How does a museum talk? Its voice lives in wall texts, whether they deliver art history or gently admonish against touching work or using flash photography. Its tone has to be serious enough to honor the histories it was built to protect, and to convince visitors that the twenty dollars they paid to get in was well spent. An institution speaks in architecture: impressing on the outside with staid neoclassical columns or dazzling postmodern forms, and on the inside with the anonymizing solemnity of its pristine white space. The institution's voice also extends to publications, in print or online. In "Personal Voice," an essay for A.i.A.'s October 2016 issue, designer and critic Laurel Schwulst reminded readers of Brian O'Doherty's insight that "the apparent neutrality of the modernist gallery space is an ideological construct that is essential to the identity of modern art," and added: "many of the most trafficked art sites adhere to design conventions borrowed from the modernist gallery."

The relationship between a museum and its visitors-the imposing way it speaks to them-is clear enough in the lobby, in the galleries, on the website. But other spaces of communication can muddle things. On Twitter, Facebook, or Instagram, the museum becomes one of many users, just like the individuals who visit it. Taking advantage of the communication opportunities offered by social media requires accepting a diminishment in status-but this, too, can have surprising benefits.

Reading museums' social media feeds shows that self-representation online conforms to brick-and-mortar analogues. After all, museums usually use social media in broadcast mode: posts are written from the institutional perspective, and they match the style of texts that go on gallery walls, or in brochures and catalogues. Institutions employ editors to standardize their language across genres of communication-from donor appeals to publications to gift-shop marketing-and social-media use adheres to the same standards. Likewise, photographs posted to social media accounts tend to be high-resolution archival images, taken from the museum's collection or from sets of the professional installation shots used for print media, rather than iPhone snapshots or screenshots-the sort of visual content that ordinary individual users post.

It can be more interesting to consider posts and forms of behavior that aren't in broadcast mode: replies to other users, likes, retweets. The Metropolitan Museum of Art uses its Twitter account like a help desk, fielding queries and requests of all sorts from visitors to their New York home, and from curious users and other institutions. Most queries get an instant reply, acknowledging the question and asking the user to wait; a subsequent follow-up provides an answer. The social media manager consults with the relevant expert at the museum and reports back: it's social media as switchboard. (The Brooklyn Museum's Ask app, which João Enxuto and Erica Love wrote about in A.i.A.'s October 2016 issue, similarly fields visitor requests, but this is a chat app produced by the museum exclusively for use in the museum.)

Replies to tweets and comments often take a more relaxed tone than broadcasting posts. While most of their tweets employ standard orthography, institutions including the San Jose Museum of Art and New York's Museum of Modern Art often add smiley faces-the simple combination of parenthesis and colon-when replying to tweets that mention them. On rare occasions, other emoticons come in handy. When a user tweeted: "My baby just took the loudest, smelliest dump in the middle of the Magritte exhibit at the MOMA. #i<3NYC," MoMA's account replied: "We know. :(." You'd never expect a wall text to smile, but when you approach an employee at a museum-to buy a ticket at the front desk, or to ask a docent a question-the interaction is usually garnished with a smile.   

Every now and then the broadcast feed offers a glimpse of personality. When Paper magazine ran a photograph of Kim Kardashian flaunting her buttocks (headline: "Break the Internet") on the cover of its November 2014 issue, the Met responded to the frenzy around the image with an uncharacteristic tweet. It said: "Here at the Met, we have artworks that can #BreakTheInternet too! On view in gallery 150." The post included a photo of a Neolithic stone sculpture, a headless woman with an exaggerated rear end. The picture had clearly been taken in the gallery with a phone. The background was somewhat blurred, the angle a little off-not like the crisp, professional collection images the Met's Twitter account usually posts. In the ensuing thread, the Met replied to a user with a link to a Tumblr called "the Metropolitan Museum of Butts," an unofficial collection of images of rear ends on display in the galleries. (Unfortunately, the domain now shows amateur erotic images-a lesson in websites' impermanence.) When another user cut and pasted the sculpture from the photo onto the Paper cover, the Met liked it.

Experiments with institutional voice online tend to be rare, tentative, and short-lived. In "Personal Voice," Schwulst describes how the CCA Wattis in San Francisco expresses an unconventional style through the design of its homepage. But on social media sites there's hardly any room to innovate with design. CCA Wattis's use of Twitter is unremarkable, but on Instagram the institution posts screenshots of its website, to disseminate its unusual visual profile, and its use of mantras ("Andrea Fraser is on our mind" or "David Hammons is on our mind") to express its approach to exhibiting artists. Instagram isn't a place for discourse, but it can be used to remind followers of an institutional identity-in this case, screenshots of the website work like selfies.

A social media "takeover"-when the institution delegates the use of its account to an artist-can be a way to vary voice and raise interest. It's usually employed by smaller institutions, who have less at risk when they hand over their Instagram or Twitter password. However, for two days in August 2010, the Los Angeles County Museum of Art gave control of their Twitter account to actor Rainn Wilson, best known for his role as Dwight on the sitcom "The Office." The takeover was part of "Cell Phone Stories," a summer-long series of experiments with visitor experience through mobile technology (other initiatives included iPhone-sized sketches of dresses inspired by works in LACMA's permanent collection by the fashion label Rodarte and Steve Fagin's novel about the museum told through Facebook posts). By inviting Wilson, who at the time had a Twitter following of over two million, to tweet reasons not to visit LACMA, the museum hoped to mix celebrity, persona, and humor to expand its online audience. Results were mixed. Wilson's tweets were hashtagged #rainn to set them off from the rest of LACMA's feed, but their crass, silly humor was confusing to a lot of the institution's fans: "Our artists are simply pre-modern reductivists in a post-ironic milieu seeking to blah blah blah (fart) #rainn," and: "LACMA fully supports the 'Birthers' & demands to see Prez Obama's signed birth certificate! Show up @8PM for BIRTHER RALLY @LACMA. #Rainn." LACMA has not repeated the stunt since then.

Perhaps a safer version of the takeover is to give control of the account to people who aren't social-media professionals but still have a stake in the institution and its online representation. This already happens on #AskACurator day, when members of museum curatorial staffs field questions with the hashtag, whether or not they are addressed directly to the institution. Expanding that principle beyond a single "holiday" could break up a monolithic voice and convey the reality of an institution as a collection of individuals, a mixture of personalities working in counterpoint. An institution's identity can be multifaceted-stoic at some times, quirky at others-and this already comes through in the variance of broadcast mode and individually addressed replies. When the museum becomes a user, it can become more human.