to be in that situation.” And then we will keep folks updated on the status of that so then they can understand, hey, they can come back and get updated another time.
JASON HOWARD: Well, interestingly enough to that point, one of the things we talked about in a webcast earlier this year with some folks from the deployment team and some of their peer teams was if a user had hit one of these, say, issues that required a pause, right, “Hey, we're not going to deliver the update to you yet,” the messaging that the end user would see was a bit generic, just says, “Hey, your machine's not ready or something to that general effect, right?” It's kind of changed a little bit over the course of time.
But now it's a bit more prescriptive, right? It's not going in depth and show you file paths or anything, but it's saying there's a known compatibility issue with this driver, this hardware, something a bit more specific, so that the user can potentially like let's just say there's a problem with an application, right, there's a known compatibility issue with an application. It'll actually say before we send this update to you, there's a compatibility update with application whatever it may be.
And the user can make a decision at that point to say, “Hey, this isn't something I use very often, I'm okay not using it for a little while.” potentially remove it from their device, and then actually take the update at that point.
Or, if they're going to say “This is extremely important, this is part of my daily personal business, whatever routine, I can't do without it.” Hey, it's actually a good point to say, well, almost like saying, “Thanks, Microsoft, I appreciate you guarding my experience, I'm going to keep doing what I'm doing, and then once it's ready to go, once an update's been provided, I'll take the update and move forward from there.”
CHRIS MORRISSEY: Absolutely.
JASON HOWARD: When you mentioned the seeker experience earlier, let's talk a little bit more about that, where a user was potentially going into the OS to look for, I don't know, a Defender update or potential driver updates or something else, and the next thing you know there's this big update that shows up on their device and you're like, well, that's not necessarily what I was looking for, right?
So, this is one of the things that you all have actually implemented kind of a separating feature in, if you can talk a little bit about that.
CHRIS MORRISSEY: Yeah, no, that was a big point of feedback that we've gotten from both Insiders, our seeker community, and then just influencers at large, interested. And if you think about it, like, when you go up to an ATM machine and you check your balance, you don't expect it to spit $20 at you. (Laughter.) That would be nice.
So, what we've done actually is we've gone back and looked and said, okay, if you're going in there to check for updates, what you're really looking for is information. And when you see the information, you can see what updates are available.
To your point, it could be a driver update, it could be a security update. It could be a feature update. We now have made it and broken those two things into two pieces and now you can see what's available, and then you as a user can choose which update you want to take. So, if you want to take that feature update, you see it's available and you can now make that decision. You can now download and install. And what's even cool there is you can download it and we'll even give you more in choices of when to install.
The whole point here is to make sure that you have your device when you need it to use it as you want. So, you're in control, you use your device as you need it. And we give you the information. That's the transparency aspect.
So, what we've kind of been talking about here is control and transparency, and you need both. Like we were saying before, you want to know if we pause something because there might be an incompatibility. If we don't tell you what that is, which software it relates to or which driver, hard for you to take action, hard for you to make an informed decision. And that's what this is really all about.
JASON HOWARD: Yeah, sharing that level of information, not only does it garner trust from users because they have a better understanding of what's happening, it helps demystify the process a little bit so that they're not kind of sitting there wondering, what's actually happening here, did I do something wrong, is there something wrong with my hardware, right? It's just filling in those little gaps like that seems to make quite a difference.
CHRIS MORRISSEY: Absolutely. And, in fact, it's a journey that we're on. In fact, the team I'm on, we lead what we're calling the transparency for the win. And our approach here is to continue to grow in our ability to be transparent. When I say ability, we also have to have the mechanisms that scale to the size of the Windows platform.
We are doing things at a massive, massive scale. I'll give you a couple numbers. We're now at over 800 million Windows 10 active monthly users. And when you add on the down-level versions of 7 and 8 and everything else, you can see the numbers get humongous. And when we're updating, for instance, at scale with monthly updates, we're exceeding 1,000 devices per second on a global basis.
JASON HOWARD: Excuse me?
CHRIS MORRISSEY: A thousand devices per second.
JASON HOWARD: Per second?
CHRIS MORRISSEY: Per second. So, when you're updating at scale with a machine like this, you've got to have an ability to communicate with people to be transparent.
So, one of the things we're doing is a couple of factors, a couple of different methodologies. One is developing a better outward-facing tool. So, you might have heard of our new public-facing health dashboard that we just launched this last week for down-level, so it's 18 version, Windows 10 1809 and before.
And when we come up and we're ready to make the new version, the May 2019 update available, we'll obviously have that content on there, and then we're going to get that, the news of that out there a little bit further. But people can go check and play with this now.
We are posting on there, for instance, all the known issues, whether they're still open, the status that they're at, all the dates that go along with that, all the KB numbers in one place where you can find everything, and for the last six months, all the remediated or fixed issues.
So, what we're doing there is giving people a really easy one-stop shop to find everything. In the past, we knew it wasn't optimal. We think this is the right direction, and we have plans to continue to invest in this area, and that could be another podcast in the future, but we are looking at always both in-product, out of product, and on these types of surfaces like these dashboards, to keep our customers, to keep our users informed and in control.
JASON HOWARD: As a side note to our listeners who want to check out the health dashboard that Chris just mentioned, here's a link that you can follow: aka.ms/WindowsReleaseHealth
So, we've talked about Mike Fortin's blogs a few times here. I want to give listeners a quick link that they can reference. So, if you want to go back and read some of the historical or current blogs, they'll be able to do so. So, for those of you listening, I'm going to give you a quick link. Here we go: aka.ms/fortinblogs. And I'll spell that out real quick: F-O-R-T-I-N-B-L-O-G-S.
So, grab that link in your spare time after you're done listening to the podcast. Or if you're on a PC right now, you can pull it up in a separate window.
But let me transition just a little bit, because one of the things you just mentioned was data, right? And in the current environment of news cycles and everything going on in the world of tech, data's a very hot-button, hot-topic-type issue. Users want to know what is being collected, what's being gathered, what's being done with it, so on and so forth.
So, for Insiders out there, especially who are really into this type of stuff, you know, there's -- there's always an appetite for as much transparency as we can offer. And it's -- obviously, we're on the path to do that to the best of our ability within, you know, government regulations, so on, and so forth.
So, for anybody who's looking for that transparent view into how we capture data and turn it into something actionable that we can actually improve Windows with, there's one of the blogs that, you know, was just referenced is there's a specific one titled Data, Insights, and Listening to Improve the Customer Experience, right?
So, quick plug for that one in particular. Definitely go check it out. There's actual screenshots of real information that we've looked at, it's not redacted, which is super interesting, given Microsoft's history of kind of holding data and information close to the chest, but it's a new world, right? With Satya at the helm and making customers a top priority, we're doing things a bit differently.
So, please do take a moment, check that out, go read it. The article walks through the complexity of capturing the data, which you know, if people are looking for updates at 1,000 a second, can you imagine, you know, the amount of data that we're trying to work through to say, hey, these are compatibility issues, these are driver issues, these types of machines are having problems taking updates.
You talked about scale earlier, but I think we're going to need to invent a new word, because "scale" isn't necessarily going to cover it at this point.
CHRIS MORRISSEY: That's how we feel sometimes. But I would encourage the listeners here to actually go back and look at that blog series. We worked on that with Mike, and there's a lot of great information there. And to your point, we really do kind of open up the curtains to explain how our processes work in great detail with, like you say, a lot of screenshots and prescriptive definitions that help the user kind of see what we do and why we do it and the outcome of it and that there really are some very robust processes that we utilize and lever.
JASON HOWARD: There's an entire group of folks out there, and interestingly enough, I love giving my mom plugs, she asks me random questions all the time because she's like, oh, you work at Microsoft, I'm going to ask you questions.
And (laughter) it's funny, right? On-demand tech support, so to speak. She'll call and ask me questions and, "Oh, hey, I noticed this changed." And interestingly enough, if I rewind two or three years ago, she wouldn't ask me stuff like that. But it seems like now that we're being more transparent and showing folks things, it's not done out of, I don't understand what you're doing, it's almost like driving innate curiosity, where we're bringing people from being just a consumer of Windows to hopefully converting them into being fans of Windows.
CHRIS MORRISSEY: Yeah, absolutely. I think it's also people becoming more familiar and comfortable with tech. And the more we can break things down, the more we can explain things, the more we give them information, of course you're going to be more comfortable. If you don't understand how something works, what's a black box, it's kind of like voodoo, it's hard to get around, it's hard to like something, it's hard to feel comfortable with it.
When you can kind of understand it, it makes a big difference. And that's why we say we're -- it's a multi-pronged attack to -- or approach, I should say, not an attack -- to get people comfortable and give them the information they need where they're using something and when they're using it and the appropriate format.
And that's what we're really looking at now is how do you get the information to that person in the right format? And the other thing is, make the update experience, make the whole process as seamless as possible. That's our North Star, seamless.
JASON HOWARD: So, I know we've talked about feedback repeatedly through this discussion, right? And there's both short-term tactics that are being implemented, many of which you've talked about, and of course there's going to be some longer-term things that will change, whether it's an idea that's kind of in the planning stage now or something that we haven't thought of yet, which that's the way things go in tech, right, there's always something new around the corner.
But, hey, given all of this, I'm going to ask you my favorite question, right? What tidbits can you talk about that your team might have up its sleeve as, you know, we move forward, right? Nothing top secret, you know, I'm not trying to get you to, you know, give us the super-secret recipes here, but maybe a teaser or two of something we can talk about.
CHRIS MORRISSEY: Absolutely. Well, we are big fans of the Windows Insider Program, and we'd love to come back on your show here. This is excellent. We've also been benchmarking what you guys have been doing. So, I think you'll see us doing a little bit more outbound engagement, let's just say, through non-blog, through non-traditional channels. So, I think that's something we're looking at.
And in the spirit of transparency, this journey that we're on, I think you'll see us trying different ways, different techniques to get some thoughts across about what we're doing and also just what's going on with the product to make people more comfortable and hopefully injecting a little levity and humor. And, like I say, using some very different channels.
So, stay tuned. I think by mid-summer, some of this might manifest and that might be a great time to come back on and explain what we did, why we did it, and where we're going next.
JASON HOWARD: Injecting humor, it sounds right up my alley. (Laughter.)
CHRIS MORRISSEY: There we go.
JASON HOWARD: So, as we wrap up here, any parting insights or other advice you want to offer to our listeners?
CHRIS MORRISSEY: I would say one thing is, first of all, thanks for being Windows fans. Thanks for, you know, listening to this podcast. What I would say, though, is as we come out with these new updates, new products, whether it's monthly or feature updates, we really do appreciate and need your feedback. So, continue to give us feedback. Any suggestions as well.
And one thing I'd add is, and maybe we can add it in later on is, we're interested also in what areas are you interested to hear more communications on, if there's different subjects, different areas, certain parts of the update experience, could be anything. We are very interested to hear what folks want to hear, rather than us just trying to think in a vacuum. And that helps us prioritize and then figure out what's the right channel to get that message out.
JASON HOWARD: Awesome. Well, Chris, thank you so much for being with us. Again, if any of the listeners want to check out the blog post that we mentioned several times over, aka.ms/fortinblogs, F-O-R-T-I-N-B-L-O-G-S, and then of course send us feedback and, you know, we'll take it from there.
CHRIS MORRISSEY: Awesome, thanks again.
JASON HOWARD: Thank you so much, Chris.
JASON HOWARD: That's a wrap for Episode 20: Quantum Leaps and Windows Updates. If you liked this episode, you were born to be a Windows Insider. If you're not yet a Windows Insider, come and join the fun. It's free and easy to register at insider.windows.com. Not only do you get access to Windows features that haven't yet been released to the public, but you'll also get access to exclusive contests, giveaways and other opportunities to grow your skills.
As we close out this episode, I'd like to share one more note for the Windows Insiders out there. Having just finished Microsoft's annual Build Conference for 2019, there is a ton of great content and recorded sessions now available for viewing. If you'd like to check this content out, visit aka.ms/Build2019. That's 2-0-1-9. You can watch Satya's keynote, view featured talks, and navigate all other available on-demand sessions.
And with that, I'm your host, Jason Howard. Catch you next month for another episode of the Windows Insider Podcast.
NARRATION: The Windows Insider Podcast is hosted by Jason Howard and produced by Microsoft Production Studios and the Windows Insider team, which includes Allison Shields, that's me, Michelle Paison, and Kristie Wang. Listen to our previous podcasts and visit us on the web at insider.windows.com. Follow us @windowsinsider on Instagram and Twitter.
Support for the Windows Insider Podcast comes from Microsoft, empowering every person and every organization on the planet to achieve more.
Join us next month for another fascinating inside look into Microsoft, tech innovations, careers, and the evolution of Windows 10.