Beyond Classification, or: new legs for a dead horse

This paper was delivered at the ATypI conference in 2013 in Amsterdam.
Since then, the Vox classification system was abandoned by ATypI in 2021
For anyone confused by the sub-title: to ‘flog a dead horse’ is a colloquial English expression meaning ‘to waste energy on a lost cause or a situation that cannot be changed’

I would like to start with an apology: I haven’t discovered the perfect, comprehensive type classification system.
(I hope no-one’s too disappointed)
What I have been trying to do is to revisit some rather tired arguments by interrogating some of their underlying assumptions, to suggest a basis from which the topic might be moved forward.
(re)Defining the problem
Since its conception in 1954 The Vox system has been the closest we have to a global consensus on type classification. Ever since that time, the limitations of its scope and the flaws of its method have drawn criticism from type designers and type historians. It was augmented as Vox-ATypI in 1962, and formalised as a British Standard in 1967.
Catherine Dixon’s paper at the 2002 St Bride’s conference made some timely observations in reminding us that Vox’s conception was considerably more nuanced and organic than the rather doctrinaire, reductive interpretations which have survived.
Discussion of typographic taxonomy and reform, including the recent ATypI special interest group on the subject, has been characterised by flurries of activity which tend to exhaust their momentum and drift into interesting theological wrangles. One participant likened this to ‘a party game’.
Leaving to one side the many design flaws for which the Vox system is criticised: (the conflated categories and invented words, the uneasy mix of historically-defined and formal characteristics), why do we find it necessary to revisit this question?
The perceived need for reform is often attributed to the changed conditions of the digital era, the most obvious of which is the sheer number of digital typefaces , deepening the challenge of meaningful choice, and bringing many interesting new classificatory problems.
These have included the postmodern fashion for hybridisation and stylistic juxtaposition, by which typefaces like Max Kisman’s Fudoni presented an explicit and deliberate challenge to the authority of classification.
These factors have been so dramatic as to obscure other considerations. As a set of finite practical problems, they might suggest a larger and more rational version of the Vox model as a possible solution. But to propose more or better-defined categories, is to accept a basic premise rooted in the conditions of the mid-20th-century.
I’d suggest rather that the change is not simply quantitative – the number of type designs – but qualitative, in the need to address different communities of users and a far greater diversity of user needs.
Alongside developments in typeface design, the activity of selecting type has been democratised and is no longer the exclusive province of informed specialists within the confines of metal type and photosetting. It has become an everyday occurrence, considered (however briefly or casually) by a wide range of the population in the course of their working day.
Classification and its discontents
As designers, we’re aware of the truism that appropriate solutions depend first upon accurately defining the problem they address.
I’d propose that the reason that the sporadic interest in this topic seldom reaches much of a conclusion, rests not in the relative merits of proposed methods, but in a tendency to make broad assumptions about the nature of the problem. Attempts to locate the philosopher’s stone – a single, comprehensive taxonomy for type – are limited by the fact that different users think of type in different ways, and thus require different classificatory tools.
What do we need it for? Thinking differently
If we consider type classification to still be a viable topic, I’d suggest we might reconsider our objectives, and recognise that the conditions of twenty-first century typography differ so substantially from those of the 1950s as to require not only different systems of classification, but quite different ways of thinking about what classification is, and what we want to use it for.
– not a different system, but a different way of thinking about systems
so: what is the purpose of type classification in the 21st century?
who is it for?
what is its job?
how has that job changed?
Thomas Christensen observes:
‘ the question I don’t hear asked is what the purpose of such classification is. Is it just a filing system to help us find things, is it a guide to function and usage, or what? Unless the purpose of the classification system is explicit it can never be satisfactory.’
In 1954 the idea of the ‘non-skilled type user’ had no significant currency, the design of typefaces was still largely a specialised adjunct of the printing and typefounding industries, and the purchase of a typeface still took the form of cases of metal or matrices. Classification was therefore the common language of a relatively well-defined specialism.
Since then the user base has broadened. Classificatory systems will have different meaning and purpose for different users: for the font distributor addressing a non-specialist audience, for the educator developing an informed understanding of type use and type choice in their students, for the skilled typographic practitioner, or for the type designer considering precedent and defining characteristics in new designs or revivals
What kind of system? diverging purposes
The development of objective classifications and categories to suit the different expectations of each of these audiences, will lead us in very different directions.
We can see this when contrasting the categories proposed by Robert Bringhurst, in which the practice of typographic classification is used to articulate aspects of cultural history.
In the commercial arena meanwhile, the early days of digital type distribution brought us new categories such as Ironic, Random, Handwritten, Amorphous, along with the worrying use of ‘Historic’ and ‘Typographic’ as discreet categories, as well as the one from which I’ve adopted my title: ‘Beyond Classification’.
The need to adopt such a term indicates a clear failing in existing classificatory language.
Many current font retail sites use multiple key-words to denote ‘similarity’ at the point of sale.
What kind of classificatory system can better serve the conditions of 21st century type? I’m reminded of that famously unhelpful direction to travellers, “you don’t want to start from here”. Rather than try to standardize an improved vox (uber-vox? supervox?) we need to reconsider the questions a classification system is expected to answer, and how we redefine the nature of those questions.
Voices in the room: recent work on the subject
Excellent, thoughtful work has been done in this field; too much to credit in detail in a 20 minute presentation. I’d mention Jonathan Hoefler, Indra Kupfersmidt, Thomas Christensen, Alessandro Segalini and Denise Gonzales Crisp among others. Outstanding research has also been carried out by Catherine Dixon and by David Shields, addressing some of the areas least adequately covered by the Vox system.
Ultimately, these contributions might tell us as much about typographers as about typefaces; dialogue about classification has become a medium through which to talk about type, and these represent a spectrum of individual philosophies and perspectives. In comparing them it is clear that they cannot be viewed as ‘like for like’; they are not simply different solutions, but solutions to different conceptions of the problem.
This need not be a bad thing, as it provides the basis for thinking differently about what a classificatory system means and does.
Methodologies: approaches to classification
John Hudson made a particularly lucid distinction between the two main approaches to classification as based in either intrinsic or extrinsic characteristics, grouped according to formal properties or historical/cultural origins. A defining fault-line in the idea of an ‘omniscient’ classificatory system is exemplified in Vox; the idea that the one system can address both intrinsic and extrinsic, combining historical context and formal description.
Classification as type history is compromised by the wide spectrum of ‘revival’ faces and the variable extents to which contemporary typefaces reference historical models.
Classification by intrinsic method, though tempting to the systematically-minded, is of limited value in determining a type’s appropriateness for use. Data sorting by letter profile, as recently proposed by Julie Chauffier among others, presents a seductive prospect of methodical consistency, but in practice serves to demonstrate the limitations of classification based upon ‘like shape’; it is appealing as a way of grouping fonts but seems less effective for the grouping of typefaces and, for instance, addressing the complex and richly variable relationships between their roman and italic form.
If we wish to consider type solely as a set of shapes, detached from historical context; then formal description and digital data sorting can be applied to this problem with appealing consistency, but it would be unrealistic to imagine that such an approach satisfies all user expectations. If on the other hand we are concerned with type as a cultural artefact; a medium intrinsically linked to developments in literature, architecture, science and the history of ideas, the consideration of letters as abstract form will be of only very limited use.
I’d suggest therefore we need to introduce a new level to the sorting process. To make sense of the practical problems we face, any meaningful use of classification for the selection of type would involve at least one prior phase: that of selecting our classificatory model.
This will depend upon our priorities.
Universality, post-modernism and post-modernity
The supposed ideal of a single, omniscient system may be not only unachievable, but redundant as a philosophical position. The idea of a ‘universal’ system seems characteristically modernist in its ambition, assuming the cycles of successive improvement in a manner which post-modernism set out to critique. The idea that the Vox system could be ‘improvable’ is, I’d suggest, a modernist response to what is now a post-modern problem.
A defining characteristic of post-modernism has been the rejection of ‘big narratives’; the chronologically defined orthodoxies of history. The Vox system (or, still more problematically, part of it), presumes a ‘big narrative’ and I’d suggest it is this that has proved inadequate to the demands of current conditions.
Post-modernism as method, not style
John Hudson, Hrant Papzaian and Nick Shinn contributed to an interesting Typophile node where they discuss the potential of ‘post-modern’ as a possible stylistic ‘category’. Though it was a lively discussion, it seemed to me symptomatic of the way the design professions have short-changed themselves over postmodernism; identifying it as a transitory ‘style’ while neglecting its potential scope as a critical perspective.
Hudson identified an instance of which both are true; an example from the heady days of post-modernism and the heroic period of early digital type design, the late Scott Makela’s Dead History.
If we move away from the rather reductive perception of post-modernism as a ‘style’ (now dumbed down into popular design history as ‘po-mo’ and dismissed), and look instead to post-modernity as an attitude of mind, it shows its potential as a way of looking at methodological tools, looking differently at categorisation itself, and finding a rationale for the co-existence of diverse classificatory practices as simultaneous and equal-valued.
Postmodernism, and in particular, the language of postmodern architecture, validates the idea of ‘radical eclecticism‘ as a rationale for the mixing of contextually dissonant elements.
The idea that seemingly contradictory positions can co-exist is one which can be extended beyond the stylistic mash-up, and toward proposing the peaceful co-existence of distinct classificatory methods.
Why does it still matter? language and thought
To recap: the issue is not so much that 20th century classification systems are ill-conceived, but that 20th century expectations are no longer appropriate to current conditions. But this is not to suggest that type classification itself is a futile or redundant activity. The way we classify impacts upon the way we think about design and the way we design. It prompts detailed consideration of both typeface choice and typeface design.
The language we use, serves to drive, inform and indeed constrain the way we think, and therefore the way that we design.
Thinking differently about classification: redefining structures
It is becoming clear that the changed conditions of type production may not be adequately served by the refinement or development of previous classificatory models, but require instead a different structural approach. I’d propose that classification might now be practiced in a pluralistic and conditional manner. Rather than aspiring to the modernist ideal of a single linear system, we may need to move away from the dominance of either the extrinsic or intrinsic, or the uneasy forced synthesis of the two we have seen in the Vox system, toward the co-existence of different systems for different purposes and contexts. If we recognise that variant systems of classification can exist meaningfully in relation to each other, we can then make appropriate choices of system based upon our needs and requirements.
We have identified several divergent approaches to classification, which we might describe as ‘modes’ of classificatory thinking, and the solution may be a multi-modal one.
Within the same technological shift that has prompted this dialogue, developments in interactive media have provided practical means for exploring alternatives to linear, sequential structures; giving us instead the tools to articulate and explore radial, nodal or constellational models.
The problem we are facing, is nicely coincident with the idea of conceptual clustering and the development of concepts such as fuzzy set theory.
Addressing the different methodological camps and user groups I have identified, may require a move away from a two-dimensional linear structure towards the three dimensions of virtual reality, to accommodate and compare extrinsic and intrinsic systems, to offer both a contextual and formal axis.
Conclusions (of a kind)
Are we any closer to a single descriptive vocabulary for typographic style? probably not. But in mapping the circumstances that make such an idea redundant, I hope that I’ve started to outline the possibilities for a redefined status: of classification as a tool for reflection. It may be that like other aspects of typographic practice, classification systems as we know them are entering a cycle of redundancy, out of which their significance can be reviewed in the light of new conditions, something like the contemporary re-appraisal of letterpress or the renewed interest in punchcutting.
Through this we may view type classification less as an objective tool than as a medium of enquiry; not a means to unexamined decisions, but a choice of perspective.
Bringhurst’s classifications exemplify this; representing not a proposal for consensus of usage (after all, no-one else describes types in that way) but an authorial position.
The best writing on type classification is more type philosophy than rule-making. A typeface is a cultural artefact, and like other cultural artefacts it is susceptible to different modes of interpretation. Classificatory systems are an aid to this process rather than a set of absolutes, and classification is above all an index of philosophical inclination and creative style; an interpretative tool.
It has been suggested classification may be an obstruction to thinking about type. With respect, I’d propose the opposite, that a pluralistic approach to classificatory method serves to illuminate and deepen the quality and diversity of thought about the subject, helping us to reflect upon what it is we talk about when we talk about type, to bringing some conditional order to its multiplicity. Not perhaps to sort more efficiently, but to think more expansively.
It is increasingly difficult to think of the relationality of typefaces in the linear terms presumed by Vox or indeed Bringhurst; these relationships are more like constellations, and the way in which we organise such constellations will be determined by the affinities we favour; intrinsic or extrinsic, formal or cultural.
Historically, interpreting the constellations has prompted the development of cosmologies.
…and as our cosmologies inform our belief systems, our perspective upon the constellations determines our view of the typographic universe.