The Shape of Understanding

Yesterday, the Museum of Modern Art in New York opened the exhibition Full Disclosure: The Edge of Information Design. Its 30 digital and analogue works explore how designers turn data into forms that people can see and understand. In an interview, curator Paola Antonelli recently expressed the motivation sharply: »The objectivity of data, in particular, is increasingly under scrutiny, as public confidence in institutions and in the intelligence they produce is decreasing.«¹

»The real distinction isn’t between diagrams to trust and diagrams to doubt, but between an exclusion we can see and one we can’t – between not yet knowing enough, and having been reassured just enough not to ask«

Can a diagram tell the truth? Maybe diagrams were never really in the truth business, only the legibility business – and the two have been quietly confused for a century. Or maybe that’s too simplified: the people involved in both businesses cared, each in their own way, about getting something right.

In 1920s Vienna, polymath philosopher Otto Neurath and his collaborators developed ISOTYPE, a system of pictorial statistics designed to make complex information legible to people outside the target audience of conventional statistics. The heartfelt ambition was usability over spectacle: standardised symbols, repeated using strict rules, so that a visual reader could take in unfamiliar connections at a glance.²

Technical draughtsman Harry Beck drew electrical circuit diagrams for a living. His 1931 map of the London Underground treats the railway network as such a circuit diagram, making the same point as Neurath but in an entirely different way. Beck abandoned geography for connectivity – straightening lines, distorting distances – which worked precisely because his wasn’t a correct depiction of London. We know Baker Street isn’t really at right angles to King’s Cross. We accept the distortion because we understand what the map chose to preserve: topology instead of geometry, how to get from A to B instead of where A and B actually are in the world.³

Mathematician Claude Shannon’s information theory gives that choice a name. Communication means encoding information into a signal that can pass through a channel, and often something is lost in that encoding.⁴ Representation establishes a relationship between what’s retained and what’s lost. Making something legible isn’t just making the invisible visible – it’s deciding what needs to be on display and what can remain invisible. Beck’s map hides geography. ISOTYPE hides numerical detail. Every useful diagram builds a field of attention by excluding almost everything else.

Perhaps that’s just the cost of any communication, no more troubling than speech leaving most of the world unsaid. Or perhaps something changes once what is left out stops being visible as an omission – once a diagram no longer reads as a selection, but as the thing itself.

Architecture has something to say here. In their book Learning from Las Vegas, architects Robert Venturi, Denise Scott Brown, and Steven Izenour distinguished the »duck« from the »decorated shed«: the duck’s form carries its own message, while the decorated shed is an ordinary building wearing symbolic meaning applied to its surface.⁵ The distinction transfers: an honest diagram is a duck, its structure doing the work its appearance promises. A decorated-shed diagram wears the signs of legibility without doing the actual work.

That’s the sharper version of what’s at stake. Information design was never going to fail by excluding things – it has to, that’s the whole method. It fails when it hides the fact that it has excluded something. A duck’s omissions are legible as omissions. A decorated shed’s aren’t. Good simplification hides information. Bad simplification hides the fact that it has simplified.

Think of a dashboard reduced to a single number – a credit score, an election-forecast needle. Nothing about its clean presentation admits that anything was left out. The cleaner the number looks, the easier to forget what was measured, excluded, and assumed, and how uncertainty was handled.

Statistician and data visualisation pioneer Edward Tufte’s long attack on »chartjunk« tried to prevent exactly this, stripping away decoration that competed with the data itself.⁶ His answer wasn’t to distrust diagrams, but to make sure the visual economy earned its keep. The trouble is that the same competence serves both purposes. A building that lies about its structure eventually meets gravity, an engineer, or an inspector. A dashboard only has to meet us. A diagram can borrow that trust from its appearance without earning it.

We increasingly inhabit systems we can’t examine directly – algorithms, financial models, filtered content. We only ever experience their representations. The real distinction isn’t between diagrams to trust and diagrams to doubt, but between an exclusion we can see and one we can’t – between not yet knowing enough, and having been reassured just enough not to ask.

Harry Beck’s map has been wrong about London for nearly a hundred years. It never collapsed, because it never once pretended to be right.

References