Liveblogging a panel at South by Southwest 2005 (SXSW2005; SXSWI) with Joshua “Rock Queen” Darden, Shaun Inman, Mike Davidson; 2005.03.15 09:53
Josh: “Me first.” His background in typography is indeed with screens and Web sites. His first company was Scanjam; first projects were for Web sites and specific Web projects “addressing things that I didn’t think were being addressed fully by existing [faces].” Were mostly bitmaps at first, then outlines “that were specifically designed to render good bitmaps on screen, and that’s sort of how I got started.” One shortcoming of his image-based type was that they were “hardwired into GIFs,” meaning they were not searchable. (Hence Josh essentially makes an accessibility case for live text.)
Now is working at extended text at very small picture sizes or in bad printing conditions like newsprint. “How things will render in specific media… drives what I do.”
Will define screens narrowly. “As Mr. Clark will tell you, screens encompass many things we can read – captions and subtitles on the screen, video/DVD,” destination displays in rail stations. Square pixels and phosphors or LCDs are components specific to Web fonts “that interest me.”
Text is typically handled by core fonts like Georgia and Verdana, “and that’s terrific”; display is handled by a core font “that’s been tagged somehow” or an image. Display is where things begin to get a bit tough. “Fonts for extended text always have to be comprehensively useful; they in fact tend to be a little boring.” You can’t set extended texts in Zapfino; he shows an example in Zapfino caps (an unlikely scenario at best) and a semiserif screenfont.
Readability and differentiation of onscreen text are big issues for him. He shows 356890 in Arial, “based on a font that was designed around 1826 – your typical British or American grotesk…. The shapes are closed; they’re very circular.” Blurry or small type makes it impossible to differentiate; Arial isn’t useful for some applications, like bus schedules, “U.N. relief information.”
These problems have been addressed “by two of our best typefaces to date, Georgia and Verdana. Granted, they’re boring” but they are successful at being distinguishable and readable. The ClearType Font Project expands from Verdana to include a larger character set.
He shows a page from the bible in a previous century. Sticking to core fonts for that same text “loses a little something in the translation. It’s just not quite as cool – and coolness is important.”
“Typography is not just about communication in a clinical sense; it really is about expressing emotion or grandeur… whether it’s lyric poetry or deathmetal lyrics or love notes, it really does need to express, really does need to capture the spirit of the moment.” Has manageress passes around an edition of Erasmus’s colloquial Latin phrases that is 337 years old. It fully predates computers, “but it manages to clearly capture what it’s all about…. Unfortunately, onscreen typography cannot now express what we had as a matter of course 300 years ago, and that kind of sucks. My personal mission as a typeface designer is to try to embody content. I do it through typefaces; designers do it through design of documents.”
Wants to pass something down to future generations that is “warmer than Georgia and Verdana…. We don’t have to lose our typographical culture just because we’re now reading things on screen.”
How do we get there? Type designer; operating-system developers; rendering-system developers; standards developers; interactive users and developers; and the users, “who are at the mercy of everybody else” and tend to be separated from everybody else. But users and type designers are the easiest to put together in the same room for a discussion.
Mike: My interest in typography grew the day that I realized I couldn’t draw. That was the day I first tried to draw, and that day was his first day of art school. Not that many people understand the labour involved in designing a typeface – 2,000 work-hours, according to Josh. That’s a year. “We’re silly not to view that for all that it is.”
Designed posters for sports leagues out of “college.” It was a great exposure to type, since the audience – kids – wasn’t all that picky about type. But the Web was a bit of a disappointment because everything was Times typeset in black and white.
Shows found type from Austin. The more words and letters you see, the more readable the font will be. He, like me, loves the neon in this town.
In his last ESPN redesign, he wanted to bring type to the forefront. You go to ESPN to find out not just what happened, but how important it was, and type is a great way to do that. Compares “war mode” ESPN homepage to Sportsline.com’s, “with black Arial text over a rectangular photo, which is the same thing they have every day.” Yahoo Sports has “an even smaller photo with even less interesting text.”
“ ‘Skirmish mode’ is not quite as serious as war mode, but we use it when something pretty important happens.” It’s more rectangular; war mode uses a large type/photo illustration with a curved side.
They designed a Flash 4 component that would let them inhale Akzidenz, their corporate font, and render it. It still used an alt text and an h1. Shaun “The Wolf” Inman did a much better job at using Flash for type without changing the HTML. (“Shaun has solved so many of my problems that he has begun to remind me of the Harvey Keitel character in Pulp Fiction, who was known to make problems go away.”)
Shaun polls the room to see who’s familiar with Flash replacement – nearly everybody.
Shaun: Mike’s the pusher, the advocate. Looked at it from an educational standpoint; wants to talk about what we need to do.
One easy way to maintain typographic control, which designers certainly like, is to use px, which IE/Win cannot resize (due to its own browser bug). According to the Noodle Incident, set your body element to 76% and other faces in em; that always resolves to 12 px.
Some image-replacement techniques were inaccessible due to bugs in screen readers; you can fix that by positioning an image -3000px or so offscreen.
Question from Brent: Which is really best, px, em, or %? Shaun recaps the Noodle Incident technique. Mike finds himself caring less and less about Internet Explorer every day; there are better browsers to use whether or not you have “special needs.” “I find that taking little things away from them can be fun, and this is not such a bad thing to take away from them. I still size my type in pixels. Yes, the 76% technique is better.” Two more questions, one from Andy Budd, clarify this.
Question from Jason Santa Maria: Are there any further movements to get more core fonts on computers? Josh: Unfortunately, there are invariably issues with licensing. Getting a font into the hands of millions of users is a problem. Just selling a license to several million users requires several million dollars (in theory). “There are quite a few of us, though, who are developing typefaces for heavy information as well as [display] typography.” We’re used to 72-/96-dpi screens, but at TypoTechnica somebody was using a 200-dpi screen “and it changed everything” about type on a screen. The onus is on OS developers interested in doing the work and getting them to actually design the faces.
Mike: Traditionally, this has been mostly a technological problem. 1984-era Macs used aliased type where even print fonts “looked like crap on screen.” Now we’re left with business barriers – getting the right people in the room together. sIFR came from an idea to do something “that’s not perfect” but works.
Question from me: I’m in favour of any innovation that makes pages look better that also is proven not to harm accessibility. (Mike: “Is somebody recording this?”) But with sIFR, isn’t it true that all you get is a better choice of typeface and not a better usage of that typeface? You don’t have kerning and letterfit, for example.
Mike: “I would agree with that, but that’s half the battle…. The tests that you need to pass in order to use it is a technical test” – CSS, JS – “whereas I would much rather have it as a design test – for example, if I could lock the zip file and ask a question like ‘What year was Helvetica invented?’ before you could use it.” “If you use [sIFR] improperly, it can kill your site.”
Flash cannot programmatically kern text; you can do it manually inside of Flash. But that’s not a parameter you can pass through in a script. “I don’t care which one people use,” Mike says, referring to IFR and sIFR (or his and Shaun’s versions). “It’s all about using the best method possible.”
Question: What are the barriers to embedding fonts in Web pages? Josh: Mostly technical. Embedding a digital font lets the user distribute it. Type designers earn nearly all their income from licensing, and licensing could not be worked out. Most licenses forbid embedding of any kind, even in PDF. “Unfortunately, no one at the moment is speaking to the people who actually use fonts. We’re just software developers sitting in a room bitching.”
Mike: Wants an easy way for people who work on the Web to buy just a subset of a font, sort of like the iTunes Music Store where you can buy a song rather than an album. He wants a reasonable fee for a reasonable license.
Question from Brent: Has anyone done any testing on speed between client- and server-side font replacement? Shaun: No, but somebody is working on it. (Your scribe missed the name.)
Question: sIFR headlines tend to load last on a page, and that’s a usability issue. “It’s kind of a problem when headlines are the last things to load.” Shaun: There are technical ways to solve it (though not good ones). That will happen if you use images, too.
Mike: That’s also a bigger issue on Safari; he’s talked to Hyatt about it. They have to fire their replacement event onLoad because Safari begins to render the page before it’s fully loaded. It’s a tradeoff from the Safari developers. sIFR measures the boundaries of where the browser text should be and replaces it with a Flash block, so if those boundaries change, it takes longer to load sIFR. Person in audience suggests using CSS to style the headline in a font of similar measure so the difference in width is smaller.
Question: Any way to make sIFR text resizable? Mike: It is resizable on load and follows the zoom settings of your browser. (So I guess you resize your text and reload the page.) Mark Wubben is working on that.
Question from Jason: Could a central host of fonts be a good idea instead? (That could work for browsers you have to pay for and for embedded systems.) Josh likes the idea.
Question from woman in audience: Is teeny-tiny pixel text a trend or a long-term solution? They’re tuned for today’s large pixels compared to laser printers. Josh suspects it’s a trend, “a local hack, if you will. It’s certainly not a bad thing, and it’s resulted in some terrific typography, but pixels are eventually going to get so small that you can’t have a single-pixel stem.” You may have different tiers of information, like footnotes (at several millimetres) and headlines (much larger). Mike says you can design an outline font that looks like a pixel font, but pixel fonts “will always remind you of the ’90s… and is always going to remind us where we come from.”
Question from Andy Budd: What’s your favourite font? Shaun: FF DIN. Depends on the situation. Mike: For body copy, Lucida Grande (“I think it’s one of the most readable fonts in the world ever”); display, Mythos (“Do not use this for body copy”); print, Myriad (“One of the few sansserif fonts that I can find myself reading easily”). Josh: “My stock answer is ‘the next font,’ but I’d have to say Lexicon.” Comes in several hundred styles “but is strangely incomplete – they released an @ sign a couple of years ago…. Virtually unknown in North America, which means you can scoop the whole design world” if you start using it.
Question about subpixel anti-aliasing. Shaun: I haven’t explored it at all. Mike: It’s a bigger issue with small text. Let the user decide how they would like large passages of text rendered on their machine. Josh: Has “baked subpixel anti-aliasing into fonts in the past.”
Question: Are there any issues of transparency in sIFR? Mike: Yes, but it’s “local to plugins,” not sIFR, and mostly in Safari and older browsers and Opera. “I never recommend using transparency with sIFR, but some people have used it to fine effect.”
Mike hates every font manager out there. Shaun drags fonts in and out of his ~ font folder as necessary. Josh tried Suitcase for a while “and it was horrific,” since he installs so many fonts.