Presentation at BarCamp Apache Oxford on 5 April 2009


I am an information designer and that has made me a web designer and developer almost by default. I got to this position by enjoying drawing letters and printing; I studied typography academically at Reading University, and that led me to the work I do now. I was occasionally bored as a student and I was given a piece of font-creation software called Fontographer which gave me an excellent grounding in font editors and a pretty strong inclination not to become a type designer. It’s very habit-forming and it panders to my worst instincts, I’m afraid to say!

I’m going to talk about a project that is not on the face of it directly concerned with software. But I think you will very soon see that it is in fact quite closely connected in several respects. I’ll discuss the background, diverting into font techy stuff a bit, and then I’ll finish by bringing things up to the present day.


The Open Font Library is a web site where you can find fonts licensed under permissive licences. Anyone can sign up and contribute their own fonts as long as they’re licensed under permissive terms.

Open Font Library Download fonts, upload typefaces, branch and merge designs!

Similar projects

The origins of the Open Font Library lie in similar projects: ccMixter and Open Clip Art. ccMixter was commissioned and is run by Creative Commons. It’s a music sharing site targeted at amateur musicians and appears to have successfully proved the point it was intended to make.

ccMixter - Welcome to ccMixter

Open Clip Art, born of the same impulse, uses the same technology and sometimes the same terminology and is a place where you can contribute and make use of vector artwork for use in presentations and the like. Reflecting the different nature of the content, there is a different emphasis placed on the various capabilities of the site.

Open Clip Art Library Drawing Together

In both cases the Creative Commons license, something I came across firstly at Flickr, was the key. Creative Commons reframed the concepts of copyright in a way that made them comprehensible to amateurs. This meant that the democratisation of activities like music production and graphic design - which PCs made possible - moved forward another step.

The Creative Commons License is not just a rephrasing: it’s an invitation to individuals to push their creativity in the world. Sites like ccMixter and Open Clip Art follow up on the promise of the license by providing somewhere that can host the contributed material without diluting the sharing culture message by being covered in ads or offering any other inducements.

The Open Font Libray was begun by people who had seen these existing sites and wanted something similar for typefaces. This is a logical desire and it reflects the fact that type, like clipart and music, is part of the creative world.

How fonts are different

So the Open Font Library was duly created. It uses ccHost, the same software platform as ccMixter and Open Clip Art. And so it also shares the model of those sites: a model in which there’s a assumption that the material is out there, the demand is out there, and building a site is a question of providing upload links, search facilities, and download links.

But I think there are some differences, and it’s some of these that form the bulk of the talk.

How fonts are made

Firstly, I want to look at how fonts are made. In contrast to photographs, clip art, and even music I believe fonts can require very much more time. I’m not suggesting that because a photo is formed from a slice of light over a fraction of a second of time that photographs are ‘easy’, or that because a lot of clipart is drawn over photos it’s trivial, but nonetheless fonts can take years to create whereas the other items I’m considering generally don’t.

Obviously, not all fonts are alike. Geo is a font that I made years ago in four hours. But that’s because it’s a copy and paste job built as if from Lego with the minimum regard paid to legibility, aesthetics or the character set. Open Font Library needs to encourage better fonts than this.


A font is a set of glyphs. Glyphs are usually drawn and stored as vector artwork. In common with other vector artwork the data is a set of points with information about how to join them up.

Unlike other vector artwork formats, the font file is a collection of glyphs, each filed under the character the type designer is trying to represent with that set of points and joining instructions. There is also some extra information that provides help to the software that will be used to display the glyphs. It can, for example, tell that software that when an A and a W are next to each other it’s advisable to reduce the gap between them. This is called ‘kerning’ and type designers (or their assistants) frequently put a lot of time and effort into it. Remember that for Latin typefaces, each character can appear before or after every other character and that there are roughly 26 times two letters.

I could talk about the complexities of non-latin typefaces, which frequently feature complex and subtle rules of letter or word construction where elements must be placed above or below other elements and all kinds of cleverness in typesetting is required: but that takes me way out of my comfort zone, so I’ll leave that as an exercise for later.

Here is a quick peek at a font editor, FontForge, which is a good example of niche free software of high quality.



I don’t have professional type design experience but I imagine that revision control, and certainly version tracking, is a key part of typeface development wherever it’s done to professional standards. I’m not so sure that is the case for material in the other libraries I’ve mentioned.


Earlier this year a thread on Typophile, which is the preeminent forum for discussion of type design practice, suggested the formation of a group to develop an ‘open’ typeface collaboratively. This drew much venom. Most of the venom was directed at the idea of a genuinely useful free typeface, which many professional type designers seem to find threatening. But there was also skepticism that collaboration could work out. I wonder whether that is a real problem or not. I don’t know the answer, but in a studio working on a type design in the days before software could be used for the whole process there would usually be more than one person involved in the production work.

What fonts do

Fonts are used in a very different way to music files, clip art or photos.

Fonts are descended from writing systems, and writing systems are the means by which we can encode, store, transmit and re-use our thoughts. Type has supplanted handwriting as a means of mechanising the system, but the requirement remains.

Fundamentally, it’s impossible to use a computer without a way to communicate with it. As soon as we go beyond lights and switches we start needing to use characters to input and output information and fonts follow fairly swiftly after that.

In other words, fonts are message-carriers not message-originators. That makes them a very different proposition to music or art.

Minority languages

As I’ve implied above, creating a font usually requires a fair investment in time. Mechanising language has always been costly. So one might imagine the presence or absense of font software for minority languages is one unconscious form by which these languages are kept that way. Of course this depends on whether the writing system for that language is not found elsewhere. I doubt anybody has rejected the idea of creating font software because they actively oppose the spread of a language, but it is easy to imagine budgetary constraints having a similar effect. I’m ignorant in this field, and not equipped with statistics.

Poor quality fonts

Some writing systems do have a body of fonts already in place but lack many of quality. Examples include Arabic and Urdu. I should distinguish between aesthetics and technical quality here and say that on the evidence of a talk I attended a few years back the Arabic fonts were in two groups: technically good but aesthetically poor, or aesthetically promising but poorly made. For Arabic, a writing system that embodies so much of the culture that formed it, this is an unfortunate situation.

The new site

I’ve given some reasons why simply changing ‘Clip art’ to ‘Fonts’ might not be the best way to establish a vibrant and useful web site with fonts zipping to and fro. These are the ‘deeper’ reasons, addressing the model of font development and usage. But there are more straightforward reasons why the site might not be thriving and it was these that I hoped to address when I got involved with an update in Autumn last year.

I think there were two straightforward reasons.

Firstly, the site looked boring. It wasn’t clear how you uploaded a font file and what the outcome would be. Which formats were acceptable? Was anything else required along with the font? Could I upload the font forge file too, so others could have the ‘source code’ for my font?

Secondly, there was an absence of background information. What was the rationale for the site? Why was the license choice Open Font License and where had this come from?

I felt that if the site was to deliver on its promise, it needed a rethink.

Fonts are ‘consumed’ by creative people who are often visually sophisticated, so it seemed wise to kick the site’s visual design into shape. What we have come up with may not set the world on fire, but it’s a more generous, open look where the elements are controlled better. The intention, in any case, is to let the type that is on display shine through. We encourage users to upload sample images and we have a previewer that’s capable of working over many character sets and keyboard layouts in the works.

I had also noticed that things were moving again on the font linking front, and it struck me that the Open Font Library could play a part in this. Font linking allows the use of font files hosted on a web server to typeset the text of HTML pages. It’s a bit of a nightmare for the font foundries, because it suggests that generating revenue from font licensing to web sites may be something only the toughest can manage. But for freely-licensed fonts, it’s a no-brainer. The only bar that is set is the quality of the font itself.

We now generate a vary basic chunk of CSS for each font, which can be copied and pasted into any site’s CSS. We also provide a brief tutorial explaining the code. This is brinkmanship, because we are effectively offering to host free fonts for the web. We have yet to discover whether it will have any impact on the world or our bandwidth requirements. Firefox 3.5 promises to support font linking, and the World Wide Web consortium has also confirmed that it prefers the straightforward use of OpenType fonts over Embedded OpenType. We have yet to see how this will play out.

@font-face - MDC

Web font linking with @font-face - OFLB

To these early ideas, we have been able to add several useful additional features that should really help the Library to fulfil its potential. Ed Trager, a software developer who displays considerable knowledge of writing systems, has been at work on a number of analysis utilities that make the site better. I have already mentioned the first of these; it is an interactive previewer, allowing users to get a taste of each typeface as they browse without requiring a download. The second analyses the coverage of character sets for each font, something that a we can use to help drive up the quality of the fonts on the Library.

Ed Trager


The Open Font Library is a collaborative project that moves forward when there is will power, time and money available. The site improvements I’ve described exist in Subversion and on the development site; other improvements exist only as items on our wishlist. Some important things that are not yet implemented include supporting Linux distribution package maintainers and offering version awareness. I hope that what eventually emerges will make a useful contribution to the pool of fonts and font expertise. I’ve enjoyed working on the project.