Igalia's Brian Kardell chats with Chris Lilley, Technical Director at the W3C, about his long history with the Web and the W3C, ranging from line-mode browsers to CSS to SVG and more.
Brian Kardell: Okay, hi, I am Brian Kardell. I'm a Developer Advocate at Igalia. Welcome to another episode of this Web History that we do with Igalia Chats and BlinkOn. We try to capture sort of an oral history of the web with the people who helped build it. It should be interesting for all of us and hopefully for posterity. In 100 years, maybe somebody will look back on this and learn something really interesting. With me today, I have Chris Lilley. Chris, say hi.
Chris Lilley: Hi there. Hello, everyone.
Brian Kardell: Chris is known for a whole lot of things, is maybe the only person that I know personally who has an Emmy, right? Yeah, we'll talk about that, but also known as sort of like the father of SVG, has the greatest handle online, SVGeesus. You've played a whole bunch of roles. We'll talk about some of them. Chris, you were involved in the web since, wow, very early days.
Chris Lilley: Yeah, '93, I think. Maybe possibly late '92.
Brian Kardell: How did you get introduced to the web in '92 or '93?
Chris Lilley: Well, if I can go a little bit back from that. I quit my job. I was in biotechnology and I quit my job and went back to the University of York in England to do a master's. Networking was the hot thing, right? There were terminal rooms. People would get on, they would go to FTP sites, they would go to chat rooms. They would do all this sort of stuff. That was all brand new to me. I hadn't been in that sort of environment before. Yeah, that was kind of the social thing there. It was like people would sit all night in the room doing stuff and then groggily make it to their lectures or not the next morning, but that was when the web was being done. That was '89 to '90, but I didn't actually come across it myself until a few years later. I was working at University of Manchester. I was a technical author writing training materials about computer graphics. I came across the web. There was a thing called NetNews then. I don't know if people remember that, and NTP. There were posts about this. The first ... This is a bit embarrassing, the first browser I ever used was the CERN line mode browser. I literally telneted into CERN on port 80. There was a screen and I was clicking around or typing. Actually, no, not clicking, typing. Yeah, link four. I'll have four, please. It was like, this is rubbish. It was a plain text interface. It gave me things like a list of all the phone numbers of researchers at CERN. I'm like, 'I don't care.' Also, this looks just like Gopher. So I confidently but inaccurately stated that this web thing was rubbish. I was working in a computer graphics lab, right? We were doing radiosity and visualization and that sort of thing. I heard that another lab in the US, NCSA, had a web browser, which would download images. You could actually have an image. It wasn't in line in those days, but there was a link. You could click on it and the image would pop up in a new viewer, and you could share images. I think the second browser, the one that I actually used, was X Mosaic, like 0.12 beta or something like that.
Brian Kardell: Right. Yeah.
Chris Lilley: There was new versions every few days, right? This was cool, and I could immediately see the application. We put up a web server for the lab. It was called Info.mcc.ac.uk, because the CERN server was called info.cern.ch. This was before www was the thing that everyone used, so, yeah, that's what we called it. Also, nobody cared at that point. We didn't ask anyone for permission. We just downloaded the software, put the thing up, and started adding content, which was what everyone did for a few years, until suddenly departments would go, 'Oh, wait, wait, no, I have to take all of this down. This is all unauthorized.' They would put up a four-page brochure for the university, and everything else was forbidden. That was the early days. If I could, I'd like to mention the third browser.
Brian Kardell: Totally. Yeah.
Chris Lilley: I'd heard about HTML three, by which I don't mean 3.2 by the way. 3.2 was rubbish, but HTML three was interesting. It had figures with captions and all sorts of stuff. I went to a demo of that at the first web conference, which we'll get to later. I didn't actually see it because his HP workstation was broken and he couldn't get it to project. It was just full ... The audience full of people shouting things like 'Sink on green, and try this and do that.' It didn't work. That was fine because when I got back to Manchester, we had this thing called Andrews File System, which was kind of a proprietary thing, but again, worldwide networking, right? You could CD to slash. You'd be at the roof of your universities, whatever, but then you could go up from there. You could go to slash AFS and then a domain name, and then you were on whatever they'd decided to export as their file system. I went to actually his area at CERN and I downloaded the arena browser, which was supposed to be an HTML three browser. That was fun and interesting, one of the first to support PNG, for example.
Brian Kardell: I think that was originally ... It was Dave Raggett, right?
Chris Lilley: That was Dave Raggett, yeah, that's right.
Brian Kardell: Was it based on the HTML plus largely proposal? I believe it was kind of shook ... What shook out of that?
Chris Lilley: Yeah, it shook out that. It didn't last very long because the ITF basically went, 'No, no, no, you can only standardize things that are all done and implemented. We're not having this. Stay with HTML two.' I was involved with the HTML two. I was on the ITF group that did that, but there we are. Yeah, that's the early browsers that I used.
Brian Kardell: I know that both math and graphics were apparent needs for the web very, very early. You talked about like PNG, and you talked about your work in graphics. I'm wondering ... I would like to come to the start of the W3C in a minute, but were there efforts around something about vector graphics before the W3C even started?
Chris Lilley: Yeah. There was an ISO standard called Computer Graphics Metafile. Some people promoted that literally just because it was an ISO standard. Also, because my boss at the time, Mr. Hewitt, he was actually on the ISO committee that did it, so he was all ... That was my first introduction to standards, all these 3D standards never went anywhere, like figs and GKS and stuff. Yeah. They were pushing that, but the problem was ... You could get a plugin for it. I was involved in registering the MIME type, but you couldn't do anything with it, right? I remember, actually, this was a few years later, but I had some slides and they were ... I forget what they were, what colors they were using, but it wasn't very good. There wasn't contrast. There were complaints, vocal complaints from the audience. 'Next time, can you change the colors to be, like, say white on blue?' Or something like that. Me, full of bravado says, 'Oh, I can do that right now.' Brings up the CSS, changes two lines, hit refresh, and all the text has changed color and all the graphics haven't. I was like, 'Oh, we need something where you have styleable graphics.' It took quite a while to eventually say, 'No CGM, isn't it.' All sorts of people were coming up with all sorts of different ones. HP had one, Adobe had one, IBM had one, all these different ones. Some of them were based on Java. There are all sorts of different approaches. Eventually we said like, 'Okay, we are going to start something on vector graphics that's explicitly for the web.' I wrote a requirements document and we got a working group together and we all went, 'We're not going to start with any of these, actually, we're going to make our thing.' It was so refreshing, right? Up till then I was talking with network engineers, I was talking with people that did HTML, and anything about graphics was really ... They're like, 'You mean pixels?' Whereas with these people, I was like, 'Well, we have to Bézier curves, obviously.' Then we spent five minutes deciding whether to have cubic or quadratic. It's like, well, we'll have both. We'll just have segments in each, and that's fine. There was so much agreement right from the start. It's like, of course we need this, of course we need that. I did get into trouble, though. I said, 'We have to use CSS to style it.' The people who wanted a pure XML thing were absolutely furious. In fact, some of them actually raised a formal objection and tried to get me fired for contaminating their lovely standard with CSS, but I think I was blown out in the end. That was the right way to go.
Brian Kardell: Okay, that's really interesting. We've gotten a little bit out of order now, but ...
Chris Lilley: Yes.
Brian Kardell: W3C was ... Started in 1995, I think?
Chris Lilley: Yes.
Brian Kardell: You were very early employee, right?
Chris Lilley: Yes, I joined in April '96. Before that, I had worked on a couple of things. I worked on the PNG specification, and I'd worked on HTML two, which I have a funny story about. We should maybe get back to that at some point. I went to the first ever web conference at CERN, and then I went to the fourth one because I got a paper adopted that was about graphics. Graphics on the Web: Not Just Decoration, it was called. I went through all the things that were wrong about it and I wanted to fix. On the basis of that, they basically hired me. Then my first day there, Håkon Lie said, 'Yeah, Chris, graphics are important of course, but I think you should do fonts and stuff as well, because it's not enough to fill up your day just doing graphics.'
Brian Kardell: You I think were either one of the first chairs of the CSS working group.
Chris Lilley: I was the first chair.
Brian Kardell: The very first chair of the CSS working group. That's very interesting.
Chris Lilley: Yeah. The way that happened was they were still finding their feet. They were basing themselves on what the X Consortium did, which didn't always work. They had what they called an editorial review board that would somehow review specifications. It was a bit weird. Everything was in there basically, HTML and CSS and the DOM, and then they split it. It's like, no, this makes no sense, so they invented working groups. I was made the chair of the CSS one.
Brian Kardell: Then, also web fonts?
Chris Lilley: Yeah, web fonts. There was a printing workshop in the first couple of months, so May or June of '96. We were trying to get ... What do we need to make the web, not just online, but on print? But also some things from that that would make the online experience better. One of them was web fonts. People were talking about ... The main thing they were concerned with was IPR, right? The technical aspect of it was barely considered, but everyone was terrified, frankly. I went to a typography conference, A Type I in reading in '97. I felt so out of my depth. I turned a corner to the building where it is and there's people with safety goggles and chisels, and they are carving in stone. I'm like, 'Oh my God.' Then I tell them, 'Hey, we're going to put fonts on the web and they can be just downloaded like that by anyone, and it'll be great.' They're like, 'You killed my babies. We are the last generation of people who will ever create fonts because you've taken all the money out of it,' and blah, blah, blah. We did actually get a spec together about ... Which was where app font trace came from, but then the group dissolved into fighting and deciding what format would be used. We didn't decide on a format. Eventually that group got closed down and the stuff got moved into CSS, CSS two. I'd been involved in one, but from the outside. For CSS two, we had our font face, but again, we didn't have an actual format that you could use, which made it kind of useless. Microsoft came up with one, but it had all this — Embedded OpenType — but it had all this DRM stuff around it, and people didn't like it. They were very worried that effectively the browsers would have to become the police. The browsers could be held responsible if someone used the browser to circumvent copyright and download something, so that really didn't work. Then, with SVG, we noticed that people were creating stuff in graphics programs. They would select all the texts and they'd go convert to curves, ding. You get exactly what it looks like, but it's not text anymore, and you can't search it and you can't edit it and you can't ... It's not accessible. We basically made the simplest possible thing that said, 'Convert it to curves, but wrap it with this markup so we know what each curve corresponds to.' This was really basic font format, which persisted embarrassingly long. It was just to get ... But that was implemented on all the pre smartphone things that had little SVG things. They could display those, but then it was taken out with the CSS spec, because nobody in quotes had implemented it, apart from all the SVG guys and Microsoft. That didn't count. It wasn't until much later when we came up with this WOFF format, which was really just a compression of open type with a bit of metadata at the top so you could link to a license. It turned out that the font guys were perfectly happy with that. They're like, 'Ah. Every time they download a font has a link to the license. If we sell a font to someone, and then they're using outside of license, then we can take them to court.' Yeah. What does the browser do about that? Nothing whatsoever. Nothing. Everyone was suddenly happy. The browser guys who thought, 'Oh, the font people, they want cast-iron DRM. They're going to break the web,' and the font guys who thought, 'Oh, these people, they just want to give away everything for nothing, and it's ridiculous.' Within a few years, the typography conferences that I was still going to, I was no longer a pariah at that point, there were big announcements. Adobe is making all of our fonts available for the web. You can license them with web, or you can license them for print, or you can license them for both and pay twice as much money. Yippee. Suddenly it was a business model and it worked. The percentage use of web fonts went from, well, 0% in 2011, and it's now 85 to 90%. It's become a thing that everyone uses everywhere, except China and Japan.
Brian Kardell: Web fonts are what won the Emmy, right?
Chris Lilley: Yes. It was all that work putting together the earliest CSS work and then WOFF, and then WOFF 2 is more ... Has better compression and things. Yeah, that's what won the Emmy.
Brian Kardell: That's amazing. Then, SVG. Were you the chair of SVG?
Chris Lilley: I was also the first chair of SVG, yes. As I said, there was an attempt to oust me because I kept doing unpopular things, like insisting on CSS and stuff. Eventually, because of that ... They didn't fire me, but they introduced the idea that there was a staff contact who would be separate from the chair, and the chair would always come from a member company, not from the staff, whereas before the chairs all came from the staff. I got moved over to staff contact, and then it continued on like that. It continued on for a while. It was initially successful, but a lot of the browsers didn't like it. Then they basically absolutely refused to implement anything new that we did. The first version became the only version, and after a while, that actually really preyed on me, the amount of antagonism and that we put all this work in, and that people would be saying, 'We're never ever, ever going to implement this stupid stuff.' Eventually I actually left that group and I moved on to other things where I felt my time would actually result in something like CSS.
Brian Kardell: What I like about the ... I don't know if you know this, but on the co-chair of the MathML working group, which was restarted.
Chris Lilley: Right. Yes.
Brian Kardell: Along with SVG, they're the two other ... Like I said, math and graphics were two very obvious gaps. There were ... Even in HTML plus, there were ideas about how we fill both these gaps. They both had sort of this similar problem where there's a lot of interest, there's uses, people get excited about it, but ... Yeah, it's difficult to get implementers to want to do that, but it is ... Both of those are the only two things that are integrated with the HTML parser.
Chris Lilley: Right.
Brian Kardell: They're in the HTML standard under other embedded content. You can kind of embed them in one another, which is very cool.
Chris Lilley: Yeah.
Brian Kardell: Igalia has been working for ... We did the MathML implementations, but we also have been working on a new SVG engine. I don't know if we ever talked about this, but my ...
Chris Lilley: You mentioned it, but I don't think you've told me much in detail.
Brian Kardell: My colleagues, Rob Buis and Nicholas Zimmerman are the ones who wrote the SVG engine in WebKit originally, all the way back, the original one.
Chris Lilley: Right.
Brian Kardell: Which then became the one for Blink. They really liked and believed in a lot of the things that were done with CSS, the CSS integrations.
Chris Lilley: Yeah.
Brian Kardell: But the code paths were basically completely separate.
Chris Lilley: Yes.
Brian Kardell: There's been a lot of work on that. We're really hoping that that can restart. Anyway, fingers crossed. We'll get there eventually.
Chris Lilley: Does that integration include ... A thing I've noticed, right? We add something to CSS, like white color gamma colors, and it's like, will it work or will it not work in a given browser? You flip a coin, and maybe it doesn't and maybe it doesn't. Can you do this? Can I point to a gradient? Oh, no, not those gradients. You can't use those in SVG. You have to do it this other way. There's a lot in SVG to hate, right? The document object model frankly. At that point, HTML didn't really have a document object model. It had a ... Am I in Internet Explorer or Netscape? And forked to completely different paths with completely different ... The idea of a consistent parse tree was also a distant dream that we like to pretend happened, and it wasn't. The only thing we had to copy from was the smile DOM, which was never widely implemented, but that was the only standards track document. A lot was copied from that, which is why you have all this thing.val.baseval.animatedval stuff, because it had all that, because the idea was you could always get the animated value of anything. You could change it through script and it would all ripple through this. Anyway, it made it a very horrible and much hated document object model [inaudible 00:20:16]. It's good to hear that if there's more integration, I think that's the important thing, not just the parsing obviously is already there, but, yeah, if CSS stuff can be used in HTML and in SVG, you can do all these things ... For a newcomer, it's very hard to understand why you can do that over here. Oh, no, you can't do that there.
Brian Kardell: Yeah, there was similar problems with MathML because a lot of the things in CSS didn't even exist when they were making the first MathML. There's duplication where ... There were things copied that were copied from the days of font tags and stuff like that.
Chris Lilley: Yeah.
Brian Kardell: Where you put colors in as a attribute of its own. Yeah. We did for MathML the aligning of those things as well, where we said, 'Okay, well, these things, they're presentational hints now, and they take a color.' Yeah. Any CSS color. Whatever CSS color you can put in there is going to work. Yeah, I think the more we get to where there's as much as humanly possible one web, right?
Chris Lilley: Yes.
Brian Kardell: I think that's my ideal situation, where you have these three great technologies that work in as much as possible way interchangeably so that you ... The logic that you use over here, you can use over there. Yeah. When we started our work on MathML, there was no DOM at all. You couldn't say ... You could get a reference to a MathML and say .style. It was like, it would throw.
Chris Lilley: Yeah.
Brian Kardell: It would say, 'What are you talking about? There's no.style.'
Chris Lilley: Yeah.
Brian Kardell: But now that's all addressed. Anyway, I think it's really positive, and I'm excited about that. I hope that it makes you feel good that somebody still really cares about SVG, that there's still efforts. Yeah, I don't think that any of your work is in vain. Another thing that we could talk about, I guess, is there is in the W3C also this technical architecture group that is sort of like ... I think its role, you can clarify this, but I feel like its role has changed and evolved and adapted over the years, but you were appointed to the TAG in ...
Chris Lilley: Yes.
Brian Kardell: 2001 by Tim Berners-Lee, right?
Chris Lilley: That's right.
Brian Kardell: You served there for four years basically?
Chris Lilley: Yeah. the TAG was originally started because when documents came towards maturity and they wanted to make a recommendation, sometimes Tim would say, 'No, this whole thing, this is doing it wrong. It shouldn't be like that.' 'Why?' 'Well, in my head, it should be like this and this and this.' It was too little feedback too late, and it was very arbitrary and very much the benign dictator model, which we didn't really work. We wanted to actually capture it, put it down, argue about it, and have something that would have more of a consensus to it. Honestly, one of the reasons that I was put forward onto the tags is because someone said to me, 'You know, Chris, you're one of the few people that can just stand up and say to Tim you're wrong, whereas most people are like, oh, no inventor of the web.' I did that for a bit, which didn't make me popular, but we spent so much time arguing about RDF and whether it did or did not break the web, and all of this stuff, and how many angels can dance on the head of a pin. Although I'm proud in some ways to be one of the authors of the architectural World War Web volume one, on the other hand, I look at it now and go, it misses the point so much. Some of the things were very popular then, but not now, like the halt and catch fire method of error detection. Seriously, what is a user supposed to do in that point? It's like, yeah, your tab just crashed. Good luck with that. Some of it was wrong, but yeah, it was good work. Then, of course, it all got changed. The new guard came in and they started reviewing specs and reviewing them much earlier, which meant the feedback was actionable. That was a big change. My wife, Lea, was also on the TAG, but not at the same time. It was much, much later until just last year, actually.
Brian Kardell: I was just going to say, I think you're the only husband wife team to ever serve on the TAG, I believe.
Chris Lilley: I believe so, yes. Yeah. I really want to thank Igalia again publicly now because of the childcare support, which meant that both of us could attend the meeting and see her at the breaks or whatever, but we didn't have to worry about that. We could both attend, which was very, very valuable. Recently, there was a meeting hosted by Apple in Cupertino, and they did not provide childcare, but Zoe was very good. She sat on the side of the meeting with a tablet watching cartoons and was told very strongly, 'Don't come and interrupt us unless it's really, really important.' She was great, but it meant she had a little badge, because they apparently had to get permission for her to be a technical whatever, something. It actually says invited expert on her badge, and she's five and a half.
Brian Kardell: What else can we talk about? Do you want to talk about your experience on HTML two?
Chris Lilley: I actually have a funny story about that. Yeah, I started working on HTML two and I kind of parachuted in. They were already very well established. There were a lot of people in ... Dan Connolly and these sort of people very keen on formal methods and document type definitions, and all this sort of stuff. It was all being done in STML, but there was also this ... Of course, we don't expect anyone to have an STML implementation. You don't have to follow all the rules, just do the best you can. It was all very wooly. All of these people had years or decades of experience doing this stuff. I was kind of intimidated, but I also thought, 'No, this is stupid.' I actually started to write an email saying, 'Why don't we have a grammar which defines exactly and precisely what we actually expect people to pass, and then we' ... Never mind DTDs and whatever. Totally new technology stack, let's do that instead. Then we'll have a consistent parse tree, I think that will help with the DOM as well, and blah, blah, blah. Then I realized this John Bozak and Michael Sperberg-McQueen, and they've got decades between them of experience. I was like, 'You know what? They're probably right. I just haven't got it yet.' I never sent the email. I didn't send it and have it shut down. I just ... Never mind. Delete. If it had, and if it had been accepted, which of course is far from given, we would've had consistent parsing between browsers 15 or so years before HTML 5 rescued us by giving us consistent parsing in browsers. This is one of my lifelong shame things, the sort of thing when I can't get to sleep at night, I'm like, 'Oh, you idiot, if only you had done.'
Brian Kardell: I wonder if we would still wind up with WHAT working group in the end, because the very first thing that they took up was the parser, but I think that that was mainly because they had to.
Chris Lilley: It was also because they wanted to describe the behavior, whereas the old HTML specs were literally just here's the grammar. By the way, the A element is about links and stuff. You know. It's like, no, no, that is not ... You know is not a specification. You have to say exactly how fetch works and what happens and what happens when it's interrupted, and how does it work with caches, and all of this. That was the ... That's why HTML took off. Not because you could have bold and things, but because it had elements like tables and forms that did stuff. The same for SVG. It had things that actually did stuff, so you, just by writing a small amount of markup, you would get that behavior. It was the encapsulating behavior I think that was a crucial thing.
Brian Kardell: You had said that also you were at the first World Wide Web conference, right?
Chris Lilley: That's right. Yeah.
Brian Kardell: What was that like? Because I wouldn't be involved in the web for another four, four or five maybe even years. I definitely didn't get to attend those. Tell me something about it.
Chris Lilley: Well, as I said earlier, I was working on writing training materials about computer graphics. Most people were making what they called multimedia CD-ROMs, right? You'd get the CD duplicated, you'd stick it in your machine, you'd have this point and click type stuff, and you'd do things, usually using some proprietary framework, and that was it. We didn't do that. We wrote student notes and whatever in a frame-maker think, and slides. By slides, I meant things that you could print onto acetate and put on overhead projector and project. That was what we used. That was what the state of it was. I realized that because the web now existed, we could ... Instead of having to order these pre-printed or wherever, we could put the postscript files up on the web and we could put documents that we wrote on the web. We could put programs that were demonstrations up on the web and people could download them and run them. We could update them over time when we found mistakes rather than that was that addition, and maybe we'll do another one in 10 years. It's like, no, the one we put out last week, we realized it says this and it should say that. We changed it. I wrote a paper about that, went to Barcelona to present that at a graphics conference. That got me more involved. Yeah. That was where I got to go to the first web conference, because there was a part of it that was about use of the web in education. That wasn't my primary interest, actually. I was sneakily going there for all the other stuff. That's where I met Dave Raggett. That's where I met Tim. That's where I met Dan Connolly for the first time.
Brian Kardell: You were talking about CD-ROMs and it clicked a thing in my head. I'm curious, were you into hypermedia before the web?
Chris Lilley: A little. Not that much, honestly.
Brian Kardell: There were things like HyperCard and FileMaker Pro.
Chris Lilley: Yes.
Brian Kardell: Things like that that had hypermedia style creator bits where you could do a lot of that. I wonder, have you ever heard of or met this guy, Ian Ritchie? He made a bit of software. I think his software company was called Owl. The software that he made was called The Guide. It was a hypermedia thing in the early nineties basically, when Tim was making the web, before there was even a browser. It is this great short TED mini talk that's five, 10 minutes, something like that. It's really short. You can find on YouTube. Maybe we can link it up in the show notes, about how Tim actually approached him at a party in Vienna and was like, 'Hey, maybe you could make the web browser ... Because I'll just even contribute the ... All I want is a URL. That's just ... Can we just add the URL?' Because it already had networking. It already did everything. Actually, this is really interesting, it was specified in a markup format. It was called HML. There's this joke that Tim added the T. Yeah, I don't know. I think that's really interesting to think about. I don't know what ... But before the web, there was almost the web.
Chris Lilley: Yeah.
Brian Kardell: Even the way that Tim defined HTML was not by sitting down and saying, 'What tags do we need?'
Chris Lilley: That one is very easy because he looked at some CERN documentation on a thing called the IBM Starter Set.
Brian Kardell: Right.
Chris Lilley: Which was for STML, and that's where we get H1 to H6 and P and UL and LI and all that stuff.
Brian Kardell: Yeah. Even GML. Before that, right?
Chris Lilley: Yes.
Brian Kardell: Dates all the way back to the sixties. That original set and the first set that's in there, and also the set that's in Markdown, is just exactly the set that you would think that would be, because they're the ones that would be on a typewriter if you were just typing on a typewriter. They're like, which things are the headings? Which things get some kind of emphasis?
Chris Lilley: Right. Speaking of other hypertext things, I did actually meet Ted Nelson a couple of times from Xanadu. We had some interesting conversations, but I found him excited, kind of like Tim is, but also he would casually say things like, 'And of course to do this, we need a complete worldwide overhaul of the copyright system.' I could see that putting that as a prerequisite was not a path to success, but it was interesting ideas.
Brian Kardell: That's interesting. How did you meet him?
Chris Lilley: He was in Japan at the time. I was there for a meeting in Japan. I had dinner with him and Tim and Robert Kayo.
Brian Kardell: Has he been involved in anything for the web?
Chris Lilley: Not really.
Brian Kardell: Because I know he likes to kind of diss it a lot.
Chris Lilley: Yeah. Yeah. Well, he doesn't do the things that his system did, but his system also doesn't actually exist.
Brian Kardell: Yeah. What else? The World Wide Web Conf. You ...
Chris Lilley: Oh, yeah. Actually, that reminds me. On the way there, right? I'm on a plane to Switzerland and some guy comes up and says to me, 'Hey, you must be Chris Lilley.' I'm like, 'Yeah, I am actually.' 'You going to the web conference?' 'Yeah, I'm actually. How do you know?' He's like, 'Well, your photo's on your web page.'
Brian Kardell: Nice. I guess that's ... Probably there's a lot of people who can relate to that now, but in the early nineties, not so much, right? That was ...
Chris Lilley: Oh, not at all. People were mostly using dial-up, so you had tiny, tiny little images. I was looking at some of those graphics now and they're like ... It looks like they're like 5 million by three on the screen nowadays. It's tiny, tiny little things.
Brian Kardell: Also speaking of this, you know I work at Igalia with Eric Meyer. Eric has told this story I think that you were the one to invite him to the CSS working group.
Chris Lilley: Yes, absolutely. Rather than working in a vacuum, people who were putting up test pages and pushing for interoperability, yeah, I wanted those people involved. It's like, okay, well, I don't work for a member company. No, but you're doing really good work, so we want that. I did quite a bit of that in those days there. I also got Daniel Glazman to join and eventually become chair, because he's like, 'You're doing it all wrong, this, this, and this, and you shouldn't do that.' I said, 'Well, come on and fix it then. Come on. Put your money where your mouth is,' and he did, so that was great.
Brian Kardell: Yeah.
Chris Lilley: We haven't talked about color yet.
Brian Kardell: I was just going to say, you also are involved with color. You have forgotten more about color at this point than I will ever know. You and Lea, and I believe Una now as well, have been working on the color spec.
Chris Lilley: Yeah.
Brian Kardell: I have seen some amazing talks that you gave about this. How long have you been studying color?
Chris Lilley: Okay. Back in Manchester, I did write some training materials about color. This was a modular thing. We were supposed to take three months on each module. I took I think probably a year and three quarters, and they were very annoyed at me, but I kept finding new stuff and not letting it go. I got really hyper focused on it. That was a good thing and it was well received. That fed into PNG, because I was the one that insisted we have to have tags for the gamma, the display, and the chromaticity of the monitor so you could actually get ... That was the days where, is this a MAC graphic or a PC graphics or an SGI graphic? And that sort of thing, so I wanted that tied down. As to how long it took, in 2018, October I think, I gave a talk to the International Color Consortium. It was an evening talk. It was after the meal and it was an invited speaker sort of thing. I was just getting involved with them. They said something light. I called it Color at W3C: 20 Years of Pain. I described all the attempts which had failed, and why they failed and what had ... Etc. It ended up saying, oh, we have this thing called CSS color four. We just had the first working draft. I don't know if that's going to work or not, but I really hope so. It's interesting to think about what was different in that. Instead of writing a spec and saying, 'Okay, guys, implement it.' One thing I noticed was that people found color very worrying, very off-putting. They thought ... It is very complex, right? But at the basis of it's actually quite simple. What I wrote was the simplest little bits of JavaScript. It's like, yeah, if you've got display P three and you want to convert it to X, Y, Z, then you do this. If you are X,Y, Z and you want to convert it to SRGB, you do that. It's basically mostly a matrix. It's like multiple ... Linearize it, multiply it by this, and then add gamma back and you're done. By making it simple, I kind of tempted in the developers to use that. It also meant that we could give examples. In the GitHub issues, we could say, 'Well, this code would go to that.' Everyone was using the same code, but it was horrible code because it was simple and procedural. You had to ... You'd have six or seven nested function calls to do all the things. If you didn't do it in the right order or the right way or whatever, you missed one out, then you get completed the wrong result. With Lea and I, we did a different thing. It's the same math, but it's object-oriented. Instead of two lines of gobbledygook to convert one color thing to another, you would go RGB.2, and then whatever you wanted like OKLab, and it just converts it for you and it knows how to do that. It can follow the chain of what it needs to convert one to the other. That was great because people could use it. They could simply prototype things. The browsers were using that to check what their implementations were. In many cases, they were copying the same matrices and whatever, so they got exactly the same results. That really helped. It broke the chicken and egg thing. It meant that people could put that in and see what happened and they could compare what their implementation did. I think that really helped us move it forward. The other thing, of course, was Apple. Apple suddenly decided that P three screens, wide gamut screens, were important, and then they put them on laptops and watches and phones and everything. They all had the same wide gamut one, so they wanted people to use that. That was really the impetus. Suddenly a major browser wanted to have this stuff as opposed to, Chris, no one needs that stuff. That's just ...
Brian Kardell: This is really funny because I remember in 2019, I was at Mozilla CSS working group face to face, and I went ... You and I talked, I think we went to dinner. We're discussing this and you were saying, 'This is what we need. We need to find somebody to do it and get it going.'
Chris Lilley: Yeah.
Brian Kardell: I wrote about it. I started talking to people about it. Funny thing is that a lot of people ... I think I had a similar reaction. I grew up my whole life with CRT TV. You watch the CRT from across the living room and all those little dots pull together pretty well, and it's convincing.
Chris Lilley: Right.
Brian Kardell: If you had the Sony Trinitron nice ... Somehow you thought, 'That's a nice sharp picture.' Then at some point it hit a ... If you saw it next to a modern display, you're like, 'What is that?'
Chris Lilley: Yeah.
Brian Kardell: Right? But you never felt like it was missing. I feel like that's the experience that a lot of developers told me that they were having. I've watched slowly shift this Overton window to where they understand, oh, wait, no, that's not the end of color. All the things I could express with SRGB is not where color is limited, right?
Chris Lilley: Yes.
Brian Kardell: I thought I had all the colors, but I don't. Actually, those are beautiful and there's all these other things that I can do with it with these new specs. Then I think people started really getting excited about it. It's funny how that shifts, right? How you don't know until you see it that you missed it.
Chris Lilley: Yes, absolutely. Absolutely.
Brian Kardell: Yeah.
Chris Lilley: Another thing that helped, Sam Weinig, who was working at Apple at the time, took a big leap. He started implementing all this stuff. Some of it would change every couple of weeks. He would just track. He was in at the issues, he would change the implementation. There were all these previews coming out of Safari. Each one would have in the change notes, and done this, use the new matrices, do this, do this. The fact that he got that going and it was always tracking the spec was amazing. Also, gave the other browsers something to aim at because they didn't want this to be an Apple only thing, which it could have been, because the Windows machines and whatever, they took longer to get wide gamut screens. Yeah, I think that really helped, getting developer interest. Now that's a given, right? You have to get developer interest. There's no point in writing spec otherwise, but that really helped it. The one other thing that we really need is understanding. That's the thing. The specs are not written for the general public. People don't know what they can do, why they should do it. I kind of cry sometimes when I see some new bit of color syntax and all ... In a blog post or whatever, and all the examples use SRGB. Sometimes ... Well, often actually, because I'm an annoying person, I write after the authors and say, 'Why did you use SRGB?' There's two answers. One, I didn't think about it very much, and two, well, the colors I'm playing with are in SRGB, so I have to use that, right? I'm like, 'No, it didn't say that at all.' Oh. It's the understanding and it's what can I actually do with this that is kind of missing.
Brian Kardell: Yeah. I think that for some of us, maybe you're an exception because you've been so inundated with color for so long, but for me ... I don't want to say I learned to think about color in SRGB. If I turn to write a color, I can kind of hunt and peck my way there in SRGB. I frequently do that because it's the thing that I'm used to. What are the ... I guess we'll see in another five, 10 years what the next generation of people who aren't limited by that and don't have this burned in memory, what they do. Hopefully it's not that.
Chris Lilley: Yeah. Things like perceptual uniformity help and make it easier, but then they give you new problems. It's easier to go out of gamut. You take a color, you rotate the hue, and now it doesn't work anymore. Depending on what the browsers do with that of gamma colors, which is frankly terrible at the moment, then you lose a lot of the benefit. Yeah, I think having something where you can say, 'No, this color visibly looks between these two. I want a gradient between this and that,' and the middle color looks like what you'd expect the middle color to look like, that wasn't a given before.
Brian Kardell: Oh, yeah, that's for sure. That's for sure. Those are great examples actually. The gradients are amazingly better. I think that's about all the time that we have.
Chris Lilley: Can I tell you a quick funny story first?
Brian Kardell: Absolutely. Yeah.
Chris Lilley: There is a great video on YouTube by Alex Sexton I think. It's about the X-11 colors and what stupid names they have, and blah. It goes through the early history of color. My involvement was quite early on, so I was a bit more abrasive then than I am now in my nice refined and polite self. He said, 'Ah, now we have Chris Lilley, basically an asshole. Also frequently correct.' I don't know. I kind of want that in my gravestone. It's funny.
Brian Kardell: Excellent. Thank you.
Chris Lilley: You're welcome. Thanks, Brian.
Brian Kardell: I've never found you to be an asshole, but great. This is really fun. Thanks, Chris.
Brian Kardell
Chris Lilley