Personalized Exhibits

    Personalization isn't only for newcomers to cultural experiences. While the entrance is an important place to imprint visitors with a sense that they will be responded to personally and uniquely, personalization can be employed throughout the visitor experience based on their profiles or self-identification. Profile-making requires a time investment, and it should be a starting point for a deeper personal relationship with the institution, not just a one-off interaction.

    And so we move from the entry experience to the visitor path through the museum. How can the museum continuously be responsive to your changing needs and interests as you move through the experience? Museum exhibit designers and educators already do this in aggregate for all visitors by designing varied exhibit spaces, punctuating object experiences with interactives, and connecting visitors to different levels of content throughout the institution. But people derive meaning from different aspects of the museum experience. While one visitor may love messing around with a magnet interactive for the enjoyment of the mechanism, another may be fascinated by the scientific theory behind the phenomena on display. How can we serve the "right" content to each visitor?

    This is an important question not only for satisfying different types of visitors but serving visitors over time. One of the biggest complaints that visitors express about museums is that "nothing ever changes." From the perspective of a curator or historian, there is so much rich content surrounding each museum artifact that a visitor could spend a lifetime learning with one collection. But visitors don't see it that way. They see one object, one label, one story. If we can deliver content to them that is personalized to their immediate self-identification as experience seekers, students, and social agents, museums will be perceived as offering experiences relevant to different stages of life and needs. And from the staff perspective, rather than being required to create "one size fits all" interpretative content, curators and exhibit developers can then develop materials that address different audiences' needs, from the novice to the expert.

    Many art museums have experimented with offering different content streams on audio tours, featuring channels for children, channels featuring artists in their own words, channels featuring curators, and experimental channels featuring music or evocative sounds. This kind of personalization isn't just for art museums. Consider science centers. My first museum job was in the Acton Science Discovery Museum, a small hands-on science center in Acton, Massachussets. The museum was filled with really fascinating interactive exhibits, including many whose explanation eluded me despite having a degree in engineering. I found the exhibits to be beautiful and mysterious and loved exploring and playing with them. But label text was only offered at a child's level, and the resulting experience attracted families with young children exclusively. If we could have offered a different track featuring scientists' takes on the interactives, or more complex levels of interactive challenge and explanation, the same interactives might have been able to serve a more diverse audience.

    The challenge with serving different audiences personalized content often comes down to the distribution mechanism. Best practices suggest that labels should be brief and accessible (with the exception of introductory panels in art museums, which appear to be written with confuscation in mind). In the model of discovery learning, interactives are often built with a "win condition" that dictates a fairly linear experience in which the visitor tries to achieve a single goal. Visitors may find audio tours to be socially isolating and cost-prohibitive. And while many museums are experimenting with multi-media guides to the content, including those that employ visitors' own media devices, no one has cracked the code on creating an on-demand content experience that delivers visitors the information and depth level they want when they want it. Either the secondary layer is highly immersive (and therefore, constrained) or it requires so much pushing of buttons and picking of options that it breaks the overall experience. In some cases, interpretative staff can provide flexible interpretation, but these people are often locked into particular modes of content delivery or are unable to devote significant attention to individual visitors for extended periods of time.

    Some institutions have attempted to solve this problem by creating a physical device—typically associated with a barcode or RFID tag—that visitors can carry with them and use to access each exhibit and associate it with their unique identities. For example, the 2009 Brain exhibition at the Experimentarium offers each visitor an RFID ring. When a visitor first checks in at an exhibit, he is asked to make a basic profile including contact information and some information about his thought process. The system then directs him to particular exhibits based on his “brain profile” and at each exhibit, gives him personalized information about how the exhibit relates to his mental abilities. He can access any of the exhibits, but the system encourages him towards particular ones, giving a sense that there is a unique track “for him.”

    While this system offers some unique personalized aspects, it is plagued by the same challenges that have accompanied tracked exhibitions since their introduction in science centers in the early 2000s. There are two fundamental problems with these systems: they disrupt the social experience of exhibits by forcing groups to use an exhibit one by one (or to watch as a single member of the group uses the exhibit and records her experience), and they force a strict narrative on what is often a highly chaotic exhibit usage pattern. You can’t come up and use an exhibit “in the middle” if you must initialize the experience with a swipe of your tag. Particularly for families who are the primary audiences for science centers, the requirement to wait in line until other visitors are done, keep track of the tagged object, associate an exhibit experience with just one member of the family, and swipe it before each exhibit experience can be onerous. And from the institutional perspective, RFID systems are still too new to be seamlessly employed in ways that are both understandable to visitors and easy to maintain technologically.

    How could we design simpler personalized "threads" of museum experiences that are easier for visitors to integrate into their typical patterns of museum use? The ideal system would be available across the institution (not just at some special kiosks) and is associated directly with each object or exhibit. It would accommodate both individual and social use. It would "learn" what you might like over time and offer good suggestions, but it wouldn’t force you down a reductive, singular path. It would provide diverse forms of content. It would remember where you've been and integrate your self-defined interests and expertise into its offerings.

    This is no easy task. This kind of system has two parts: a rich content base of different types of interpretation for any given exhibit or artifact, and a recommendation engine that steers you to that content. Let's look at several examples of these systems, from low-tech to high-tech.

    As a low-tech example, consider "staff picks." Walk into almost any locally-owned bookstore or record store, and you are likely to find handwritten cards affixed to certain volumes featuring a few sentences from a staff member expressing his or her ardor for the content at hand. These picks are a kind of alternate aggregation of the content. In a bookstore, the books are separated by type (fiction, biography, best-seller, discount). But the staff picks are distributed across the types and provide a second layer of interpretative material, pointing out individual items and offering a more personal recommendation than that offered in the dust jacket. As a customer, you have to decide how reliable you perceive the staff to be to evaluate how you will integrate the staff picks into your browsing and purchasing. Bookstore staff are somewhere between expert book reviewers and your average Joe, but the tone of staff picks tend to be friendly, personal, and easy to follow. Some people hate the informality of staff picks. Others find it charming and useful. 

    Staff picks are a good start, but they have several limitations as a recommendation system. They are intermittent; patrons can't see reviews and thoughts on each book in the store. While in almost all cases the staff picks are labeled with the staff member's name, patrons don't have any background on individual reviewers' expertise or preferences beyond the knowledge that they work in a bookstore. And most importantly, staff picks are not personalized to each particular customer. They are functionally just an alternative layer of content that happens to be more informal and personal in tone than the base content available about every book in the store.

    Some museums have experimented with staff picks in exhibits and programs and have tried to solve the problem of the interchangability of staff picks by profiling the "pickers" more robustly. Personalization isn't just for visitors; highlighting the unique perspectives of scientists, designers, and educators both gives those staff members an identity and offers a more nuanced blend of interpretative material. Art museums have a long history of inviting curators or guest artists to aggregate custom shows from the collection that highlight particular narratives or juxtapositions. These can be done formally, as in Fred Wilson's famous Mining the Museum exhibition at the Maryland Historical Society (1992) or Damien Hirst's exhibit of favored works at the Rijksmuseum (2008). But they can also be done informally in programming. The Seattle Art Museum's "After Hours" adult program invites guest artists to lead tours of "My Favorite Things" throughout the collection. The Denver Art Museum's similar "Untitled" program features unusual experts (Jungian psychologists, perfumers, neurologists) offering "de-tours" of the collection in which they offer their own expert interpretations of the art at hand. And art museums aren't the only ones experimenting in this space; in 2008, the Exploratorium launched a few "Staff Picks" signs featuring diverse members of the staff sharing informal thoughts on what they love about particular exhibits.

    In a museum, staff picks could be expanded to highlight the distinctions between the ways that different kinds of staff interpret exhibits on display. For example, a general museum with art, history, and science content is likely to group the exhibits by those general topic headings. But it would be quite interesting to see how a design curator interprets a historical piece of furniture, or how a scientist sees a piece of landscape art. Because the roles among museum staff are more varied than those in the bookstore, there's an opportunity to promote learning from multiple perspectives using this technique.

    How would you display these different perspectives on exhibits? No one wants to clutter exhibits with long label text or multiple panels that might confuse or overwhelm visitors. The key is to make the interpretative material available via random access. Random access is a strange term to describe what is really "direct" access--information that can be consumed out of sequence. While your eye can skip around on a long label, you are more likely to feel compelled to follow the sequential narrative of the text and read it from top to bottom. But if you can select from several labels (all of which are equally easy to see or bring up), you can quickly (and randomly) access the material of greatest interest to you. Random access was the technological innovation that transformed museum audio tours from forced narratives into open-ended explorations. Museums with multiple-channel audio tours geared towards different audiences often use different visual icons for each tour, so you can see that a particular painting has an audio commentary on the teen channel and the conservator channel, whereas another sculpture in the same room might just have audio commentary for children. You can pick what you want to hear piece by piece thanks to random access.

    Random access is not always a good thing. If you are trying to immerse people in a very specific shared narrative, it's important for everyone to consume the same thread, the same emotion, the same story. The International Spy Museum is a fascinating example of this (disclosure: I worked for SPY from 2004-2007). The Spy Museum provides a multi-sensory, high immersion museum experience. Everyone takes the same winding path through the museum.  We focused the lights and the sound and the physical design to choreograph particular emotional responses to the stories and artifacts on display. And while the Spy Museum visit experience starts with a personalization activity, it is almost the opposite of the profiling discussed in this chapter. At SPY, rather than expressing some aspect of your own identity, you start by assuming someone else's identity by selecting a "cover story" in the first gallery. SPY isn't a place to be yourself. It's a place to be someone else, to mold yourself into the museum narrative.

    For many people, this design leads to a highly enjoyable experience. The Spy Museum is successful because it tells great stories well. But some visitors report feeling suffocated by aspects of the story that are too dissonant with their own perspectives and experience. Visitors with strong political feelings on the right and left complain that SPY glorifies the bad guys (whether you think the bad guys are the CIA or the Soviets). And while visitors to any history museum might complain about political bias, people feel it more acutely at SPY because they are trapped in highly immersive and persuasive environments. It's like Disneyland.  Some people believe it's the happiest place on earth, and every component that builds on that narrative gives them more delight. Others feel like they are caught in a hellish nightmare, unable to escape.

    At the extremes, random access and immersive narrative are oppositional, but there are many experiences that blend the two comfortably. There's a faulty perception that you can't have a deep narrative experience in the context of random access. Of course you can. We use random access all the time to find the things that will resonate most with us, whether at the store or flipping TV channels. In the library, we use random access to select a book off the shelf and then we curl up and immerse ourselves in the story. We turn the radio dial to a favorite song and rock out. And while some purists would say that a song is best experienced as part of an album, or an artifact as part of a set narrative path, most of us consume individual experiences randomly all day long.

    So let's go back to multi-vocal interpretative material.  The New York Museum of Modern Art (MoMA) offers random access via an audio guide, but you don't need an audio guide to introduce people to many voices. There are low-tech ways to offer multiple points of view as well. For example, consider the most frequently-used interpretative material: the label. In the last 50 years, museums have poured lots of money into developing alternative interpretative media devices to augment the exhibition experience, and yet there have been few projects focused on the physical design of labels. More people read labels than use any other interpretative content in museums, no matter how fancy the device. And yet as professionals, we mostly focus on improving the content of labels but not their technology. Labels are mostly printed in the same way on the same substrate. They have a standard length and are written in a consistent tone. There have been some minor innovations, such as the "flip label" which features a question on one surface which has to be flipped up, slid, or rotated to find the answer. Some exhibit designers hate flip labels and consider them a stupid way to pander to desires for interactivity. I like how they create a kind of suspense and use three-dimensional design to change the conditions of the narrative, however basic the interaction may be.

    What kind of label design would be most useful for presenting multiple perspectives on or versions of interpretative material? The ideal label would allow you to easily access the perspective of interest and to flip from perspective to perspective without having to follow a designated sequence. It wouldn't be a video kiosk that forces you to watch a whole 90 second video from one commenter before checking out another, or an extremely long label that provides a paragraph from each perspective one after another. The ideal multi-vocal label wouldn't take up more space than a standard museum label. It wouldn't be a whole bunch of labels clustered around the artifact. The ideal label would give equal weight to each voice (not unfairly biasing you towards the curator who got fancy font or the visitor wrote in pen on an index card). And it would let you know if there were any interesting connections or frictions between different voices, giving you an idea of where to go next.

    So here's one idea for a label design that addresses these goals. Imagine a set of window shades or projection screens, the kind that have one solid piece of fabric and can roll up into the ceiling or window frame when you pull a string at the bottom. Imagine making very small versions of these shades so that the fabric piece is the size of a typical museum exhibit label. Now, imagine mounting these shades in layers, one on top of another, a different layer for each perspective with the text printed on the shade. The shades themselves would be tinted, with each color representing a consistent perspective that could be followed throughout the museum. Hanging from each shade, instead of the same white string, is a colored tag with a one or two-word description of the perspective (i.e. the artist, biologist, preacher, loves it, hates it). Pull the tag of interest and you will bring down the label of choice. Pull another tag to see another perspective. Labels with interesting intersections or arguments could have two tags on them. That's it.

    This is clearly not the only way to make this happen.  You could print the labels tiny on slides with perspective labels and people could drop the slides into a lightbox to see them projected bigger on the wall (problem: slide theft). You could print text on the walls in different colors and give people magic glasses with different colored lenses to let them "see the content through someone else's eyes."  Or, you could do this digitally with small touchscreens where people could select the content of choice.

    No matter what technique you might use to display multiple perspectives, you will eventually run into a problem of scale. Imagine doing the herculean effort of collecting a diverse multiplicity of voices--expert and novice, artist and scientist, visitor and guard--on the majority of exhibits in your museum. Now you have a problem that is very similar to the basic wayfinding problem in museums. How would you display them? And how would visitors know which one they'd like to hear? The multi-channel audio tour or multi-shade label is manageable on the scale of a few perspectives, but is unsustainable for more than five or six alternatives. Visitors would have to remember the icons or tags and confront the boggling tyranny of choice at many stops. They might just give up.

    This problem of information overload leads to an argument for simplicity, for fewer channels, fewer stops, shorter labels, less interpretative material. But there are other ways to solve the problem, to have your thirty-seven "channels" of personal content and consume them happily too. What we need now is a recommendation engine.

Continue to the next section, or return to the outline.