Keeping an eye on the dashboard

Data dashboards help us to make sense of increasingly complex and constantly refreshing information. But what do they miss out? Jamie Bartlett and Nathaniel Tkacz introduce their ESRC-supported project on the subject.

In late 2012, the Government Digital Service created a new way for the Prime Minister to keep on top of what was happening in the UK. It was called the ‘Number 10 Dashboard’. This bespoke iPad app provided performance indicators on certain government services, real-time information on aspects of the economy, trends from social media and expert commentary, all integrated into a single screen and with the capacity to ‘drill down’ as needed.

The images accompanying the media reports of this new tool were particularly striking. The Daily Mail printed a photograph of Cameron ‘testing’ the app while seated at a breakfast table – piece of toast in hand and ready to swipe the screen. An article in The Guardian featured the Prime Minister on a moving train, leaning back in his seat, with pursed lips and eyes focused on the screen. Everything is moving, besides Cameron’s gaze. This was real time governance: data on the move, as it happens. A way of relating to a world manifested through data, somewhere between glancing and monitoring, of selective interaction, of ‘lean back’ governance, constantly in motion.

Dashboards have been with us for a long time of course. In simplest terms, they are single screens that aggregate and display multiple flows of data. In the security sector, they are the screens that populate the walls of ‘control rooms’. In finance, they are embodied in the Bloomberg Terminal. In management, they belong to the histories of ‘business performance management’, ‘intelligent systems’ and ‘key performance indicators’. And, of course, the term came to prominence with the automobile over a century ago as the ‘board’ which separated the driver from the elements.

Dashboards have become one of the ways that decision makers across many sectors in society now receive and digest information. Federal Reserve Chair Janet Yellen consults a dashboard – known as Yellen’s Dashboard – to see where the US economy has been and is likely to go. China is seeking to build a dashboard of over 40 economic indicators to ‘measure the efficiency and quality of growth’ as part of a general shift away from current growth measures. There are dashboard applications in all the app stores. Personal fitness dashboards, social media dashboards, user profile dashboards, advertising dashboards, blogging dashboards. The ‘morning dashboard’ for the iPad is designed to de-throne the newspaper as the modern subject’s source of enlightenment. Companies such as Geckoboard and Tableau exist to create bespoke dashboard data solutions.

This all suggests that dashboard’s functionality is becoming generalisable. In academic circles, we would call this becoming ‘diagrammatic’ meaning that the format is starting to leave a mark. The reason they are now so important for modern decision makers, from government to traders, is because of the ‘data revolution’ and the related challenge of the so-called data deluge. Over the last decade, there has been an explosion in the amount and variety of digital data being created, documented and shared – and thanks to commensurate improvements in computing power and storage, this is increasingly amenable to collection and analysis. Because there is so much data available, it is necessary to present it to those that use it. These small screens of pre-selected, pre-formatted information are presented as the ‘solution’ to the deluge, ‘noise’ and other constituent excesses of big data. As Google Now puts it, ‘just the right information at just the right time’.

This all matters. When dashboards become integrated into the routine practices of government, they change processes of decision-making. Although dashboards are increasingly our analytical window into the world of data, they are not necessarily neutral purveyors of that data. They invariably shape and prioritise the information that is presented. As NYU Professor of Media Lisa Gitelman recently put it the notion of raw data is an oxymoron and the dashboard adds another hermeneutic layer to the mix. Which metrics are privileged? Who decides when a particular indicator moves into the red? How regular is the refresh rate, that is, what kind of temporality is built into the dashboard and how does that move us to act? Which metrics are not available, or deliberately left out?

And dashboards can often obscure more than they enlighten, because many of them present data without the user really knowing how it was created. We call this the ‘black box’ problem. Sitting behind a dashboard is a complicated world of data scraping, API calls, word based sampling methods, natural language processing algorithms – and any number of new modes of collection and analysis. Some of these are highly complicated and difficult even for experts to comprehend, while others are alarmingly simple and reductionist. Every algorithm that classifies data and every sampling method employed, carries certain biases and caveats that influence what the dashboard displays. This all happens in the black box, often beyond the reach of manipulation of the user. At their worst, dashboards might dazzle with visuals rather than illuminate with insight. Cameron was reported to have liked his dashboard because it was ‘simple and very easy to use’. But that is also the dashboard’s great weakness.