 Good afternoon. I'm here to talk about the history of the web from its humble beginnings in the early 1990s through the high and low points in the browser wars to today. And a lot of this talk for me personally is nostalgia. I hope there's a lot of see what impact history had on today's web. As George Santana said, those who cannot remember the past are condemned to repeat it. Let's definitely not do that. I'm going to talk a lot about Internet Explorer's history. You'll find out why in a second. But I still believe a lot of this drove getting us where we are today. But much like Mel Brooks's movie I have to skip a lot. So the first deck I pulled together for today I think was about an hour and a half long. Hopefully I'll make it through and land on time. So hi I'm Chris Wilson. I've been working on the web platform a super long time. I started over 26 years ago. I co-wrote the Windows version of NCSA Mosaic, first real mass market browser after after Tim Berners-Lease browser. I've been an engineer at PM, a lead PM, a GPM, a developer advocate. I'm currently on the TPGM team in Chrome. I've worked on four different browsers I think. And I've been a Googler since 2010. By the way that's since Chrome version 6. Chrome version 6 shipped a week after I started. Don't take this the wrong way. That sounds super impressive. I don't think I'm the smartest person in the room. Certainly not in this room. I have been in the room where it happened to borrow a Hamilton quote for a second. Many many times over the years which is pretty cool. But so I want you all to cast your minds back for some of you before you were born to 1988. 1988 is when I went to the University of Illinois as an undergraduate. And to give you a picture of the Internet at the time, this chart is great. You can't really read the the labels but 1988 is here basically. And you'll note it almost doesn't exist. No one had really heard of the Internet at that point. In fact it was still commonly known as ARPANET in academic circles. And my incoming undergraduate class was not issued email accounts by default. I actually had to go to the computer lab, ask for one in person. And in fact we logged into a mainframe like people didn't usually have the ability to connect. And you had a time limit of seven hours a week on this mainframe system. And I would usually burn this up playing ultimate larn because you only had seven hours. I'd usually burn it up in the first couple days. And then I couldn't check my email again until Saturday morning. Thankfully engineering students got a second account that was not time restricted on a different mainframe. So clearly I did pass my computer science classes. This was my first internet terminal. It was a cast-off terminal from a job that I was working at over the summer as an HP 2622A terminal. It had an integrated thermal printer in the top. It was awesome. I connected this to the University of Illinois system via a smoking fast for the time 2400 BOD modem. For reference I calculated my cell phone gets something like 6,000 times faster connection over LTE and a 60,000 times faster connection when I'm on Wi-Fi at home. Or it's about twice the speed of my hotel Wi-Fi here in Mountain View. But this was just a text terminal. Like the internet was mostly text. In fact that's one of the points I made here is it really was email. It was Usenet. It was IRC. It was Gopher. FTP was cool too but frankly used a lot less except for sharing software. Now don't worry my entire presentation is not just Douglas Adam quotes. There are a couple more I think. But this is really this is something to remember all the time. Anything that's in the world when you're born or when you're very young is just normal. It's the way the world works. I have two daughters. The internet is the way the world works. They don't even think about it. Mobile devices? Absolutely. Anything that's invented between when you're 15 and 35 is new, exciting, and you can probably get a career in it. In the internet the web was definitely invented between when I was 15 and 35. Now I'm not really here to say get off my lawn you young whippersnappers. The web's been done. Everything you can come up with is against the natural order of things. The challenge is really how do you take it and make it so that you can adapt to the natural order of things. Think about what it can become in the future because otherwise I turned 35 before Chrome was released. So that would be a problem. In 1991 I got a part-time job working at the National Center for Supercomputing Applications. This is the last time I'm going to spell that out. The NCSA is one of the five original NSF supercomputing centers and our group's mission inside NCSA was to build software tools to enable researchers to use the power of the super supercomputers on their desktop. So we did Telnet. We literally built Telnet. In fact that's what I was hired to do, work on the terminal. Now keep in mind this was before Windows 95. Whoops. This was before Windows 95. There was no TCP-IP network stack. PC Telnet was DOS software. We had to eventually build Windows software. We also did visualization tools and things like that, data loading and formatting tools, that kind of thing. But this fit right in when we had this project come up where a couple guys came up with this X-Windows software called Mosaic that let you access this thing that some guy in Switzerland had come up with called the World Wide Web, used a prototype document language called HTML. That was kind of, well that was the next 26 years of my life. I co-wrote the Windows version of this. Incidentally, NCSA Mosaic had three versions, Mac, PC, Windows and X-Windows. They were totally different. They did not share code at all. We looked at each other's code, but cross-platform software, not really a thing. I also, this utterly ruined the word Mosaic for me. If you say the word Mosaic, I will not think of a piece of art. But keep in mind again, we didn't have network support in Windows. Windows didn't have a built-in network driver. Like they didn't have sockets. You could buy a separate package that went at it. Even Windows 3.11 with networking. I can't remember the whole name. Windows for Work Groups, I think. It didn't have TCP-IP built in. Didn't have sockets. It had networking, but through Novalp or something else. So we already had our own network stack. We actually literally wrote the device drivers for network cards. Like I would debug this stuff. I debugged the TCP-IP fallback algorithm at one point. And so we got that working in Windows. We built the HTML parsing and rendering stuff, all that kind of thing. And NCSA had this tradition of shipping lots and lots of beta releases. Like basically build it on your machine, put up a release. I think I actually put up several debug version releases because I couldn't find a bug once that turned out to be pointer initialization, of course. But we put Windows Mosaic up on the FTP server for the first time. And I remember we got super excited because a week later we were at 1,000 downloads. Now I gave this talk I think in 2011. And I calculated like how long it took Firefox to get 1,000 downloads. And it was measured in fractions of a second. So a little different scale. But still, what was really surprising was the game DOOM came out the same year and we still managed to get work done on Mosaic. Because we were college kids. I mean, we were students. NCSA had this great overall charter of, hey, go enable education and research. And we got to explore a lot of things. And if we didn't have that there, we wouldn't have gotten to go build a web browser. It wasn't in our software tool groups charter at the time. We were chartered to build, you know, image visualization tools, Talmud, et cetera. And this is one of the things, one of the parallels that I draw with Google. Because Google has a lot of that flexibility. It's one thing that I really like about working here. Now, but overall, like, we made a bunch of foolish mistakes in Mosaic because we didn't really know what we were doing. We didn't really have a whole lot of support. We didn't have a business plan. And plenty of dot coms didn't have business plans too. But we didn't really like, there was no user experience class in my computer science degree. There was no, you know, theory of software testing or anything like that. And we didn't even have revision control systems. Like there were two of us working on Windows Mosaic. We didn't have network drives. So literally, we would like work on the same code. We'd save it to a zip disk, walk across the office, stick it in the other guy's drive, copy the right files, hope for the best. So we did make some foolish choices. And one in particular that I personally own is this. I'm personally responsible for overlapping bold and italic tags working in the early web. For some reason, I thought this would be cool. The rampant violation of all common sense and SGML structured document decency. And by the way, yes, we do still support this in Chrome. And every other browser probably. We do it in an interesting way that doesn't violate everything that's decent. But in my defense, such as it is, I thought it would be cool at the time. But the early web, even the really early web, because remember, I'm in like 93 ish here. It was really compelling. And the reason it was compelling was that you got access to all this information. I mean, Google's mission is organize the world's information and make it universally accessible and useful. And I was the one chanting in the back when Paris asked this question earlier, because it's in my deck a couple times. And you have to remember at this day and age, the world's information wasn't on the internet. The internet was like we're off at the low part of that ramp, right? A lot of that information wasn't even in digital form yet. Like you had to go to a library to get information. Now eventually, of course, like anybody who's watched a bunch of information before knows how we tend to think of the web platform, you'll recognize part of the slice acronym here. But at this point, it wasn't indexable really, or it was indexable, but there were no indices to it. In fact, I meant to bring along in my on my bookshelf, I have this book, like a bound paper book that is literally lists of URLs and descriptions of them. It was published, I think in 93 93 or 94, I can't remember which and it's just a solid book filled with, Hey, this is what you can find on the Honolulu Community College website, or the like. So most critically at NCSA, we had this, we recognize that the coolest thing about this web thing was that everyone could be an author because most of the systems out there weren't really designed for anybody to put servers up and share information go for in particular, like go for clients were easily accepted, accessible, go for server software you had to pan arm and a leg for like basically only universities ran it or private enterprises because it was so expensive to run the software. Now, so we actually implemented a bunch of stuff at NCSA to try to encourage this, this sharing of information. We actually had even in 1993, we implemented a group annotation feature, like you could annotate web pages, and then anybody in your group who shared the same annotation server, they would see your annotations when they went to the same web pages. We also had this idea that the web server software should be free. And everyone should run a server on their machine. Now, putting this in context, this was several generations of massive security problems away from where we are today. We actually I ran a web server on my PC in my office at NCSA. It was world accessible. Literally anybody could just FTP into my personal. In fact, we wrote a network space dog fighting game me and a group of my friends at the university and we borrowed the NCSA PC telnet code to do this. And after we've been playing it, like we got it up on running, it was like awesome, those early land parties was great. After we had it, I've been running for a while, I realized that I never disabled the FTP server in the PC telnet code. So sure enough, I could like FTP into my friend's machine, while he was playing the game, and you know, start a big download. He didn't really like that. But the cool thing was there was lots of wacky content, like this was groundbreaking at the time because you didn't have access to this content. Like there was a guy named Kevin Hughes from the Honolulu Community College, who they have a really cool dinosaur museum at this Community College in Honolulu. And so he thought, Oh, this is awesome. I'll take some pictures of it with my digital camera, and put it online. And he did. And it was like, you would just get this weird kind of stuff. And this is more than anything, I think what contributed to the growth of the web was this concept of the long tail. Who has not heard of the long tail before? Just to Okay, there's so there's a few, which is good. This is where the large portion of content out there, the overall population is in this tail. And unfortunately, the colors are messed up. But like, all of the stuff out here, measured by interest is just as popular as the most popular things. So another way to look at this is the web is incredible, because people get super interested in uncommon topics, more so even sometimes than, you know, Taylor Swift's latest video, which is great, by the way. This is taken for granted today. Like this is one of those things where you're not in your head and saying, Yeah, whatever. This was new back then, because you couldn't really get access to the long tail, not in the early 90s. In fact, in the early 90s or mid 90s, actually, 1996, I took a class learning to play the didgeridoo part of the class was how to build one. I was like, Oh, this is really cool. I put up a web page how to build a bamboo didgeridoo. It's pretty easy, by the way, just poke a hole in it, more or less. But how many of you remember geosities? Right on nostalgia buffs, I love it. I put up a geosities site on how to build didgeridoos from pieces of bamboo, complete with ASCII art, because I didn't own a digital camera yet. This, this was seven years before Canon came out with a digital rebel, just for reference. So it really wasn't that popular. And I put my Microsoft email address on it. And yes, that seems like a crazy thing to do today. But I did. And this page stayed up until geosities shut down in 2009. And that entire time for 13 years, at least once a month, I would get an email from somebody saying, Hey, this is great. I was wondering if you know where to get raw bamboo here in Finland. By the way, this page is still accessible. If you really want to see it, the URL is in the speaker notes for this page, but I'm not going to go to it now because I'm a little ashamed of the artwork there. But I tell this story not because I think you're all interested in building bamboo didgeridoos, although I'm happy to talk about that. Even fewer of you are actually interested in, you know, the finer grain details of where to get raw bamboo here in California. I do know, because I did live here for a little while. But you probably have some interests that are just as uncommon. In fact, by contrast, for the last few years, I started making ukuleles in my spare time. And I buy my materials and my tools online. I watch videos on YouTube of how to bend the sides, right? Or that thing I compare notes on forums about different web wood choices, et cetera, browse through exotic woods on eBay. This is a thing like Ossert's ukulele wood on eBay. And it's not like five pages. And it's like hundreds of results. And I'm showing my age because I find this amazing, like this is mind blowing the access to information that you have. This is a massively new natural order. And this leads me to something that I've said for decades about my career is that the web and the internet just helps human interactions happen faster. And fewer guardrails, fewer speed bumps in the way. I used to think, by the way, this was an unreservedly good thing. I don't think that anymore. So this is kind of foreshadowing. Be careful about how you enable human interactions. But it does definitely remove geographic barriers and information barriers. So anyhow, back in 1994, the core mosaic team all pretty much left NCSA, nearly at the same time. Mark Andre has left the previous year. I quit and headed out to Seattle to take a different job. And then Mark came back with Jim Clark scooped up the rest of the engineering team before they could leave and go off to the winds. Because they saw that I was the first one that left, but I was not going to be the last. But I went to this company called Sprye. They built a product called internet in a box. I literally have one of these on my bookshelf too. Yes, the entire internet in a box. But it was kind of cool, because this was the whole package for what you needed to connect to the internet, which again, at the time was pretty cool. It had a Usenet browser. It had an email system. It had a gopher client. It also had a web browser. And it still wasn't clear that the web was the winner, by the way, it had all these. Like when I was building this talk, I like, you know, did the command tab on my Mac and it's like, literally the only things running are two copies of the browser and the calculator. And that's just because it was easier to get to the calculator by clicking the icon. Now, I don't want to talk a whole lot about Sprye in particular, because they they went away for good reason. Although it's not Spyglass, by the way, Spyglass was a totally different company. And in fact, a year later, I left Sprye and I joined Microsoft and the Internet Explorer team. Spyglass was actually this company that NCSA licensed their code to, because they realized that the business of software licensing was not really the business NCSA wanted to be in. And then Spyglass reselled it to everyone else. And Microsoft actually licensed it from Spyglass. And Spyglass said they rewrote all the code from the ground up. I'm just saying a lot of that code looked really familiar. And a lot of it, the architectural choices. Yeah, it had some of the same idiosyncrasies as code that I had written, not that I was proud of necessarily, but was the same. But I joined the IE team right before we shipped IE 2.0 for context. And I'm sure a few of you have probably heard this quote before. Mark Andreessen was quoted as saying, Netscape will reduce windows to a set of poorly debugged device drivers. Now, in his defense, by the way, Mark actually said that Bob Metcalf said this, and he was just retweeting it. Well, quoting it, because yeah, JavaScript wasn't around, Twitter wasn't around. This may seem like a cheap shot against Microsoft. It actually isn't because I didn't believe it then. I continue to work for Microsoft for 15 years. I don't even believe it now because those device drivers are really well debugged at this point. But I don't think any of us expected what this statement kind of foreshadows that the world is just going to become the web platform and software sitting on top of it, not a whole bunch of software sitting on top of windows. And this really did kick off the browser wars. And my favorite example of the hideousness of the browser wars was the blink tag. And how many people have heard the tech, the tale of how the blink tag was invented. Wow. Okay, so this one was pretty good. I was not in the middle of this one. I was not in the room where this one happened. The blink tag was created because some Netscape engineers went out drinking one night. I'm not kidding. It's on Wikipedia. Look it up. And one of them, or more of them were like, Oh, man, we should make a tag that makes everything blank. And they were out drinking. So one of them decides to go back to the office. And the next morning shows their coworkers, the blink tag. Now, and Microsoft, we hated the blink tag. We're like, Oh, my God, why would you make them? Why would you do this? Like you're gonna, I don't know, trigger people who are sensitive to blinking lights or something. This is horrible. In fact, when the Netscape team demanded that CSS include text decoration blink, I said, fine. We'll parse it, we'll apply it, but we can't be forced to make it actually blink. It was in the CSS one spec at least, you're not actually required to visually make the thing blink. Now, we still had to out blink them now. And that's what the marquee tag was about. And the marquee tag. Just search for marquee tag in Google search. It's actually awesome because it has now my favorite Easter egg in Google search, because it actually replicates a marquee tag like the tag results start scrolling back and forth. And that's basically what marquee did. It scrolled text back and forth. The joke actually ended up being on us on the IE team, because we also made it scroll vertically and stack text vertically, like a marquee, you know, when you walk down Times Square or something, and there's vertical text. And apparently this is really popular in Korea, because it matched how their text and fonts worked at the time and it was useful in developing Korean text. And we literally couldn't remove it for like decade, more than decade afterward, because it was so heavily in use and we would have broken a lot of content. Now, in 1996, as when we shipped IE three, so this is kind of the error we're in. And around this time, it's important to understand, IE actually would dynamically layout pages. Like if you resize the window, you just relay it out. It's actually how I wrote the original rendering layout system in NCSA mosaic. This is also by the way, the idiosyncrasy I was talking about earlier was I would break up runs of text. As you narrowed the window, it would have to break it up at different places and it like the runs would never recombine. They would just stack. So like if you stretched it, if you squeeze the window all the way down, it would take up slightly more memory and have more runs in the overall system. They had the same thing. Now, let's get navigator by the other way. But on the other hand, when you when you resize the window, it would literally reload the page from the network, not even from disk. Like disk caching was only going to start being a thing in a year or two. So they would literally go reload the page. And in fact, if you had form fields on the page, and maybe you realize that you couldn't see one of them, and you would resize the window, it would reload the page and lose everything. They started caching things across page loads for exactly that reason. But we did this kind of feature for feature comparison, but it started being more clear that they had a they were going for cheaper features than we were. We did do feature for feature on JavaScript on tables on font tags. I am actually the person who implemented CSS first in browser. This is because my my boss actually came into my office and said, Hey, we're planning out features for I three. We want you to work on frames. Because we didn't do frames yet. I frames, frame sets, etc. I was like, Hey, that'd be cool. But there's this thing called CSS Netscape's not even looking at this yet. It'd be super cool. We can do all this neat visual stuff. And he was like, Oh, let me think about it. Came back in half an hour and said, Yeah, go do that. And that was how we got CSS in the browser. I mentioned earlier about, you know, coding, coding quality, let's say, software systems. And at this point, we did at least have revision control. Like we had network drives. But in the three days, our revision control system didn't work so well if multiple people tried to check in at the same time because we frequently were checking in many different files. And we had what we referred to as the check in bear. This is literally a physical teddy bear that sat in a hallway next to a white board. And if you needed to check in, you had to go get the bear, write your name on the board, take the bear, the bear had to sit on your desk while you were checking in. And then you went back. And if someone else had written their name below yours, you went and gave it to them instead. It was an interesting semaphore, I guess. But after we shipped IE three, we started looking ahead to IE four. And at Microsoft with IE four, we actually built an entirely new engine, the Trident engine, what became known as the Trident engine. And I found out about this the first time when we were invited to a talk by this group called the forms cube team. No idea what that came from. And there were rumors that they were building something. And I didn't have a chance to go. I was busy doing something. But the other guy who worked on the rendering engine, my friend Jeff, he went and he comes back, he comes into my office and he's really, really down like how to go. It's like, we're screwed. They can do everything. They can edit, they can change things like it's a completely live dynamic environment. We're out of a job. And he was half right. Because I had done style sheets. And I gave a talk to the Trident team like a week later, where I was doing, I was showing off all the stuff I'd done on style sheets. And I was like, does this, does this, I'm hoping to get in these two features before we ship. Literally someone's in the audience said, could you not? Because anything you ship, we have to be compatible with. And like just for that, it's three features. And so they asked me to move over to the IE four team, work on style sheets. And when I was asked to join, that I was told I would be in the object model team. And I'm like, what the JavaScript team, you asked me to do style sheets. But this was kind of the key thing about this error was everything became dynamically programmable. Before that, in IE three, up through IE three, and still in Netscape Navigator, all you could reprogram were images, not the size, because you couldn't change the layout, but you could swap the visual out. The links that a particular hyperlink would point to an input fields, nothing else, nothing that changed the layout of the document. And the only way you could change visuals was input fields and images. And suddenly you could redo everything. I mean, you could, you could change inner HTML and put a whole new document in there. You could change things in the document object model. There are a couple of things you couldn't change dynamically. There's a whole 45 plus minute talk about has layout here that I'm just going to skip right past, other than to say it really wasn't my fault. People do say that sometimes. It wasn't. It was super important to understand that Netscape Navigator didn't have a dynamic system. They couldn't change all this stuff dynamically. Like they, their new system, what became Mozilla, was delayed to combat this. Instead of changing everything live through the object model, they proposed, hey, we should have a new tag called layer. And you can put things that change inside a layer. But that whole CSS positioning thing, let's not do that. We're like, no, I think we're going to do that. There's a lot of jockeying back and forth. And I think this all culminated when we shipped active desktop in Windows 98. And this is because you could suddenly drop web content onto the desktop. And all of the window decoration there was actually done in web content. This was horrifying to those of us who really knew what was going on because it was a very heavy weight system being dropped in to do a very low level system. But this was also when the dawn of Ajax happened, right? This was Alex Hotman, a guy who later worked with an IE. They needed a data transfer system for outlook web access. So they came up with this control, shipped it in MSXML and the shipped in the IE 5 timeframe. I'm not sure which of the IE 5 releases because literally we had IE 5.0, 5.01, 501A, 501B, 501C and 501D. We were trying to ship together with Office. It took a few tries. And after that, we kind of stumbled at Microsoft. We were largely just re-architecting. Like IE 5.5 did this native frames thing and IE 6 fixed all the things we broke in IE 5.5. And that was about it. Really, like we didn't really know what to do. And a lot of it is because of this chart. And this is browser adoption during this time period. You notice there's a nice peak up there. And most people don't do well at peaks, particularly Microsoft doesn't, but most people don't do well in peaks. But as we shipped IE 6, right about here, we had the best browser on the planet. I mean, we were much better than Netscape Navigator in my admittedly somewhat biased opinion. Our Mac product was actually better than our Windows product. Our Mac team is applauding somewhere and they don't even know why. But the web didn't actually have rich layout. It did great on flow text. It did great on tables, but it didn't do stacking and it didn't do grids at the time, which was kind of bad. Compatibility was becoming a real problem for us, particularly because Microsoft had this really, really hardcore compatibility philosophy. I remember having to ask for approval to break something, to fix our standard support, but it would break something in the help pages for Microsoft Cart Precision Racing, a game that we had shipped like three years before that. And overall, like the web didn't lead developers back to Windows and Microsoft started questioning why are we doing all this for something that doesn't get us more developers and we already have all the users. So why do we need to do this? They just wanted to retain the developer market. So we created this Avalon project, which became that framework 3.0 and I'm not there's at least another 45 minute talk in there. And of course, this just ended up in them losing that market to and then declining it. And of course, at the same time, the world just kept going, right? This pattern of mashup apps became more and more important. I wrote this, this white paper internally about mashups. And this is basically multiple components and data started coming together again. This seems natural today, but the idea that you had the same system that had Chicago crime statistics and mapping software and that you could just get them to work together dynamically. That was pretty awesome at the time. And of course, WebKit started gaining traction, Firefox was gaining traction. And the what WG was formed around here to because the W3C kind of lost its way. There's several hours with the talks there. And in 2004, Microsoft finally woke up said, oh, wait a second, we need to do something here. And this was actually the release that I was most proud of personally, not because of what we actually shipped, but because we had to build a team from basically ground zero. And the unfortunate part is you read this is out of the Wikipedia page on I7 and it kind of says new features, including tab browsing and page zooming integrated search box of feed reader. And oh, yeah, Web standard support got a little better, but it still doesn't pass the basic tests. And that we had to get used to being vilified at this stage. I did actually have someone in blog comments posing as Satan, offer me a job at that point, asked how much I made. And if I wanted a job and I was kind of worried, so I didn't delete the post because you never know, but don't read the YouTube comments. Don't pay too much attention to people. And I'd point out that only now are we getting into the period of time where Chrome began. So I'm not actually going to cover any of that because Peter will do a much better job than I would. But I will say Microsoft continued to kind of invest the minimum into the platform, not the browser, but the platform. And that didn't work so well for them. And a lot of this led me to the lessons that I learned out of the IE experience. Don't become complacent just because you're at the top of that heap. And Chrome can definitely learn something there. Don't focus on the competition was a huge part of it that works OK when there's only two of you, not so well when it's a lot of different people and a lot of things. We're actually pretty good at not doing quirky stuff. We try to pay attention. We try to build lots of lots of test pages and make sure that we're following a spec. But more than anything, not all the smart people work at your company and, of course, don't ignore disruptive factors like the mobile revolution, which came along right about this time, too. And this caused us some challenges on Chrome as well. But this was a huge deal, right? Like I went in 12 years from having this Motorola StarTac, which I promise you was a super sexy when it came out. It was the best phone out there. I loved that phone. Right up until the point I dropped into Lake Washington to having an iPhone, right? And the iPhone, it had all these sensors and the touch screen was super engaging and just so much power in it. And by the way, I was over 35 when the iPhone was released. But I still managed to adapt because I remember saying two years later yeah, baseline. My cell phone has to take the place of my guitar tuner now. Like it just has to work. I'm not carrying a guitar tuner around anymore. I had a guitar in my office and I'm pitch sensitive enough that I wanted it to work. And unfortunately, the web wasn't just the mobile platform at that point. It would have been great. But frankly, at that stage, the web was still designed for 1024 by 768. Literally, it was hard coded in web pages. When Apple came up with their somewhat wacky rules about how to remap onto their device, they weren't depending on this and they weren't wrong, really, to make most pages look correct. But it wasn't really open to doing touch input. Occasionally, you needed a keyboard. And the network was still pretty spotty. Like people weren't expecting this and it's super important. This is actually from around that time period, too. Ben Galbraith, well before he worked at Google, he said this at a conference and he said Web 2.0 isn't about a set of technologies. It's about caring about your user experience. And this was the big evolution then was people started really caring about their experiences offered in a browser. And prior to that, they really weren't. Post that, we actually had yet another version of that when we had to bring out the progressive web app banner a couple years ago, where we really had to focus on the fact that bandwidth was still required. Like it didn't, the web didn't work well if you didn't know you had a connection and given many talks about that. But at this point, the things that I want everyone to focus on is a big reason that I'm convinced the web platform is still that universal platform is that those access to capabilities is catching up we just have to be super careful about security and privacy and the user experience and more than anything how that platform works on other devices. This is also one of the traps that Microsoft fell in because even when Microsoft had two browser teams, the Mac team and a Windows team, the Mac team was often a corner by themselves. They weren't integrated. They weren't thinking about it. And like I co-chair of the Immersive Web Working Group, we have to think about Magic Leaps AR headset and how Web XR works on it, even though Google is not currently to my knowledge, building in our headset. So it's kind of important. Guidance for that really, this is this is Google 101, right? Respect the user, respect the platform, respect the community. Don't be the smartest person in the room for any of you working in standard stuff. Be the best listener. That's a much better way to take that on. As Parisa mentioned earlier, today this is our mission on the Chrome team. And this, by the way, overall, this is what brought me to Google. Like this is a strong part of Google's DNA. It's a strong part of the Chrome team's DNA. This is exactly why I stay here as well. And with that, I think I'm going to close with my final two Douglas Adams quotes. Thank you.