If an algorithm can construct a narrative, what is the role for a museum or gallery?
Chicago-based company Narrative Science began as a research project at Northwestern University bringing together computer science and journalism students with professors in the Intelligent Information Laboratory. They created a piece of software called StatsMonkey, which automatically generated stories of little league baseball games by turning statistical data into narratives. The company would source statistics and data about who hit what, and what the scores were and so on, and then the computer would use that information to write stories about the games. In other words, Narrative Science found a method for automatically turning structured data into human-readable narratives.
Given how pivotal storytelling is to the work of museums and to much of what the arts does, the idea that an algorithm might be able to construct meaningful narratives – these things that seem so human – might seem foreign, but it’s starting to happen.
What makes this concept particularly interesting to me is the way it starts to play with the scale of storytelling. Most people are not interested in stories about particular ittle league games. Few news organisations would ever send staff to cover every game in a city, because the economics of that just aren’t practical. But stories of these games are interesting to some people, they’re interesting to the players and their families, to members of other teams, and once the initial algorithm has been written, this computational approach to storytelling makes it economical to tell data-driven stories all the way down to an audience of one.
But this approach is also a way of scaling up capacity to tell stories. ProPublica, a public interest newsroom, recently employed Narrative Science to create short narrative descriptions about almost 52,000 American schools, to show differences in access to advanced classes between wealthy schools and poor ones. The information already existed in a database, but they wanted to make it so laypeople could comprehend what the data meant. Writing 52,000 individual stories would take significant resources; getting an algorithm to do the work offers a very different and potentially much more effective approach.
Now just think about being able to have stories – if only simple ones – about every object in your museum collection; stories that told when it was created and what it was made from, stories that shared how and when it came into the collection. This is something that Koven Smith, from the Denver Art Museum, has been thinking about. At MuseumNext 2012, he asked whether we could turn an approach like this to collections data in order to give meaning to the thousands of collections records we have that are never going to be the focus of more in-depth curatorial research. Smith proposes that rather than “changing one line of text in a book to reflect current research, [a] curator [would change] one line of code to affect a change in thousands of records at once.” While this is a very different way of imagining curatorial work, Smith argues that this is not a radical redefinition of the role, but rather the same mission as has always existed, simply “refactored to work at web scale and speed.”
And it is here, in the question of how we work at the scale and speed that the web demands that I find some of the most interesting questions in the museum sector. There aren’t enough of people working in our museums to make sense of all the stuff that we already have in our collections, much less the digital world and the new objects and ideas that we are and should be collecting. How do we, like Narrative Science, tell stories that might only be meaningful to one person? Likewise, how do we scale up our own capacity to tell important stories, and connect the knowledge we have inside our institutions to discussions happening far beyond them?
The megatrends research of Stefan Haikowicz showed how some Chinese cities have taken 20 years to come into existence, while the equivalent in Europe took 400. This increased pace of change is not dissimilar to what has been happening with digital data since the creation of the World Wide Web. The way the Web is growing mimics China more than it does Europe. In a 2010 presentation at the Palo Alto Research Center Marissa Mayer, the former Vice President of Google and now President and CEO of Yahoo!, noted that in recent years Internet data has changed across three axis, being speed (we can collect real-time data), scale (we have previously unprecedented processing power), and the inclusion of sensors that allow us to collect new kinds of data. This means that there is more data now than ever before and it’s coming from more sources than ever, human and non-human.
How can we – either as museums, or as society – make sense of it? The sheer scale of the problem creates new challenges. Whilst previously the process of filtering information and culture could be trusted to curators and academics, to publishers of books and editors of news and journals, the burgeoning data has broken the informational floodgates. (See Cairns and Birchall 2013) We need new methods to make meaning of what exists “out there”; and just as importantly, we need new ways of connecting our work, our collections and exhibitions to the global online conversation, to ensure that the important research that happens in our museums has impact far beyond their walls.
There are some tactics that we can adopt. One example comes from the Walker Art Center in Minneapolis, who hired an editor for their website whose role includes sourcing content from within and external to the museum. His role, therefore, is in part to curate the Internet in order to connect the work of the Walker to the broader art conversation. As Director Olga Viso explains, ‘Understanding that we exist as part of a diverse media ecosystem, we’ve instituted a feature called ‘Art News from Elsewhere,’ which provides a curated list of annotated links to relevant stories about contemporary art that provide greater context for the work we host and produce.’
Similarly, in an attempt to be ‘of the web’ and not just ‘on the web’ the Smithsonian Institution’s Cooper-Hewitt Museum has recently started automatically pulling in unmoderated content from external websites; so that when a visitor looks at a page with all 43 objects from creator Hugh Ferriss, he or she also sees Ferriss’ biographical information pulled in from Wikipedia. In addition, the Cooper Hewitt have started a practice of assigning unique object identifiers to all objects and creators in its online collection, which makes it possible to attach a machine-readable-tag to this unique ID. This enables members of the public who comes across a photograph of an object from the Cooper-Hewitt’s collection – or one like it – on a site like Flickr to tag the object in a way that allows the museum to digitally locate it. They then intend to ingest that information back onto the collection website. With this process, Cooper Hewitt are beginning to collect and curate the Internet using the eyes and ears and judgement of the public in order to make new meaning of and from their collection, and to connect it to the broader context in which it exist.
These kinds of approaches raise lots of questions for the museum sector. How we should be collecting the Internet? How do curate the information that exists far beyond our walls and connect it to what we have in our collections? How do we manage the context that all our work – both on and offline – now exists in? What cale of work we can achieve when our staffing numbers don’t go up, but expectations do.
Some of these are technological problems. But they’re also curatorial problems; they’re problems for museum educators. And perhaps more than anything, they’re problems that ask some foundational questions about museum practice. How do we, as institutions that value permanence and longevity so highly, come to peace with the dynamic and necessarily incomplete, impermanent and imperfect nature of the digital environment? How do we further hand trust over to people and processes that we cannot see, and know that we are still preserving what matters? And how do we collect things that can never be un-entwined from their digital environments?
Clay Shirky has, in the past, argued that ‘Institutions occasionally get to this moment… where you’re given a choice between conserving your mission and conserving your practices. Institutions tend to want to preserve the problem to which they are the solution.’
Museums stand at this junction now, where we have a choice about how we want to proceed into the future, and whether we want to play a part in preserving the most significant cultural phenomenon of our generation. The challenge for all of us is that to build something really wonderful, something truly appropriate for the future and for coming generations, we may need to be prepared to destroy at least some of what we have in the present; to blow up some elements of museum practice that have served us well until now, but that have outlived their usefulness. This is not a proposition without risk, but I feel that the risks if we don’t do it, if we fail to look towards the changes happening in the world and adapt accordingly are greater still.
