The scorpion express:
Thoughts on variable fonts

Have you heard the story of the scor­pion and the frog? It goes like this: a scor­pion wants to cross a river. But he can’t swim. So he asks a nearby frog to carry him across. The frog says, “how do I know you won’t sting me?” The scor­pion replies, “Don’t be silly. We’d both drown.”

Mol­li­fied, the frog in­vites the scor­pion onto his back. Halfway across the river, the scor­pion stings him. As the venom par­a­lyzes the frog, he says, “Why did you do that? Now we’ll both drown.”

With that, the scor­pion un­furls a set of high-tech mo­tor­ized wings and flies to the other side of the river. With his dy­ing breath, the frog says, “So you’re a rich scor­pion? Why did you need me at all?” The scor­pion says, “I may be a rich scor­pion—but I’m still a scorpion.”

If the end­ing sur­prises you, maybe you should spend more time on the com­mit­tees that make tech­nol­ogy standards.

Last week, an up­date to the 20-year-old Open­Type font stan­dard was an­nounced, of­fi­cially called Open­Type Font Vari­a­tions but more com­monly known as vari­able fonts. It’s be­ing dri­ven by the usual sus­pects—Google, Ap­ple, Mi­crosoft, and Adobe. Also par­tic­i­pat­ing are an as­sort­ment of in­de­pen­dent tool de­vel­op­ers, and—al­ways val­ued for their swim­ming skills—in­di­vid­ual type designers.

With­out di­min­ish­ing the ef­fort that’s been put into this new stan­dard, I’m not con­vinced there’s a plau­si­ble ra­tio­nale for it. It would im­pose sig­nif­i­cant costs on type de­sign­ers, pro­vide no ob­vi­ous ad­van­tage to our cus­tomers, and mostly ben­e­fit a small set of wealthy cor­po­rate sponsors.

Be­low, I’ll ex­plain my rea­son­ing. But agree or dis­agree, I hope other type de­sign­ers will give this pro­posed stan­dard the crit­i­cal scrutiny and re­flec­tion it de­serves. Be­cause if we end up at the bot­tom of the river, we won’t be able to say we didn’t know who was rid­ing on our back.

OT Font Vari­a­tions is an up­date to the Open­Type font for­mat that will al­low font files to con­tain mul­ti­ple sets of out­lines. To­day, a sin­gle font file can only con­tain one set of out­lines. So a font fam­ily with, say, weight and width vari­ants has to be ren­dered into a ma­trix of in­di­vid­ual fonts:

But un­der the new stan­dard, the width and weight vari­a­tions can be pack­aged into a sin­gle font file.

Fur­ther­more, cus­tomers will be able to in­ter­po­late be­tween styles. So rather than, say, a set of dis­crete weight options:

Weight be­comes con­tin­u­ously vari­able, and a cus­tomer can choose any­thing in between:

This idea is not new. In the early ’90s, Ap­ple and Adobe launched com­pet­ing font-in­ter­po­la­tion sys­tems. Ap­ple’s was called True­Type GX; Adobe’s was called Mul­ti­ple Mas­ter fonts. Ap­ple’s had the ad­van­tage of be­ing built into the Mac OS. Adobe’s had the ad­van­tage of be­ing sup­ported by Post­Script and PDF.

In fact, dur­ing 1993, I was one of sev­eral type de­sign­ers who worked with Matthew Carter on Skia, a font with weight and width vari­ants that Ap­ple com­mis­sioned to show off True­Type GX. In true time-is-a-flat-cir­cle fash­ion, Skia has a star­ring role in Mi­crosoft’s new OT Font Vari­a­tions white pa­per. (Pearl Jam, un­for­tu­nately, was not avail­able as an open­ing act.)

“So why haven’t we been us­ing this amaz­ing tech­nol­ogy for the last 20 years?” Be­cause it was a New­ton-es­que flop. Be­yond the Ap­ple–Adobe taffy pull, pro­fes­sional print­ers didn’t want to deal with more font headaches (in the ’90s, they al­ready had plenty). And cus­tomers were de­lighted to have any kind of se­lec­tion of dig­i­tal fonts—the in­cre­men­tal value of in­ter­po­lated styles was small. Thus, com­pa­nies that made page-lay­out pro­grams didn’t want to sup­port GX or MM ei­ther. Ul­ti­mately, too much cost and not enough benefit.

The les­son: if the cus­tomer doesn’t ben­e­fit, no one can.

The stag­na­tion of dig­i­tal-font for­mats is one of the most net­tle­some as­pects of type de­sign. Open­Type has now been with us for nearly 20 years (its par­ent, True­Type, has been around for 25). One of the dis­tin­guish­ing fea­tures of OT Font Vari­a­tions is that it’s not mean­ing­fully back­ward com­pat­i­ble with Open­Type. Es­sen­tially, it’s a new for­mat that pro­poses to even­tu­ally sup­plant to­day’s Open­Type font families.

(By the way, when I say the new fonts are not “back­ward com­pat­i­ble”, I mean that they won’t work with op­er­at­ing sys­tems and pro­grams that cur­rently sup­port Open­Type. I don’t mean that the new stan­dard will break ex­ist­ing Open­Type fonts.)

In one sense, it’s in­ter­est­ing to think that the fonts I ship to cus­tomers to­day could be in­stalled and used equally well on a Win­dows 95 ma­chine. But in most senses, it’s point­less. Be­cause in prac­tice, my fonts will never be used on com­put­ers from that era.

The con­verse is not true, how­ever. Fonts made 20 or more years ago are still us­able on to­day’s ma­chines. In fact, they make up the bulk of the Mono­type and Adobe font li­braries. Many re­main on to­day’s best­seller lists. So in fonts, un­like other cat­e­gories of soft­ware, type de­sign­ers have a pe­cu­liar prob­lem: our new work has to com­pete against decades of ac­cu­mu­lated competitors.

Is that a good thing? In terms of turn­ing over in­ven­tory, no. All tech­nol­ogy com­pa­nies de­pend on a cer­tain level of ob­so­les­cence. It cre­ates re­cur­ring rev­enue, of course. But it also helps avoid the es­ca­lat­ing costs and con­straints of back­ward compatibility.

More­over, this is a new sit­u­a­tion for type. Long ago, when type was made of wood and metal, it would nat­u­rally wear out with use, and need to be re­placed. In the 20th cen­tury, ad­vance­ments in type­set­ting tech­nol­ogy meant that font li­braries would have to be tossed out and re­placed with new for­mats. So ob­so­les­cence was al­ways in the mix. These days, dig­i­tal fonts seem end­lessly durable.

Still, these his­tor­i­cal com­par­isons are strained. Now 30 years old, dig­i­tal type­found­ing seems ma­ture rel­a­tive to other dig­i­tal tech­nol­ogy. But rel­a­tive to other type­found­ing tech­nol­ogy, it’s just a blip. For most of type­set­ting his­tory, be­ing a type­founder meant spend­ing your ca­reer near molten metal and car­cino­gens, not an Aeron chair and Keurig cof­fee pods.

Fur­ther­more, though in the past I’ve com­plained about the per­sis­tence of dig­i­tal fonts, it veers to­ward one of my least fa­vorite ar­gu­ments: the idea that a cre­ative per­son has a right to make a liv­ing from their work. Sorry, but no one does. The mar­ket—though ar­ti­fi­cial and im­per­fect—sets the rules. Other fonts ex­ist. Hun­dreds of thou­sands, in fact. As type de­sign­ers, we can ei­ther deal with that fact, or find some­thing else to do with our time.

To be fair, for­mat stag­na­tion isn’t all bad. As a re­sult of run­ning my own type foundry since 2011, I’ve learned first­hand that no mat­ter how much cus­tomers love fonts, they’re look­ing for a low-main­te­nance re­la­tion­ship. Pos­si­bly com­pli­cat­ing this re­la­tion­ship is the fact that fonts are called upon to per­form in a count­less com­bi­na­tions of op­er­at­ing sys­tems, type­set­ting pro­grams, and out­put de­vices. But in prac­tice, fonts just work. This is mirac­u­lous. As an in­de­pen­dent type de­signer, if I had to trou­bleshoot every setup sep­a­rately, I couldn’t stay in busi­ness. As it stands, I get al­most zero sup­port requests.

Why is this pos­si­ble? Be­cause font for­mats have been so sta­ble for so long. I have to imag­ine the cal­cu­lus is sim­i­lar for other in­de­pen­dent type de­sign­ers. So yes, for­mat stag­na­tion is bad for busi­ness in the sense of putting a ceil­ing on what we can ac­com­plish. But it’s ar­guably a nec­es­sary in­gre­di­ent for independence.

Com­ing full cir­cle, though it’s tempt­ing to fid­dle with the mile-high Jenga tower that com­prises to­day’s font-tech­nol­ogy stack, there are risks as well. Not of the “who moved my cheese” va­ri­ety, but rather of cus­tomers get­ting an­noyed with new­fan­gled fonts that over­promise and un­der­de­liver. As we’ve dis­cov­ered, when cus­tomers get frus­trated with fonts, they blame type de­sign­ers. Or they blame cor­po­ra­tions, who in turn blame type de­sign­ers. See the pat­tern? (The most com­mon re­sponse to my few re­quests for tech­ni­cal sup­port is “That prob­lem isn’t be­ing caused by my fonts, but I wish it were, be­cause then I could fix it for you.”)

The les­son: Be care­ful what you wish for.

Though nei­ther True­Type GX nor Mul­ti­ple Mas­ter fonts caught on, Adobe and Mi­crosoft col­lab­o­rated on the Open­Type spec­i­fi­ca­tion in the mid-’90s. (Ap­ple joined in later.) The im­pe­tus for this change was not aes­thetic, but prac­ti­cal: in or­der to sup­port the more com­plex writ­ten lan­guages that are com­mon out­side the US and Eu­rope, fonts and lay­out sys­tems needed to be more so­phis­ti­cated. And with­out sup­port for those lan­guages, no­body could sell their prod­ucts in those parts of the world.

Need­less to say, with huge eco­nomic in­cen­tives on the ta­ble, the new for­mat took off. Well, sort of: the parts of Open­Type that sup­ported new lan­guage sys­tems took off quickly. What didn’t were the parts, like Open­Type fea­tures, that im­proved ty­pog­ra­phy in cur­rent lan­guage systems.

For in­stance, though the Open­Type spec­i­fi­ca­tion was re­leased in 1996, Mi­crosoft Word didn’t sup­port Open­Type fea­tures un­til 2010. Ex­cel and Power­Point still don’t. Ap­ple stopped sup­port­ing Open­Type fea­tures in Pages for sev­eral years, in def­er­ence to its iOS ver­sion. And to­day’s crop of web browsers sup­port these fea­tures with dif­fer­ent lev­els of competence.

The les­son: when cus­tomers and cor­po­ra­tions def­i­nitely ben­e­fit (e.g., Open­Type lan­guage sup­port), de­sign­ers can too. When cus­tomers and de­sign­ers might ben­e­fit (e.g., Open­Type ty­pog­ra­phy fea­tures), cor­po­ra­tions are unreliable.

The last col­lab­o­ra­tion be­tween the scor­pi­ons and the frogs was WOFF (= Web-Only File For­mat). In 2009, the near-to­tal ab­sence of fonts on the web had be­come a source of frus­tra­tion for web de­sign­ers, who blamed type de­sign­ers for em­bar­go­ing the global font sup­ply, like some schem­ing Bond vil­lain (Dr. No Kern­ing? Glyphfinger?)

Some type de­sign­ers, wisely sens­ing an op­por­tu­nity for a diplo­matic res­o­lu­tion, pro­posed what be­came WOFF—a for­mat de­rived from Open­Type that would put fonts into browsers quickly while pro­tect­ing type-de­signer in­ter­ests (mostly, by mak­ing it harder to copy and use web­fonts on the desk­top). Browser mak­ers got in­volved. The W3C got in­volved. Soon, WOFF was off to the races. That was the good news.

The bad news? Once the dust set­tled, a type de­signer who was in­volved in the ef­fort wrote me that “the foundries who or­ga­nized be­hind this ef­fort didn’t get A SIN­GLE THING that they wanted”. (Em­pha­sis in orig­i­nal.) In other words, in terms of is­sues that mat­tered to de­sign­ers, WOFF was a waste of time—even though it was a de­signer-ger­mi­nated proposal.

I wasn’t in­volved in WOFF. But it’s no crit­i­cism of those who were to ob­serve that this kind of out­come has never been un­usual at the W3C, or within any stan­dards process. Those who can pay to pro­tect their in­ter­ests of­ten do. Those who can’t, don’t.

More­over, one of the by-de­sign side ef­fects of a stan­dards process is to achieve po­lit­i­cal peace with pos­si­ble fu­ture op­po­nents. When you pro­vide op­po­nents an op­por­tu­nity to be heard, they’re per­ma­nently dis­armed. How can any­one com­plain about the re­sult of a process that they par­tic­i­pated in?

The les­son: when cus­tomers and cor­po­ra­tions ben­e­fit, de­sign­ers should think twice about stand­ing in the way, be­cause our ne­go­ti­at­ing lever­age is limited.

WOFF2 is an up­date to WOFF that was first pro­posed in May 2014. Un­like WOFF Clas­sic, WOFF2 was not a col­lab­o­ra­tion be­tween type de­sign­ers and browser mak­ers. It was just some­thing Mono­type and Google wanted. And, since they’re both pay­ing mem­bers of the W3C, they got it.

Why did they want it? The only sig­nif­i­cant change in WOFF2 was that it added a new com­pres­sion scheme that can make font files smaller. Why did Google and Mono­type want smaller files? Be­cause they’re two of the three biggest providers of hosted web­fonts. (The other is Adobe, who sup­ported it quickly.) Cut your file sizes = cut your host­ing bills. Sim­ple. For cus­tomers and de­sign­ers, it was met with a shrug, since WOFF2 didn’t change any­thing in­side the font.

The les­son: when cor­po­ra­tions ben­e­fit, and cus­tomers and de­sign­ers are un­af­fected, they get what they want.

“But cus­tomers ben­e­fit from smaller file sizes too, be­cause that makes web pages faster.” Cer­tainly, that was true in 1996. And some web de­vel­op­ers per­sist with po­lit­i­cal ob­jec­tions. But with to­day’s faster con­nec­tions—even on mo­bile—op­ti­miz­ing for file size is less use­ful than ever.

Net­work la­tency—that is, the num­ber of re­quests a web page makes to var­i­ous servers mul­ti­plied by how long each takes to es­tab­lish—is the real buga­boo. In 1996, a web page might have made a few re­quests to down­load im­ages. These days, thanks to the noth­ing-but-ad­ver­tis­ing econ­omy of the web, a page might make “hun­dreds or even thou­sands of re­quests to fully load all of the ad­ver­tise­ments and an­a­lyt­ics”, ac­cord­ing to one study by Forbes.

For Google in par­tic­u­lar, we should shed a gi­ant croc­o­dile tear when­ever it con­cern-trolls us about file sizes on the web. YouTube (owned by Google) con­sumes an as­ton­ish­ing 18% of all in­ter­net band­width, sec­ond only to Net­flix (an eye-wa­ter­ing 37%). The file sizes of fonts—geez, that seems low on the list of the in­ter­net’s band­width problems.

FWIW, since 2013, Prac­ti­cal Ty­pog­ra­phy has been an on­go­ing ex­per­i­ment in ex­treme web­fonting. I’ve pushed about a megabyte of fonts to mil­lions of read­ers, who are us­ing all kinds of web browsers and plat­forms (in­clud­ing mo­bile). To­tal com­plaints I’ve re­ceived about page per­for­mance: zero. Of course, I don’t have ads or track­ers ei­ther. So it’s a ques­tion of priorities.

For rea­sons un­clear, this claim about net­work la­tency has al­ways pro­voked howls of out­rage among the web-dev Twit­terati. Folks, let’s work from ev­i­dence, not su­per­sti­tion. For ex­am­ple, here’s a quick test I did in Sep­tem­ber 2016, with home pages ranked in or­der of load time. As you can see, load time cor­re­lates more strongly with num­ber of re­quests than down­load size. And Prac­ti­cal Ty­pog­ra­phy beats every­one but the world’s biggest cor­po­ra­tion. Since I only pay $6 a month for host­ing, I can live with that:

websitebytesrequestsload time
apple.com1.9 MB470.62s
practicaltypography.com1.8 MB101.21s
medium.com1.4 MB551.37s
alistapart.com0.54 MB451.44s
stackoverflow.com0.36 MB431.60s
microsoft.com1.8 MB1191.87s
youtube.com2.1 MB1652.39s
yahoo.com2.6 MB1172.56s
amazon.com3.9 MB1282.58s
adobe.com2.0 MB2182.65s
nytimes.com5.4 MB5015.20s

The ba­sic im­prove­ment of­fered by vari­able fonts—in­ter­po­lated styles—has al­ready failed twice in the mar­ket. What’s the jus­ti­fi­ca­tion to try it all again?

With com­mend­able can­dor, type de­signer John Hud­son, who worked on OT Font Vari­a­tions, has tried to ad­dress this ques­tion. (John is a friend & a ter­rific type de­signer, so I’m go­ing to break with pro­to­col and not re­fer to him as “Mr. Hudson.”)

This time, the cor­po­ra­tions at the ta­ble—Ap­ple, Adobe, Google, Mi­crosoft—have “de­signed [the new stan­dard] col­lab­o­ra­tively” rather than push­ing mul­ti­ple stan­dards. That’s the good news.

The bad news is that un­like pre­vi­ous evo­lu­tions of the True­Type/Open­Type lin­eage, OT Font Vari­a­tions re­quires “sub­stan­tial up­dates” to op­er­at­ing sys­tems and ap­pli­ca­tions, and has “very lim­ited back­wards com­pat­i­bil­ity”. When we see those words, we should con­clude that things are about to get very ex­pen­sive for type de­sign­ers, and very bro­ken for customers.

But some­times this is the price of progress. What will be the ben­e­fit of all this up­heaval? Ac­cord­ing to John, “a big part of the an­swer is web­fonts, and the need for more com­pact and faster ways to de­liver dy­namic fonts for the Web”.

That sounds like WOFF2. Sure, cor­po­ra­tions that serve a lot of fonts over the net­work will al­ways want to make them smaller, thereby sav­ing money. But as we saw above, that’s not likely to ben­e­fit type de­sign­ers or customers.

The file-size sav­ings may be over­stated any­how. For in­stance, a Mi­crosoft man­ager gave the ex­am­ple that a “a con­ven­tional five-weight font fam­ily” that uses 656 KB as in­di­vid­ual fonts might only weigh 199 KB when repack­aged as a vari­able font. But have you ever seen a web­site that uses five weights of a sin­gle font? Not me. Two weights, how­ever, are very com­mon, and in that case, it’s not yet clear that the new for­mat con­fers any advantage.

What else? The new fontshave the po­ten­tial to en­able new kinds of ty­pog­ra­phy for elec­tronic doc­u­ments, re­spon­sive to things like de­vice ori­en­ta­tion or even view­ing distance”.

That sounds like True­Type GX or Mul­ti­ple Mas­ter fonts, which went nowhere with cus­tomers, or Open­Type ty­po­graphic fea­tures, which went nowhere with cor­po­ra­tions who had to build sup­port for them. First, since we’re be­ing can­did, most pro­fes­sional graphic and web de­sign­ers don’t care much about fonts at all. (For in­stance: Ox­ford Uni­ver­sity Press, one of the most re­spected book pub­lish­ers, nev­er­the­less sets nearly all its books in Min­ion, a font that every Adobe cus­tomer has got­ten for free for 20 years.) Sec­ond, those that do care about fonts are look­ing for a low-main­te­nance re­la­tion­ship, not a lab-mon­key ex­pe­ri­ence. Third, think of the strug­gles that it took to bring fonts to the web at all. Hav­ing just reached the point where web­fonts have been ac­cepted within the main­stream web, what’s the virtue of break­ing every­thing again? “Po­ten­tial” is not quite enough.

What’s more, font li­cens­ing as a busi­ness can only be as healthy as the in­dus­try it serves. In this case, if vari­able fonts are largely be­ing pitched as an im­prove­ment for the web, we ought to ask: how much money is there is in web pub­lish­ing? The an­swer: not damn much. (My font-rev­enue re­ports for the first six years of the web­font era back that up.) To my mind, that’s an­other big dif­fer­ence from Open­Type in the mid-’90s, which was in­tro­duced into what was still a healthy print-pub­lish­ing in­dus­try. (My font-rev­enue re­ports back that up too.) As an in­de­pen­dent de­signer, I can’t do much with tech­no­log­i­cal po­ten­tial un­less it also im­plies rev­enue potential.

What else? Vari­able fontspro­vide sig­nif­i­cant ad­van­tages for em­bed­ding fonts in de­vices, es­pe­cially for East Asian (CJK) and other fonts with very large glyph sets and char­ac­ter coverage”.

That sounds like the lan­guage-sup­port as­pects of Open­Type. If broad­en­ing the mar­ket for tech­nol­ogy prod­ucts is the best ar­gu­ment for re­vis­ing font stan­dards, then this will likely end up be­ing the best ra­tio­nale for vari­able fonts. But for type de­sign­ers who work with West­ern scripts—which de­scribes the ma­jor­ity of pro­fes­sional type de­sign­ers, me in­cluded—it doesn’t move the nee­dle. The orig­i­nal Open­Type spec let us ex­pand our reach into other West­ern-script mar­kets (e.g., Es­peranto and Mal­tese), be­cause those lan­guages were ba­si­cally sim­i­lar to lan­guages we al­ready sup­ported (e.g., Eng­lish and French). But fonts for East Asian lan­guages are a com­pletely dif­fer­ent ket­tle of fish.

In say­ing that, I take noth­ing away from the type de­sign­ers who make East Asian fonts. For them, vari­able fonts might be huge. Won­der­ful—may they profit greatly. In time, the West­ern cor­ner of the type mar­ket might even look com­par­a­tively small. Won­der­ful—I will be do­ing some­thing else by then.

As a de­signer, I’m sure vari­able fonts would be en­ter­tain­ing to work with. But from what I can see, there’s no ev­i­dence that my cus­tomers are in­ter­ested in that ca­pa­bil­ity—let alone pre­pared to open their wal­lets for it. There­fore, the ben­e­fits don’t out­weigh the sig­nif­i­cant cost of los­ing back­ward com­pat­i­bil­ity (mean­ing, pack­ag­ing font fam­i­lies into a new for­mat that’s not sup­ported by any plat­forms or ap­pli­ca­tions cur­rently in use).

On the con­trary, based on his­tor­i­cal pat­terns, it’s easy to imag­ine a sce­nario where I spend a lot of time de­vel­op­ing vari­able fonts that ba­si­cally no­body wants, and the few who do buy them dis­cover that they rarely work as ad­ver­tised. If this seems sullen or hard-hearted, no—type de­sign­ers have to make these choices all the time. I of­ten get asked to sup­port new char­ac­ter sets—Greek, Russ­ian, Viet­namese, math & sci­ence, and so on. These would all be fan­tas­tic projects. But I can’t reach enough pay­ing cus­tomers to make any of them worth­while. If type de­sign were my hobby, I’d be de­lighted to pur­sue these projects. But since it’s my work, I have to pass.

John also sum­ma­rizes how far the cor­po­rate par­tic­i­pants have pro­gressed with vari­able fonts. With this, we can make some ed­u­cated guesses about what they’re get­ting out of it:

  1. The “Win­dows en­gi­neer­ing team at Mi­crosoft” is plan­ning to add sup­port for vari­able fonts in 2017. That sounds great, un­less you know any­thing about the cul­ture at Mi­crosoft, where every di­vi­sion sets their own de­vel­op­ment agenda, and they ba­si­cally all hate each other. Of­fice is in a sep­a­rate di­vi­sion from Win­dows, which is why Open­Type sup­port in Of­fice has lagged far be­hind OT sup­port in Win­dows: if a fea­ture doesn’t sell more copies of Of­fice, they’re not much in­ter­ested in im­ple­ment­ing it.

    Un­like Of­fice, the Edge web browser is part of the Win­dows di­vi­sion, so it’s likely to sup­port vari­able fonts sooner. John notes that the browser team is also work­ing on a “for­mal pro­posal for sup­port of vari­able fonts in Cas­cad­ing Style Sheets (CSS) for the Web”. Again, that sounds great, un­less you know any­thing about the cul­ture at Mi­crosoft. They’ve al­ways been ea­ger to make web stan­dards, and never been ea­ger to sup­port them.

    I pre­dict: Don’t bet against his­tory. Re­gard­less of Mi­crosoft’s par­tic­i­pa­tion in vari­able fonts, Of­fice may not mean­ing­fully sup­port it for decades; Edge may not mean­ing­fully sup­port the web-stan­dard ver­sion of it ever.

  2. Ap­plechar­ac­ter­is­ti­cally, [is] least forth­com­ing about fu­ture plans, but they have a head start on vari­able font sup­port in their True­Type GX in­fra­struc­ture”. GX dates from 1993, so I’m skep­ti­cal how much of a tech­ni­cal “head start” Ap­ple really has, when every­thing else in the Ap­ple ecosys­tem is com­pletely dif­fer­ent. (That was also the last time Ap­ple took any kind of lead­ing role in font tech­nol­ogy.) Ap­ple’s busi­ness has also com­pletely changed since 1993. Then, they were pri­mar­ily a desk­top-com­puter com­pany; now, they’re pri­mar­ily a mo­bile-phone com­pany. Their sup­port for vari­able fonts likely re­duces to the ques­tion of whether it will help them sell more iPhones in China and else­where in Asia.

    I pre­dict: Ap­ple will fol­low their orig­i­nal Open­Type strat­egy—let every­one else go first, and im­ple­ment the stan­dard only if it proves to be valu­able in the mo­bile-phone market.

  3. Adobe’s font-tech­nol­ogy team is up­dat­ing its tools for font de­vel­op­ers, but there are “no de­tails about sup­port for vari­able fonts in Adobe’s ap­pli­ca­tion suite”. As with Mi­crosoft, Adobe’s ap­pli­ca­tion group is a dif­fer­ent di­vi­sion with dif­fer­ent pri­or­i­ties. I don’t think they hate each other as much. But Cre­ative Suite Cloud is Adobe’s cash cow. They’re not go­ing to let vari­able fonts mess with those teats. Adobe also care­fully pro­tects its re­la­tion­ships with big pub­lish­ers and print­ers. They’re not go­ing to let vari­able fonts foul them up (es­pe­cially af­ter re­cent mis­fires like Adobe Dig­i­tal Edi­tions).

    I pre­dict: Adobe will ex­ert min­i­mal ef­fort to sup­port vari­able fonts for West­ern lan­guages. Though con­sis­tent with ’90s nos­tal­gia, they will re­lease ver­sions of Myr­iad and Min­ion in the new for­mat (and they bet­ter bring Pearl Jam to Adobe MAX this year). Oth­er­wise, they’ll re­serve most of their ef­fort for sup­port­ing vari­able fonts in East Asian fonts.

  4. Fi­nally, Google has ap­par­ently been work­ing on vari­able-font tech­nol­ogy for two years, get­ting it ready for “Google Chrome [and] the Google Fonts web­font plat­form”. That’s no sur­prise. First, we know Google saves money with small files. Sec­ond, Google Fonts shifted its fo­cus to East Asian scripts a while ago, hav­ing con­quered the west. Third, be­cause Google spends more freely than Adobe or Mi­crosoft, and they’ve al­ready amassed an army of un­der­paid type de­sign­ers, these de­sign­ers will prob­a­bly be de­ployed to con­vert many ex­ist­ing Google fonts to the new for­mat. These fonts will be just as ter­ri­ble as the orig­i­nals. But since Google has no cus­tomers in pro­fes­sional pub­lish­ing, and no taste, every­thing will still be awesome.

  5. I pre­dict: Google will con­tinue to be the most vig­or­ous early adopter of vari­able fonts, but the ben­e­fits will be re­stricted to the Googleverse.

I could’ve dis­cussed prob­lems with spe­cific tech­ni­cal as­pects of vari­able fonts (e.g., if this for­mat won’t be back­ward com­pat­i­ble, why stick with a bi­nary-ta­ble struc­ture held over from the ’80s? Or why is in­ter­po­la­tion the only kind of trans­for­ma­tion supported?)

But there’s no point. Mar­ket con­sid­er­a­tions will al­ways over­ride tech­ni­cal con­sid­er­a­tions. Thus, to fore­cast what will hap­pen with vari­able fonts, it’s far more im­por­tant to con­sider the mar­ket in­ter­ests of the stake­hold­ers, rather than the par­tic­u­lars of the for­mat. The how is less im­por­tant than the what and why.

As John says in his ar­ti­cle, “font mak­ers ... have a long col­lec­tive mem­ory” about these things. (True.) The down­side is that it ex­tends this kind of ar­ti­cle to epic length. (Sorry.) The up­side is that if you want to know what hap­pens in these sit­u­a­tions, there’s a decades-long trail of bread­crumbs for you to study. I’m sure this kind of home­work will seem crusty and te­dious to those who en­tered type de­sign more re­cently than I did. (Sorry about that too.)

But look across this his­tory, and a sim­ple prin­ci­ple emerges. It’s a cliché, but as usual, the cus­tomer is al­ways right. When type de­sign­ers have ac­cepted what cus­tomers want (e.g., bet­ter Open­Type lan­guage sup­port) the mar­ket has re­warded us. When type de­sign­ers have re­sisted what cus­tomers want (e.g., easy ac­cess to web­fonts) the mar­ket has pun­ished us.

Why does this prin­ci­ple work? First, be­cause cus­tomers pay us—duh. Sec­ond, be­cause the cor­po­rate par­tic­i­pants in the type mar­ket have to serve those same cus­tomers. Ul­ti­mately, they can’t af­ford to alien­ate cus­tomers any more than we can. So where the cus­tomers lead, every­one follows.

The corol­lary to this prin­ci­ple is that when cus­tomer de­mand is re­moved from the pic­ture, things get murky. In this case, we might ask: if the vari­able-fonts stan­dard has been built out­side the prac­ti­cal in­flu­ence of cus­tomer de­mand, then who does it serve?

The ide­al­ist might say that a stan­dard pro­vides a level play­ing field for mar­ket par­tic­i­pants. But in prac­tice, stan­dards tend to re­flect the in­ter­ests of who­ever has the most weight to throw around.

The ide­al­ist might also say that new stan­dards are nec­es­sary for tech­no­log­i­cal progress. But in the mar­ket, the costs of progress end up be­ing strictly weighed against the benefits.

Ul­ti­mately, I see no ev­i­dence that vari­able fonts are some­thing my cus­tomers want. But I see plenty of ev­i­dence that it will help Ap­ple, Google, et al. sell more prod­ucts in cer­tain coun­tries, or save money on band­width. I’m sure these cor­po­ra­tions would love more help from type de­sign­ers to make their tech­nol­ogy look good. And if you want to go for a swim, don’t let me stop you. But this frog is go­ing to stay right here on the side of the river, where it’s sunny and dry.

—Matthew But­t­er­ick
20 Sept 2016

Mi­crosoft: The newest Win­dows 10 up­date con­tains an “ini­tial im­ple­men­ta­tion” of vari­able fonts. (Trans­la­tion: buggy and in­com­plete.) But: no sup­port yet in Of­fice or the Edge web browser. This, even though Mi­crosoft claimed last year that it would “sup­port vari­able fonts on all its prod­ucts” by the end of 2017. Mi­crosoft did man­age to ship its first vari­able font, Bahn­schrift. Over­all: about what I expected.

Ap­ple: Since Sep­tem­ber 2016, a hand­ful of spe­cial-pur­pose Ap­ple sys­tem fonts have re­lied on vari­able fonts. As of last month, the Sa­fari web browser sup­ports vari­able fonts (on both iOS and Mac OS). But: still no sup­port for vari­able fonts in desk­top type­set­ting ap­pli­ca­tions (like Pages). Nor any show­case vari­able fonts in the OS. Over­all: Ap­ple is drag­ging its feet less than I ex­pected. But they’re not at the front of the pack either.

Adobe: Af­ter I pub­lished the orig­i­nal ver­sion of this piece, an Adobe em­ployee told me that “for us it really is all about the vari­a­tion, not the file size sav­ings ... I ex­pect you’ll be pleas­antly sur­prised by sup­port in Adobe apps”. A year later, some progress has emerged. Vari­able fonts are sup­ported in the newest ver­sions of Pho­to­shop and Il­lus­tra­tor. But: so far, that sup­port is buggy and in­com­plete. And no sup­port at all in Adobe’s flag­ship page-lay­out pro­gram, In­De­sign, used by a huge num­ber of pro­fes­sional de­sign­ers. Per­haps most strangely, Adobe hasn’t re­leased any show­case vari­able fonts, save for some “Con­cept” fonts (trans­la­tion: buggy and in­com­plete). And none of those are East Asian fonts. Over­all: even lazier than I predicted.

Google: Sup­port for vari­able fonts will be in­cluded in the Chrome 62 browser, which will start per­co­lat­ing out to users this month. Google also funded the de­vel­op­ment of two free vari­able fonts (#1, #2) that are, um, in­ter­est­ing tech­ni­cal demos but not ex­actly prac­ti­cal. But: Google has not made any vis­i­ble push to con­vert fonts in its ex­ist­ing li­brary to use vari­a­tions tech­nol­ogy. Over­all: I pre­dicted Google would be out in front. But they’ve put in less ef­fort than I ex­pected. Sorry to be the party pooper.

Taken to­gether, a mediocre ef­fort. So far, vari­able fonts are not emit­ting the aroma of ag­gres­sive ac­tion, but rather the milder scent of bets be­ing hedged.

The bright­est sign is the web-browser sup­port. Chrome and Sa­fari are the most pop­u­lar web browsers (the oth­ers are lin­ger­ing in the sin­gle dig­its). Given that it took about 15 years to get web­fonts in browsers to be­gin with, that seems like rel­a­tively en­cour­ag­ing mo­men­tum. Still, even in the best case, it takes months for newer browser ver­sions to dis­place the old.

But the dark­est sign is the lack of use­ful fonts built with vari­able fonts tech­nol­ogy. Google and Adobe have made some demo fonts. But no­body is ship­ping or show­ing off solid, us­able vari­able fonts. In fact, one vari­able-font demo site is so starved for demo fonts that it in­cludes fonts orig­i­nally made nearly 25 years ago to show off Ap­ple’s True­Type GX tech­nol­ogy (con­verted to the new format).

I’ve heard this called a chicken-and-egg prob­lem: type de­sign­ers don’t have any in­cen­tive to make vari­able fonts un­til ap­pli­ca­tion sup­port is more wide­spread. Mean­while, ap­pli­ca­tion pro­gram­mers don’t have any in­cen­tive to up­date their pro­grams un­til there’s more fonts.

But this over­looks a key fact: this isn’t some grass­roots cam­paign. The tech­nol­ogy sup­port­ing vari­able fonts is sup­pos­edly backed by four of the biggest tech com­pa­nies in the world. They have the money and staff to cap­i­tal­ize the ef­fort. If they want to.

Do they? The mid­dling progress dur­ing the first year of vari­able fonts leaves me even more con­fused about how these com­pa­nies plan to ben­e­fit. I orig­i­nally fig­ured it had some­thing to do with re­duc­ing the costs of serv­ing web­fonts, es­pe­cially for East Asian scripts. Adding sup­port to web browsers is a nec­es­sary first step. But the lack of ac­tual show­case fonts—and sim­ple demo web­sites us­ing them—is ex­tremely weird.

And demos count. An old mar­ket­ing maxim holds that ef­fec­tive ad­ver­tis­ing doesn’t talk about the shovel—it talks about the hole in the ground. This is why demos are his­tor­i­cally im­por­tant in the tech world: they help il­lus­trate the ben­e­fit of us­ing the tech­nol­ogy, which in turn per­suades cus­tomers that a) it works and b) they want it.

Based on their weak out­put in year one, Mi­crosoft, Ap­ple, Adobe, and Google still seem un­able to ar­tic­u­late what cus­tomer prob­lems will be solved by vari­able fonts. If that re­mains true, this tech­nol­ogy will have a short ride to the bot­tom of the river, com­ing to rest near the moss-cov­ered skele­tons of Mul­ti­ple Mas­ters and True­Type GX.

In Oc­to­ber 2017, I wrote that dur­ing the 12 months af­ter OT Vari­able Fonts were an­nounced (in Sept 2016) we hadn’t seen “ag­gres­sive ac­tion” from the four main cor­po­rate spon­sors of vari­able fonts (= Mi­crosoft, Ap­ple, Adobe, and Google). Rather, I de­tected only the “milder scent of bets be­ing hedged”.

In year two, this al­ready lan­guid pace has slowed to a crawl. Af­ter the first year, the bright­est sign for vari­able fonts was “web-browser sup­port”. Since then, re­mark­ably lit­tle has hap­pened, as mea­sured by shipped prod­ucts. The de­clin­ing for­ward mo­men­tum that once sig­naled “bets be­ing hedged” now looks more like tents be­ing folded:

Mi­crosoft: the Edge browser now sup­ports vari­able fonts. But no sign of sup­port in MS Of­fice. Nor any new show­case fonts.

Ap­ple: no sig­nif­i­cant news that I can de­tect. Still no sup­port for vari­able fonts in desk­top apps like Pages or Num­bers. No new show­case vari­able fonts in the OS.

Adobe: no news since last Oc­to­ber. In par­tic­u­lar, still no sup­port for vari­able fonts in In­De­sign, Adobe’s flag­ship type­set­ting pack­age. Adobe has con­tin­ued to plod along with its ex­per­i­men­tal vari­able fonts, though one of the “cur­rent lim­i­ta­tions” is that the OTF ver­sions “can­not be dis­played by ma­cOS or Windows”.

Google: AFAICT the num­ber of vari­able fonts re­leased this past year via Google Fonts is zero. No sup­port in Google apps (Docs, Slides) ei­ther. Once again, sorry to be the party pooper.

Though they weren’t part of the core vari­able-fonts con­sor­tium, what about our friends at Mono­type? They re­leased a demo ver­sion of FF Meta Vari­able. And also—nope, that’s it. One demo font.

Mean­while, work on the OT Font Vari­a­tions stan­dard it­self has hit at least one ma­jor snag, re­lated to sup­port for what are known as “vir­tual axes”. With­out delv­ing into the de­tails, even John Hud­son, orig­i­nal evan­ge­list for vari­able fonts, has con­ceded that the is­sue is suf­fi­ciently im­por­tant that “a lot of [work with vari­able fonts] is ef­fec­tively on hold” un­til the prob­lem is re­solved. Not ex­actly the turbo boost that vari­able fonts could’ve used.

If you’re ea­ger to try vari­able fonts, the good news is that the Axis-Praxis vari­able-fonts demo site has con­tin­ued to grow. It’s been joined by the new Most of the fonts on these sites have been made by in­de­pen­dent type de­sign­ers, who have done more than any­one else to churn out en­ter­tain­ing demo fonts. I com­pli­ment all of you. A lot of these vari­able fonts are very clever and well made.

And yet. This is only “good news” in a nar­row sense. More broadly, it ful­fills my orig­i­nal pre­dic­tion from two years ago: that the cor­po­rate spon­sors of vari­able fonts (= Mi­crosoft, Ap­ple, Adobe, Google) would min­i­mize their ef­fort. Rather than stick out their own necks, they’d let in­de­pen­dent type de­sign­ers do­nate their time to make vari­able fonts tech­nol­ogy look good.

Last year, my friend S. said “MB, you’re be­ing too harsh. There’s a lot of ac­tiv­ity hap­pen­ing be­hind the scenes.” Per­haps. But what’s hap­pen­ing in front of the cur­tain also counts. No­tably, af­ter two years, cus­tomers aren’t pay­ing for vari­able fonts. The whole no­tion of a vari­able-fonts econ­omy is still the­o­ret­i­cal. This key prob­lem re­mains unaddressed.

Though not un­ex­am­ined. Type de­signer Jo­hannes Neumeier con­sid­ered an in­ter­est­ing thought ex­per­i­ment: if us­able vari­able fonts ac­tu­ally ex­isted, what would they need to cost? He con­cluded—and I agree—that vari­able fonts are “in­com­patib[le] with cur­rent li­cens­ing mod­els”. Cus­tomers won’t want to pay the same for one vari­able font as they would for, say, a large set of dis­crete styles, even if the vari­able font cov­ers the same de­sign space. Though vari­able fonts have been touted as a bold new fu­ture for type de­sign, it’s equally pos­si­ble that they’ll lead to a new down­ward spi­ral of more work for less pay.

Still, to give my critic S. her due: I agree I was too harsh. I doubt this doom-and-gloom sce­nario will come to pass. Why? Be­cause it’s ap­par­ent that even if you have es­sen­tially un­lim­ited time and money—e.g., you are Mi­crosoft, Ap­ple, Adobe, or Google—it’s still painfully dif­fi­cult to make vari­able fonts that work well. As proof, con­sider that other than a hand­ful of spe­cial-pur­pose OS fonts, not one of these cor­po­rate spon­sors has shipped a no-caveats vari­able font. (Mean­ing, some­thing that is com­plete and us­able and not la­beled a “demo” or “con­cept”, etc.) I re­main du­bi­ous that in­de­pen­dent type de­sign­ers should do for huge cor­po­ra­tions what these cor­po­ra­tions won’t (or can’t) do for themselves.

At some point, a well-in­formed source told me that the cor­po­rate spon­sors of vari­able fonts were “ba­si­cally mo­ti­vated by one thing only: smaller font foot­print”. If so, it’s start­ing to look like vari­able fonts are the world’s most la­bor-in­ten­sive—and there­fore least prac­ti­cal—file-com­pres­sion scheme.

Mi­crosoft: No new ap­pli­ca­tions re­leased with sup­port for vari­able fonts since the Edge web browser in 2018. (We’ll leave aside the fact that Edge re­mains even less pop­u­lar than In­ter­net Ex­plorer.) Mi­crosoft Of­fice? Surely you jest. What about new vari­able fonts? Nope.

Adobe: The good news—in No­vem­ber, In­De­sign in­tro­duced sup­port for vari­able fonts. The bad news—early re­ports sug­gest that not all vari­able fonts work equally well, or more omi­nously, that Adobe might be priv­i­leg­ing its new CFF2 font for­mat. As for new vari­able fonts: none that I’ve been able to find.

Ap­ple: No up­dates for the iWork of­fice suite (Pages / Num­bers / Keynote) with vari­able fonts. Nor has Ap­ple re­leased any new vari­able fonts of its own.

Google: In Oc­to­ber, Google up­dated its font API to sup­port vari­able fonts, in­clud­ing 10 vari­able up­grades of ex­ist­ing Google fonts. Demo here.

In gen­eral, my orig­i­nal pre­dic­tions from 2016 are hold­ing up pretty well. Mi­crosoft, Ap­ple, and Adobe are do­ing as lit­tle as pos­si­ble to sup­port vari­able fonts. (Though I was wrong that the Mi­crosoft Edge browser would never sup­port vari­able fonts—it does.) Mean­while, Google has in­deed be­come the “most vig­or­ous early adopter of vari­able fonts, but the ben­e­fits [are] re­stricted to the Google­verse.” The rea­son be­ing that as the world’s biggest server of web­fonts, Google po­ten­tially saves money with vari­able fonts.

Still, 3+ years into the vari­able-font era, it re­mains un­clear how & when the econ­omy of vari­able fonts is sup­posed to emerge. Mean­ing, if type de­sign­ers can’t make vari­able fonts at rea­son­able cost and sell them at a profit to font cus­tomers—who have to get their money’s worth—the vari­able-font mar­ket will re­main an ex­per­i­men­tal niche.

Here, we would’ve hoped to see Adobe lead­ing by ex­am­ple, as the lead­ing pur­veyor of tools to the pro-de­sign mar­ket. The sup­port for vari­able fonts in In­De­sign (how­ever spotty) is en­cour­ag­ing. But the to­tal lack of mo­men­tum on the type-de­vel­op­ment side is not. For in­stance, two years af­ter the an­nounce­ment of vari­able-font demos of Acumin, Min­ion, and Myr­iad, no ap­par­ent progress has been made.

This past year, a num­ber of in­de­pen­dent foundries—e.g., Dal­ton Maag and Font­smith—have started sell­ing what may be the first no-ex­cuses vari­able-font fam­i­lies. My com­pli­ments to you. (FWIW, Dal­ton Maag has also done cus­tom font work for Google.)

In his own year-end re­port, vari­able-font op­ti­mist (and some­time Google con­sul­tant) Ja­son Pa­men­tal ac­knowl­edges the on­go­ing “frus­tra­tion about lack of sales” among type de­sign­ers and sug­gests “of­fer­ing [cus­tomers] a deal ... in re­turn for hav­ing a case study.” You mean—if we work for free, it’ll be great ex­po­sure? Where have I heard this be­fore? As I said when I first wrote this piece, there’s lit­tle money be­ing made in web pub­lish­ing. There­fore there’s also lit­tle money be­ing put into web-pub­lish­ing tools, in­clud­ing web­fonts. The idea that vari­able fonts can break even pri­mar­ily as a web for­mat is mag­i­cal thinking.

Still, if type foundries and their cus­tomers don’t adopt vari­able fonts, the for­mat won’t nec­es­sar­ily die. For in­stance, vari­able fonts could mostly be­come known as a fea­ture of Google Fonts, even as the broader pos­si­bil­ity of the for­mat fades away. If vari­able fonts save Google enough money, Google has am­ple in­cen­tive to keep them alive.

I men­tioned Google’s con­nec­tion to Dal­ton Maag and Mr. Pa­men­tal. Over the last 10 years—and es­pe­cially since the an­nounce­ment of vari­able fonts—Google has bought a lot of in­flu­ence in what is the rel­a­tively small world of ty­pog­ra­phy. Much more than Ap­ple, Mi­crosoft, and Adobe have done.

Early on, Google built its free-font li­brary by work­ing with type de­sign­ers who weren’t typ­i­cally in­volved with the pro­fes­sional type world. Since then, Google has learned to spread money around more lib­er­ally. To my mind, this cam­paign has got­ten pro­fes­sional type de­sign­ers and ty­pog­ra­phers to take Google more se­ri­ously. Though per­haps at the cost of be­ing less care­ful about scru­ti­niz­ing Google’s motives.

To be fair, Google has done some pos­i­tive things for ty­pog­ra­phy. As I’ve said be­fore, I think Google’s pro­mo­tion of web­fonts helped spur ac­cep­tance within main­stream web de­sign. They’ve also con­tributed some use­ful open-source soft­ware for mak­ing fonts.

But for Google, fonts are strictly a tool to en­hance their core busi­ness of sur­veil­lance and ad­ver­tis­ing. And the only price Google cares about is free. No one at Google has ever sold a font li­cense, or ever will. So the idea that there could be deeper com­mon cause be­tween Google and pro­fes­sional type de­sign­ers is strained. Maybe like the com­mon cause be­tween dairy farm­ers and cows. Or scor­pi­ons and frogs.

I know that some think this ar­ti­cle is just a mul­ti­year se­ries of pes­simistic put-downs. But truly—I would’ve been happy to see vari­able fonts suc­ceed. As I said in 2016, “I’m sure vari­able fonts would be en­ter­tain­ing to work with.” If cus­tomers would pay for them. If these fonts could be made with rea­son­able time and ef­fort. And maybe there are niches of the type mar­ket where these two con­di­tions will turn out to be true.

Para­met­ric type de­sign is ar­guably the grail—or maybe white whale—of the dig­i­tal-type era. I would’ve been de­lighted to dis­cover that the (very wealthy cor­po­rate) stake­hold­ers in vari­able fonts had, af­ter 40+ years of at­tempts by oth­ers, sorted out these is­sues. By and large, I don’t think they have. In 1977, Don­ald Knuth’s META­FONT sys­tem showed off the ty­po­graphic promise of pa­ra­me­teric type. To­day, it seems the po­ten­tial of this grand idea is be­ing flat­tened into a way of in­creas­ing Google’s prof­its. What an accomplishment.

Mi­crosoft: No new vari­able fonts. No sup­port in Of­fice apps for vari­able fonts. The dis­cus­sion area Mi­crosoft opened in 2017 for vari­able-font de­vel­op­ers has been es­sen­tially dead for two years.

Ap­ple: No new vari­able fonts. Still no sup­port in iWork apps for vari­able fonts.

Adobe: Re­leased sev­eral “con­cept” vari­able fonts, in­clud­ing Acumin, Min­ion, and—oops, my mis­take. That all hap­pened in 2017. None of those fonts ever moved past the con­cept des­ig­na­tion, so as of 2020, Adobe’s cat­a­log of fin­ished vari­able fonts still stands at a per­fect zero. No new app support.

Google: The only mem­ber of the orig­i­nal vari­able-fonts con­sor­tium that has re­leased a big batch of vari­able fonts at a steady pace. Though this isn’t sur­pris­ing—from the be­gin­ning, vari­able fonts have been pri­mar­ily a Google-ger­mi­nated and Google-dom­i­nated project.

Don’t take my word for it:

  1. This year we got a first-hand ac­count of the ori­gin of vari­able fonts from for­mer Google en­gi­neer Be­hdad Es­fah­bod, who takes credit for “se­cretly” in­vent­ing vari­able fonts:

    “I stud­ied Ap­ple’s & Adobe’s sep­a­rate failed at­tempts at [ad­justable-font tech­nol­ogy] from the 90s, picked the bet­ter one, and started work on ex­pand­ing it to Open­Type ... While [my team of Google en­gi­neers] worked on our im­ple­men­ta­tion, Mi­crosoft formed a work­ing group of them, us, Ap­ple, and Adobe, or the Big Four as the group be­came to be known in the font in­dus­try. The group started of­fi­cially but se­cretly work­ing on adding run­time font in­ter­po­la­tion to Open­Type. My whitepa­per formed the blue­print of the work. The group met face-to-face each month, and six months later, on Sep­tem­ber 14, 2016 at the ATypI con­fer­ence in War­saw, the four of us com­pa­nies an­nounced Open­Type Font Vari­a­tions as part of Open­Type 1.8.”

  2. Re­cently, Bruno Maag, whose com­pany has done font-de­sign work for Google and thus has bet­ter in­for­ma­tion than me, said that “other than Google no­body wanted [vari­able fonts] in the first place. The sole rea­son it was pushed for so hard is for re­duc­ing web load when hav­ing mul­ti­ple styles of a font fam­ily in a website.”

Still, all that would be ir­rel­e­vant ori­gin-story de­tail if type users had em­braced buy­ing & us­ing vari­able fonts. Have they? I can’t really prove a neg­a­tive. But ev­i­dence of this em­brace is scant:

  1. Mono­type, owner of the world’s biggest type li­brary, main­tains a page of mar­ket­ing ma­te­r­ial about vari­able fonts that tellingly in­cludes no cus­tomer uses, aside from “the world’s first vari­able font logo”.

  2. Sim­i­larly, Adobe Max is Adobe’s an­nual mar­ket­ing event. If vari­able fonts were a big deal to Adobe or its cus­tomers, we’d ex­pect to see at least a few ses­sions about that, right? But no—there was only one. and it wasn’t even given by an Adobe designer.

This is my fifth time writ­ing about vari­able fonts. I won’t say it’ll be my last. But it’s got­ten bor­ing. Noth­ing is happening.

We’re more than four years into the vari­able-fonts era. The hon­ey­moon is def­i­nitely over. If you still think that wide­spread ac­cep­tance of vari­able fonts is just around the cor­ner, the bur­den of proof is now upon you. Oth­er­wise, I agree with Bri­jan Pow­ell that vari­able fonts are turn­ing out to be “the vocoder of graphic design.”

In the de­tails, not all my pre­dic­tions from 2016 were ac­cu­rate. But in the large, they were. In­clud­ing my main the­sis: that big tech com­pa­nies would in­duce in­de­pen­dent type de­sign­ers to make vari­able fonts look good—mostly for free—while type cus­tomers would ig­nore them. And that Google would be the ma­jor beneficiary.

Still, it’s only been a year or so since Google added vari­able fonts to its li­brary. Dur­ing that time, Google’s own font-de­vel­op­ment ef­forts seem to have slowed con­sid­er­ably. So Google’s com­mit­ment to vari­able fonts re­mains an open question.

When we con­sider that Google has a record of vig­or­ously mur­der­ing its un­der­per­form­ing projects, the likely endgame comes into view. As the fi­nan­cial ev­i­dence piles up, Google will be forced to con­clude that vari­able fonts don’t ac­tu­ally save money—be­cause Google hasn’t ag­gres­sively mi­grated its users, and from every ac­count I’ve heard, vari­able fonts are painfully dif­fi­cult and ex­pen­sive to make. At that point, Google will aban­don vari­able fonts al­to­gether. Af­ter that, Mi­crosoft, Ap­ple, and Adobe will breathe a sigh of re­lief and of­fi­cially dep­re­cate vari­able fonts across their plat­forms. But the web demos will live on. In 2045, a soft­ware en­gi­neer will come across these demos, and it will be time to learn these lessons all over again.

It’s that time of the year again! Like Punx­sutawney Phil, I gin­gerly stick my head out of my sad lit­tle hole of ret­ro­grade ty­pog­ra­phy and see what’s hap­pened in the whizzy world of vari­able fonts over the past year. To mix things up, and hope­fully pre­vent me from falling asleep at the key­board, let’s try a new ap­proach and take a look at the vari­able-fonts sup­port chart on

Op­er­at­ing sys­tems: All ma­jor op­er­at­ing sys­tems sup­port vari­able fonts. Well, ex­cept Win­dows. Which only sup­ports “named in­stances of a vari­able font” (“named in­stances” are fixed points in the vari­able-font de­sign space that have been pre-se­lected by the de­signer). In other words, af­ter five years, Win­dows still fails to sup­port the sig­na­ture fea­ture of vari­able fonts, which is user-con­trol­lable de­sign axes. Per­haps it won’t sur­prise any­one to hear that Mi­crosoft’s main evan­ge­list for vari­able fonts has been try­ing to shift the con­ver­sa­tion to­ward the web, and away from Win­dows. I see what you did there!

Web browsers: All ma­jor web browsers sup­port vari­able fonts. Ter­rific. I don’t think the no-money-in-web-pub­lish­ing prob­lem has been fixed. But never mind that, I guess.

De­sign & graph­ics apps: Pretty good sup­port, with the no­table ex­cep­tions of UI pro­to­typ­ing tools Adobe XD and Figma, which are used widely in work­groups for spec­i­fy­ing ty­pog­ra­phy, so one the­ory is that their lack of vari­able-font sup­port tends to in­hibit vari­able-font adop­tion more widely.

Video & mo­tion apps: Even though an­i­ma­tion has been touted as an ideal niche for vari­able fonts, none of the Adobe or Ap­ple mo­tion-graph­ics apps sup­port vari­able fonts. So much for that idea.

Of­fice apps: None of the Ap­ple or Mi­crosoft of­fice apps sup­port vari­able fonts. That’s not sur­pris­ing—users of of­fice apps are still rou­tinely stymied by tab­u­lar fig­ures and small caps. Vari­able fonts—surely you jest.

In sum: here in 2021, the only set of apps that of­fer rea­son­ably com­plete & con­sis­tent sup­port for vari­able fonts are web browsers. No sur­prise, since that’s been true since 2017. For all other apps, sup­port is un­even (de­sign apps) or ba­si­cally nonex­is­tent (mo­tion-graph­ics and of­fice apps).

Thus, for vari­able fonts, it’s web or bust. But eco­nom­i­cally speak­ing, can vari­able fonts sur­vive purely as a web­font format?

Con­sider that a key ad­van­tage of vari­able web­fonts has al­ways been os­ten­si­ble per­for­mance im­prove­ments due to re­duced file size. But as I said in 2016—and oth­ers have said since—it’s far from ob­vi­ous that file size is still the crit­i­cal fac­tor in web per­for­mance. Or that vari­able fonts of­fer any file-size ben­e­fit at all in typ­i­cal sit­u­a­tions where only two or three weights are used.

Spoiler: they don’t. To be fair, no one is claim­ing so. And yet. This spe­cious file-size ar­gu­ment re­mains tena­cious, largely be­cause it was part of the ori­gin myth of vari­able fonts. Five years on, no one wants to pop that bub­ble, ap­par­ently out of pity.

Ear­lier in the vari­able-font era, a type-de­signer friend was con­sid­er­ing whether to of­fer vari­able fonts. He told me he ran a script to gather data on how many cus­tomers of his web­fonts used mul­ti­ple styles of a fam­ily within a web page. Al­most all of them used only one or two. For those cus­tomers, vari­able fonts would have no ben­e­fit. But one cus­tomer used 20 styles—surely this ty­po­ma­niac would pay for a vari­able font! He in­spected the logs more closely to dis­cover the iden­tity of this out­lier. An­swer: his own foundry website.

A cer­tain con­tin­gent of re­spon­sive-web de­votées would likely say that I’m (still) miss­ing the point. Once you have a vari­able web­font, the think­ing goes, you can ac­com­plish fancy ty­po­graphic tricks that would’ve been im­prac­ti­cal with old-fash­ioned fonts. For in­stance: head­lines that adapt smoothly to the width of a browser win­dow. This ar­gu­ment isn’t wrong, ex­actly. But it amounts to a large-scale shift­ing of goal­posts: OK, vari­able fonts aren’t vi­able for main­stream web de­sign as we know it, but if we just cre­ate a new global web-de­sign cult based around vari­able fonts, every­one will profit. This is the ar­gu­ment once made fa­mous by the Un­der­pants Gnomes.

Friends, how many more years are we go­ing to do this? We haven’t spo­ken of it so far, but it’s time to start: there’s an op­por­tu­nity cost to pur­su­ing vari­able fonts. Mean­ing—all the en­ergy and at­ten­tion that gets bull­dozed into this ef­fort is en­ergy that can’t be in­vested in solv­ing other ty­po­graphic prob­lems that cus­tomers might—gasp—ac­tu­ally pay for.

In one sense, this doesn’t af­fect me, be­cause I choose not to make vari­able fonts. But in an­other sense, it does, be­cause I’d like to see font tech­nol­ogy move for­ward in ways that make our work more valu­able to more cus­tomers. Wouldn’t you?

By that mea­sure, where are we? Over the last five years, the font in­dus­try has col­lec­tively placed a very ex­pen­sive bet on vari­able fonts. In so do­ing, we have thereby ex­cluded other op­tions. (Lordy, even color fonts that weren’t a shit­show would’ve been more of an op­por­tu­nity.) If vari­able fonts dis­ap­pear, a con­se­quence is that we will have done noth­ing to ar­rest the nat­ural ob­so­les­cence of our work that was hap­pen­ing in the mean­time. Tech­nol­ogy marched on; we stood still.

Worse, by bet­ting on vari­able fonts at all, the in­de­pen­dent type-de­sign in­dus­try has nec­es­sar­ily be­come even more re­liant on a small group of Big Tech gate­keep­ers to de­ter­mine how and where cus­tomers can use our fonts. (We could’ve, for in­stance, been de­vel­op­ing more type­set­ting tools that cir­cum­vent these gate­keep­ers.) Once we for­feit our au­ton­omy, we don’t get it back.

How many times do we need to learn that les­son? What’s hap­pen­ing with vari­able fonts in 2021 feels sim­i­lar to what hap­pened af­ter 2010 as Mono­type con­sol­i­dated the ma­jor re­tail chan­nels for fonts. De­sign­ers who had been sell­ing out­side those chan­nels were largely un­af­fected. But those who re­lied on those chan­nels felt the pinch, as prices and roy­alty rates faced down­ward pres­sure. In that case, how­ever, the re­tail-font mar­ket was still fi­nan­cially strong. The gate­keep­ers were still pump­ing money into the in­de­pen­dent type in­dus­try (though less as time passed).

With vari­able fonts, there has never, ever been an eco­nomic ben­e­fit for in­de­pen­dent type de­sign­ers. The vari­able-font gate­keep­ers just keep ex­pect­ing more hope. More pa­tience. More op­ti­mism. More demos. More work. All for free.

By any rea­son­able ob­jec­tive mea­sure—plat­form & app sup­port, cus­tomer in­ter­est, profit—vari­able fonts have been a fail­ure. Truly, I would’ve been glad to dis­cover ev­i­dence oth­er­wise. Sure, there have been a few suc­cess sto­ries. But vari­able fonts can’t sur­vive as a niche of a cor­ner of the type market.

At times, some have been an­noyed that I keep point­ing out these flaws. But why? First and fore­most, no one cares what I think. Sec­ond, I could be per­suaded that there’s a tra­jec­tory for vari­able fonts where they be­come use­ful and prof­itable. It wouldn’t be the first pivot in tech his­tory, cer­tainly. But that would re­quire a de­lib­er­ate ad­just­ment, which starts with an hon­est ap­praisal of re­al­ity. Ad­vo­cates of vari­able fonts who’d pre­fer to skip over the in­con­ve­nient facts of the last five years are just pro­long­ing the pain.

If any­thing, I worry that ex­pec­ta­tions for vari­able fonts have been set so low for so long—just wait, things will be bet­ter next year!—that ob­jec­tive mea­sures no longer mean any­thing. Like a meme stock or cryp­tocur­rency, per­haps vari­able fonts are now val­ued by their ab­stract po­ten­tial, not by any real-world con­sid­er­a­tion. If that’s the case, then I take it back: there’s no way for vari­able fonts to fail, be­cause the very idea of fail­ure has been vaporized.

Speak­ing of vari­able-font gate­keep­ers—in last year’s re­port I ob­served that Google had been the most vig­or­ous adopter and de­vel­oper of vari­able fonts, but that we ought to stay alert for signs of flag­ging en­thu­si­asm. In No­vem­ber 2021, the Google Fonts project man­ager con­ceded that af­ter many busy years, “Google Fonts isn’t ac­tively com­mis­sion­ing fonts.” Also, though Google-spon­sored work on vari­able fonts had been promi­nently fea­tured at five con­sec­u­tive ATypI events—2016, 2017, 2018, 2019, and 2020—Google is con­spic­u­ously ab­sent from the up­com­ing 2021 event. Oh well—prob­a­bly noth­ing to worry about.

Vari­able fonts are dead. Cause of death: fi­nan­cial mal­nu­tri­tion caused by com­plete lack of in­ter­est from the pro­fes­sional de­sign com­mu­nity and the hand­ful of other en­thu­si­asts who pay for fonts.

Some true be­liev­ers will dis­agree. They are in de­nial. Which, to be fair, is the first step along the path to acceptance.

Some peo­ple paid by Google and Mi­crosoft will con­tinue to in­sist oth­er­wise. They know bet­ter. But un­til some­one higher up re­mem­bers that they ex­ist and then cuts off their bud­get, they will persist.

Though it won’t change the minds of these de­nial­ists, let’s sur­vey the ev­i­dence anyhow:

  1. Since last year, OS and app sup­port for vari­able fonts has stag­nated. The only new app to sup­port vari­able fonts was Figma in May 2022. Though by Sep­tem­ber, Figma had been ac­quired by Adobe. If you’re new here, that’s a fate equiv­a­lent to death.

  2. On, the in­die show­case for new and nifty vari­able fonts, only two fonts were added in the year 2022—back in April. Since then—nothing.

  3. At its de­vel­oper con­fer­ence in June 2022, Ap­ple an­nounced vari­able ver­sions of its core Mac OS fonts San Fran­cisco and New York. Though the fonts also seem to con­tain vari­able axes for weight and width, only the axis for op­ti­cal size is documented.

  4. One pur­ported ben­e­fit of vari­able fonts was giv­ing users more ty­po­graphic choices. I have pre­dicted since 2016—as some­one with thou­sands of pay­ing font cus­tomers—that this was never go­ing to fly. Why? Be­cause fonts are al­ready com­pli­cated enough. Cus­tomers aren’t look­ing for new headaches. Nev­er­the­less, Google—a com­pany with zero pay­ing font cus­tomers—pushed this nar­ra­tive for years. It was only in De­cem­ber 2021 that Google’s vari­able-font pro­moter fi­nally con­ceded that “the vast ma­jor­ity of users aren’t able to achieve those art­ful ef­fec­tive uses”. No kidding.

  5. The most per­ni­cious myth of vari­able fonts—re­peated re­lent­lessly by Google since the ini­tial roll­out—was that vari­able fonts would pro­duce mean­ing­ful file-size sav­ings for web users. That seemed un­likely even back in 2016. Oth­ers have de­bunked it since then. Tellingly, Google never has put forth any ev­i­dence in sup­port. In­deed, in a rare mo­ment of can­dor, Google has ad­mit­ted it’s not true. So let’s call it what it is and al­ways was: Big Tech bullshit.

With that, we have trav­eled full cir­cle. In 2016, the im­pe­tus for this piece was to point out that the dif­fer­ences in in­cen­tives be­tween Big Tech and in­die type de­sign­ers made vari­able fonts at best an un­easy al­liance; at worst, a re­dis­tri­b­u­tion of re­sources up­ward. By that I meant that in­die de­sign­ers would be asked to put time & money into mak­ing vari­able-font tech­nol­ogy look good, while Big Tech—mostly Google, it turned out—would reap the benefits.

Six years on, that the­sis has been proved re­peat­edly. The only good news, per­haps, is that judg­ing by the rate of vari­able-font re­leases, type de­sign­ers seem to have lost in­ter­est, and are no longer putting en­ergy into this black hole.

Still, as re­cently as Sep­tem­ber 2022, Mi­crosoft’s vari­able-fonts evan­ge­list was lament­ing to type de­sign­ers about the “chal­lenge of ex­plain­ing the ben­e­fits to cus­tomers and a big need to ed­u­cate graphic de­sign­ers gen­er­ally.” Hop­ing, I guess, that some­one would vol­un­teer for this long-over­due home­work as­sign­ment? That one more frog would give one more ride to one more scorpion?

There’s one born every minute, I suppose.

As for me—I rest my case.

—Matthew Butterick

by the way
  • I’m nec­es­sar­ily gloss­ing over a lot of de­tails in 30 years of font-for­mat his­tory. I wel­come clar­i­fi­ca­tions and cor­rec­tions if you feel I short-shrifted some­thing vital.

  • I’ve ended up mak­ing more pre­dic­tions than I ex­pected. If they don’t come true, I promise to note where I was wrong. I might be up­dat­ing this piece for 20 years, however.

  • Wired pub­lished a breath­less in­tro­duc­tion to vari­able fonts. Time be­tween the an­nounce­ment of this un­re­leased tech­nol­ogy and the first claim that it has “vir­tu­ally in­fi­nite” pos­si­bil­i­ties: eight days.

  • Prac­ti­cal Ty­pog­ra­phy uses a vari­able-font-like trick to serve dif­fer­ent grades of the body text font to dif­fer­ent plat­forms. For in­stance, Win­dows users get a slightly heav­ier ver­sion than Mac users, to ac­count for lighter screen ras­ter­i­za­tion. All this can be done eas­ily with to­day’s tech­nol­ogy. But I seem to be the only per­son on the in­ter­net who was suf­fi­ciently mo­ti­vated to try.

undock move Heliotrope Equity Valkyrie Century Supra Concourse Triplicate buy font close