Jürg Lehni interviews Erik Spiekermann about the legal definition of typefaces, web font distribution models and parametric fonts.

Designer and programmer Jürg Lehni analyses the evolution of typographic technology and the nature of digital fonts, and introduces Donald E. Knuth’s groundbreaking TeX and Metafont systems. An essay complemented by interviews with Peter Biľak, Erik Spiekermann and Dimitri Bruni (NORM).


Jürg Lehni: Erik, I am curious to find out more about the reasons behind the legal definition of typefaces as software, rather than artworks. Is it simply the fact that OpenType, PostScript, and TrueType all have aspects of programming languages in them, or is it linked to the fact that digital typefaces are simulating something that was once sold as tools, a collection of cast metal letter-forms, to be typeset by hand? Could you elaborate the nature of this decision a bit more? In my research I have found very little about the reasoning behind it.

Erik Spiekermann: The main reason for the definition of typefaces as software is the fact that typefaces can hardly be protected since most people do not even see a difference between a Garamond or Bodoni, let alone Helvetica and Arial. The requirements for a text typeface to not put itself between the reader and the text, but at best to offer a slight aesthetic added value is a disadvantage for the estimation of the “artistic” contribution as added value that is required by copyright law. This is why the most bizarre ornate typefaces that no one would want to copy anyway are protected. They are clearly recognisable as creations rather than simply the reordering of generally known parameters, as would be the case with more serious typefaces. An “A” needs to look like an “A”, so why should the “A” be protected if it looks exactly like all the others, at least to the layman’s eye?

It appears to be a problem of the framework of copyright law, which has a lack of suitable rules for typefaces, since they hold a strange position in-between tools and artworks. But let us assume for a moment that all typefaces would indeed be legally defined as artworks. In this case, a royalty-based system as it is in place for example in the music industry would surely make far more sense. Would this not be a desirable solution for the way digital typography is sold today? Was the lobby of font designers not strong enough to enforce a proper legal handling of their works, or are there good reasons why the definition of software are preferable to this alternative scenario?

Of course it would be better for the type designers and foundries, since font licenses would be accounted for based on the distribution of the media in question, but who would control all this? When radio stations play music, they have to take note of every title played and send the list to associations that control the usage. In Germany this is handled by the GEMA. These associations then distribute royalties to composers, musicians and authors, based on the broadcasting’s reach. A small local station therefore pays less than for example the ARD (National Public TV). But how could the runs of printed matter be controlled? Maybe it could work for books and newspapers, but even there it would require excessive bureaucracy. For the sake of fairness, the amount of text would have to be precisely counted, otherwise the designer of a typeface that sometimes is used for a caption only would receive the same amount as one who created the text typeface for a whole newspaper, used to typeset millions of characters per issue. With music, longer pieces also pay more than shorter ones, but on radio and TV usually only one song plays at a time and the amount therefore is easily measured.

But if we look at the embedding of typefaces on websites, there could indeed be a viable way to charge licenses based on automatically measured usage. For a typeface on a private website, seen by 300 visitors per month, I would pay 10 dollars a year, respectively a few cents per “hit”. A newspaper like the NZZ on the other hand would then maybe end up paying 1000 dollars per year. The web offers the possibility to automatically register and calculate usage through visitors, as used for example by Google as their way of handling advertising. But I doubt that such models will be successfully established in the type sector, since typefaces have degenerated to a ubiquitous and cheap commodity. I therefore do not really expect anyone to be willing to abandon the current model.

Since the first part of this interview was conducted, two new such efforts have emerged, based on recent additions to browsers like Firefox and Safari that now allow embedding of fonts in websites even if they are not installed in the operating system: TypeKit and Kernest. How do you see their chances of becoming a viable business model, a new alternative way of distributing and charging for the use of digital typefaces?

Both of these companies propose a service which would provide fonts from their servers, charging not the end-user, but the publisher of a website and paying the foundries for the use of the fonts. Obviously, foundries may offer the same service. To me, a veteran of a few font wars and typographic revolutions, this sounds almost like typesetting in the old days. We would send our manuscript to a typesetter with some instruction about the formatting, and we’d get back a negative, a print or even data. The service was charged by the hour, which in turn was a mix calculation taking into account the cost of the equipment, the level of operators’ expertise and the exclusivity of the fonts chosen. These services might work, but only if they do not obstruct the user’s experience. Nobody will tolerate error messages because a certain font cannot be retrieved from a remote server or if the license does not permit the use in a given domain or country or time of day.

Another model is the way foundries handle typefaces that are licensed to big companies. Software companies like Microsoft charge up to 100 dollars per user and year for the use of their programs. If that method was applied to the licensing of fonts, it would result in massive sums for big corporations like Bosch with 60,000 users, without requiring any additional work, since updates are charged for extra. This is why smart corporations usually license the exclusive right for a typeface. This way they only need to negotiate with the designer one time, as we did with Bosch.

I would like to go back to the question of the legal definition once more. What if we turn the situation around? If typefaces really are to be defined as software, should they not act more like software, in a more modular, flexible way? At certain points in the past there were proposals of more dynamic, parametrized typefaces, such as Adobe’s Multiple Master (MM), or Donald E. Knuth’s Metafont language. At one point MM was even integrated into most of Adobe’s own applications. But in the meantime the technology has disappeared—apparently it could not establish itself as a standard. Why do you think it has failed? Was the user overwhelmed by the amount of options, or was the standard not flexible enough?

Typefaces are still parametrisable, but only by the experts. Adobe has stopped MM because nobody was using it. But many of the type designers still work with MM which is now integrated in FontLab. Typefaces are so easy to use that nobody wants to think about them. This is also the reason why all the more complicated licensing schemes are hardly feasible. Most of them would eradicate the huge advantage that for most users fonts are installed somewhere hidden in the operating system and can simply be chosen through a pull-down menu. Even professional users such as graphic designers often know very little about the underlying technology. Mostly they simply want to be able to use the right typeface immediately, at a low price, and without having to read complicated contracts first.