 I think this opening slide pretty much sums up the first part of my talk. Good afternoon. If I had to guess, when you first think about design of Facebook, this is not what comes to mind, or this, or this, or all of this stuff. Someone also made a handful of these, not really very nice looking, but it's an important point, mostly for myself, because there's probably somebody with a sharpshooter in the back of the room, just in case I say something I shouldn't be saying. Yeah, Eric's here to keep me in line. But for the 1.2 billion people across the globe that use Facebook, this is the typographic experience they have pretty much every day, as is this. Shelly, this one's for you, and this. And even this. So this is the new paper app, which has been out for a little while now, and actually sort of has its own set of typographic nuance that goes beyond what people are used to in inheriting a lot of the typographic styles from articles, a lot of publications that get posted through the service. But at the same time, it still feels like it's a missed opportunity. And one of the two things that I actually want to look at today is really how type at Facebook is sort of the exact opposite end of the spectrum from what Dustin and Matchin were talking about yesterday as far as medium goes. We face some of the similar challenges, but scale plays a big part into the decision-making process that goes into like why things are the way they are. And this is actually something that's really interested in me, not actually being on the product design team and rather being on the communication design team, just understanding why those decisions were being made the way they were and what the actual impact is of the decisions themselves. And so, you know, you're probably used to seeing something like this. This is not actually true. It's actually really only people on the Mac that see Helvetica or iOS. Everybody else gets pretty different type experience depending on what platform you're on. But based on the commentary surrounding Apple's announcement that Helvetica Noi would be replacing Lucida Grande for the first time as the default type face across the Mac and our own redesign of News Feed earlier this year, it seems pretty easy to understand why people would be kind of ticked off and just generally frustrated with some of these what seem like haphazard and just regrettable decisions around type. Foser to that point, if you posted something about this and I'm friends with you or follow you on Twitter or something, I probably liked it or saved it or filed it away somewhere. Steven actually has a nice sort of roundup quote about why Helvetica is a regrettable decision. In short, Helvetica has almost none of the characteristics that are required from readability at small sizes or long passages, rhythm, openness, moderate letter spacing, moderate contrast. Of course he's not alone. Let's switch to Helvetica to improve legibility. Said no typographer ever. Grammatical problems with this aside. Jackson, oh lord, new Facebook is unpleasant. Let's get some new fonts on there, Facebook friends. Did Facebook roll out a new design? Why is the typography so terrible? And again, Steven, sad trombone. My Facebook just went Helvetica and the global shift away from Lucidagrand continues. Now while I can't exactly comment on Apple's rationale for the switch other than an obvious alignment between their core operating systems, what's often overlooked when it comes to Facebook, though, is the effect of scalability in these decisions on type. What actually surprised me the most when I was working through preparing this presentation was that so many of these decisions are actually not design decisions. I mean, they are to one degree or another, but they're actually engineering decisions. Bear with me, though. If we go back to 2004 when Facebook first launched Harvard, the web was a very different place. For one thing, Facebook was known as the Facebook. And apparently, Mark couldn't decide whether or not it was the Facebook one word with a lowercase t or the Facebook as one word with an uppercase t. I don't know. Anyways, let's not look at that anymore. At the time, there were only about 719 million people who had access to the internet. And yes, that's a very big number compared to today that's actually fairly small. So today, we're actually looking at somewhere in the neighborhood of 2 and 1 half to 3 billion people, depending on where you try to find the numbers. And at the time, that's not the word that I think most people would use to try to describe typography on the web or especially typography at Facebook. And this is probably still largely true. In a lot of ways, consistency and beauty is not the point. This is not what we're actually going after. For Facebook, what actually matters the most because of the scale is utility first and efficiency. So what does that actually mean? It means that if these things are not useful and by the same token, if they're not usable, then they're not going to work. And this is really at the heart of Facebook's design philosophy, or at least it has, met up until now. At a high level, custom fonts have really not entered the equation for the simple reason because they add complexity. They add additional CSS to the mix. They add additional JavaScript to do things for working around problems. And they just add weight to the payload for every time you load a web page, or if you're loading a native app binary, if you try to embed a font into that, you're putting a penalty on the user for actually having to do something that they really shouldn't have to do in order to have a nice experience. And so to understand what that actually means for scalability, let's look at a spectrum of attributes that ladder up to utility and efficiency. So infrastructure, access, surfaces, and then ultimately people. So the underlying size and complexity of Facebook's infrastructure is really the first piece that is keeping web fonts out of the mix. As far as access goes, what I'm referring to here, and we'll look a little bit more at these specifically in a minute, I'm talking about networking. So the actual wires and wireless spectrum of getting data from one point to another. And then the surfaces being the actual devices and pieces of glass that we all stare at all day looking at type and various other things. And then people, which is ultimately why any of this stuff matters at all. I mean, Facebook's mission is to connect the world. And you can't do that without people. And in the case of Facebook, it's actually a lot of people. So fundamentally, these factors and more are really the things that are driving the typographic decisions affecting Facebook products. And affecting how people are experiencing them. In order to achieve utility and efficiency, it means type must first be intentionally simple. It must work incredibly hard. And it also has to be really inexpensive, both for Facebook and for people at the other end of the wire. One of the things that we throw around a lot at work is actually nothing at Facebook as somebody else's problem. It's not somebody's problem. And this actually describes that the problems that we have to deal with, they're not all ours, but at the same time, they are still our challenges that we have to try to solve one way or another. And we can attack scale problems from design or from engineering. And sometimes, probably most often, it's actually a combination of those two things. And you would think this would be an easy challenge to solve given the size of the actual Facebook employee base and just how big the services and the resources that are available. And maybe it should be. But it's actually not right now. It means code needs to be developed. It needs to be maintained. It needs to be deployed. And we need to rinse and repeat this to optimize it again and again and again to try to echo it as much performance as we can. Because ultimately, for us, performance goes hand in hand with utility and with efficiency. When you get into scale, these effects just become enormous. And networking also plays into that because we know there's a lot of people that are here in North America that use the service. But there's even more people in the rest of the developing world that are using the service. And the conditions under which they're doing so are vastly different. We've got it easy over here with our high-speed internet and our really powerful iPhones and Android phones and desktop computing and all that sort of stuff. Whereas on the other side of the world, we're dealing with low-power feature phones that, frankly, we had 15 years ago, but we barely functional at this point in time. And at a small scale, these things are actually relatively easy to manage, which is why we've seen web fonts work really, really well across a lot of smaller-scale publications and what are largely static sites where the content is known up front. We don't have to worry about dynamic loading of content and people sharing content between people that might be from in different languages and where you just never know exactly what's going to show up at any one point in time. This is really sort of what the world looks like today. All of these different phones. And this is where everything is headed. And we sort of know this. And this is where things are just the most constrained. I mean, your phone at this point in time is not as powerful as a Mac Pro Tower or even most laptops. It's getting there, but it's still got a long way to go, and it's got an even longer way to go when we think about those developing countries. And we have to think about these things. So we have to think about the capabilities of these devices. That means processing speed, storage, the actual memory of the device that you can allocate to an app, whether it's a web browser or a native app. And then whether or not there's even support for things like custom fonts. So we could want to use all the custom fonts in a native binary app, but if there's not APIs to support that, or it's just not possible to get a font onto a device one way or another, it's just not going to happen. And to a lot of extent, without us just putting pressure on manufacturers to allow these things, there's only so much we can actually do. And then there's things like rendering quality, which, of course, we've all seen not all fonts are created equal, especially when they're being used on screen and for largely UI purposes. There's things like rendering performance. Something will load. And again, milliseconds here count for everything. And then there's also the size of the app binaries themselves, the size of the payload that we're downloading. And these things all just add up quickly. And it gets even worse when we look at the problem of having high resolution displays, as well as sort of these low resolution displays that most of us have been used to for some time now. And then things like networking speed, latency, caching, cold start times. There's lots of little factors that go into these things. And we look at performance as really being a key indicator whether or not we can actually do something here. If we actually just take a quick look at a relatively standard web font file, so one weight of Neufurtiger thin, somewhere around 83.24 kilobytes. Now, most people, for just a regular website, they're not going to bat an eye at this. This is about the size of a pretty reasonable JPEG and probably smaller than even a 2x size JPEG image if you're dealing with retina screens. You actually want to optimize your site for those. On the other hand, though, for Facebook, this is the language support that we're currently providing. So we've got a whole bunch of largely Latin-based languages along with some Cyrillics. And then there's all this other stuff at the bottom here, which is way more complicated. And if we actually look at what typefaces are generally available on devices for these things, there's not a lot, including your computer. So a good reference point I found was this aerial Unicode MS font, which is typically what a lot of fallbacks end up going to for Facebook. And this thing is 15 megs. So can you imagine trying to download this thing on a shitty 3G connection in the middle of nowhere where the connection is cutting out? And you're not going to have a good experience. And even things like dynamic sub-setting to try to mitigate some of this massive download doesn't really work when you get to Facebook scale, because you're asking, the service is going to pay a big penalty. And then there's also the additional complexity, which will slow down the service at the other end, at the actual user's end. And all of this comes down to predictability. So because of the wide spectrum of these things, we actually can't really very well predict what's going to happen and how things will behave. That means both from trying to design for these things, but then also just the actual implementation and knowing there's so many edge cases. And the bugs just pile up. The ability to use web fonts becomes a problem just because of the fact that we also need language support beyond Latin. These fallbacks can't be haphazard, because due to the overall complexity of the system itself, design and type choices must play into that predictability. But as I mentioned earlier, consistency is not the goal. It's actually more sort of like harmony, I guess. Products should feel at home across these native surfaces, and ideally do not call undue attention to themselves. At the end of the day, these decisions are ultimately affecting people. And to me, really, this is where the opportunity is for us, especially if you consider that 80% of the 800 million people that use Facebook every day are outside of the US and Canada. That's a real number. For a large part of this audience, English is not going to be their first language. It will not be the thing that they're mostly seeing in their news feed or on their timelines. The Atlantic actually recently published a piece a couple days ago that echoed some of my own feelings about the challenges of operating in a global scale and how little has really been done at bringing the diversity of typography largely aimed at print into the world of these digital products and services. We're so far behind right now. If you actually look at the typefaces that come pre-installed on the Mac and on Windows over the last 10 years or so, basically nothing's changed. Or so minimally that it actually has no impact on something like Facebook where we're dealing largely with UI text. And it's really this utility design first approach, in my mind at least, that's actually hampering growth for Facebook and for probably a lot of other companies as well. And it's an engineering challenge that we've got to figure out how to make this work. And we also need the help of people like yourselves who are designing typefaces to work with these operating system manufacturers and mobile device manufacturers and get more screen optimized and just better typefaces onto these platforms to make this just generally easier for us to be able to use without actually asking the end users of these things to really pay a big penalty for a lot of one-off sort of solutions. Because we're not going to be the only ones that are going to try to do some of these things over time. So really by this point, this is probably how you feel. It's kind of how I feel, at least when thinking about this. But luckily not all things are actually created equal, which is an interesting sort of thing to think about at Facebook. And that we actually do use web fonts on certain things, some of our consumer marketing pages, business pages, and developer platform collateral, where we're talking directly to those user groups. We can get away with making those trade-offs in efficiency just because the scope and the scale and the reach is just so much smaller. Here's a couple examples of that. Internet.org, we're actually doing our normal sort of web font is actually freight sands. But for Internet.org, one of this sort of another initiative, which we're actually trying to do something pretty different with, we're able to get away from sort of anything that looks remotely like Facebook. And here we're using a Garamond web font. I think it's sort of through Typekit if I remember right. Now what I'd like to do is actually divert completely and stop talking about Facebook products and actually talk about an important part of the culture within not just the design organization, but just the organization overall. It's something that's not really known outside of a lot of the design circles and tech circles outside of the Bay Area. It's this thing called the Analog Research Lab. And this is actually where I work. It's a little corner down in one of the buildings in our Menlo Park campus. It's about 1,800 square feet, I think. And it's an important part of the culture because we focus a lot on type. And by the sound of the name of the group, the Analog Research Lab, you can imagine what sort of things we do in there. And part of that is also this Artists in Residence program. And this is where a lot of the most expressive type and design is actually coming from outside of the product sphere within Facebook. So this particular piece here was done by Jeff Canham and is rather large on the second floor just above where the Analog Research Lab is. And there's a secret code in here which the engineers are supposed to try to decipher. I actually don't know what it is, but it's really aimed at them. But here's a large typographic piece. We also happen to have a fairly modest wood type collection and a little challenge letter press known as Old Blue. And we get to work on things like this, where we're laser cutting type out of wood and letter pressing it. Or occasionally we're starting to do some work with polymer, which is a totally new thing. We haven't really done that before. We do silk screen, some posters that we made. The one right beside me was actually cut with a laser cutter. And we did that while we were in the middle of silk screening a bunch of posters. What's interesting about this stuff, too, is that the scale of it is actually so much smaller. Like we're really just dealing with our internal audience within the company and trying to inspire those people and cause them to stop and think and just notice something that they're not going to see otherwise. And get them to just recognize that we can use type in more interesting ways. And we can let type actually do a lot of the work. I mean, type is doing a lot of the work across the product surfaces already. But maybe it's not working as hard as it could be or making it as frictionless as it could be. We also do things like these publications. This is the Book of Hack, which is something that every new employee gets a copy of. And it really sort of describes why Facebook exists and what our mission is and what our values are. We also have more expressive things like this. This was designed by Eric Morinovich and gilded by the folks at New Bohemia Science. We also have a barbecue shack. So we get to do some interesting typographic design with things like this, some of these amenities within the campus. Smell the savory, smell the savory sorcery. And really the only thing that's different about this is the scope and the size. The biggest problem we have is figuring out physically how big something has to be. There's a nice little look at some 3D lettering we did based on Silas' heroic for this year's FA conference. That was just a couple months ago. And our work actually also crosses over into digital. So one of the things that most people probably don't know is that we worked with Eric Olson, Neil Summerour, Aaron McLaughlin, and Tom Grace to create a customized version of Clavica, which we've ended up using as a driver for a lot of identity projects. And there's two weights to this typeface. And it covers an extended Latin character set plus full Greek and Cyrillic. And then we also, when we launched the Facebook Home product, we worked again with Neil and Aaron to do a few little other things here, covering Japanese and Arabic and Chinese and Vietnamese Thai, I think Korean as well. I think the big thing that I want everybody to try to take away from this is that we actually make mistakes. But we keep trying. And these things are, at least as far as the product's going to go, they're going to improve over time. And we're actually at a point now where I think we've hit that point where utility can only go so far. And people are starting to be open to a new conversation about typography within the company. And so I'm hoping that the evidence of the fact that we went and created a custom typeface, I've also built an additional glyph typeface within the product org for them to use. There's a lot of interest in this stuff. And people are actually really starting to pay attention again and want to move it forward. And if we look at the fact that we're connecting more and more people every year, more and more people are just getting on the internet, whether it's through Facebook or whether it's through Google or whoever it is that's flying balloons up in the air or sending drones with internet signals around in the sky, people are just getting online and they're getting connected. And it's up to us to make sure that that experience that they have is going to be a good one. So what I'm saying is there's still hope. Thank you.