Understanding Context

by Andrew Hinton

  When Twitter was first launched in 2006, it was mainly as a supplemental service for cell phone “texting” via Short Message Service (SMS),[346] allowing friends to subscribe to one another’s messages when sent to the Twitter SMS address of 40404. For example, rather than my having to text something like “Heading to the bar on my corner to watch the game” to a bunch of different people individually (some of whom might not be interested in my status), I could just send it to 40404. Then, Twitter would forward it to friends who have subscribed to (or “followed”) my Twitter messages. The SMS origin is why tweets are limited to 140 characters: SMS protocol has a limit of 160 characters per message, and 140 leaves room for adding contextual information such as the username of the “tweeter.”

  As part of the protocol for Twitter, users can use commands of a sort to follow or unfollow people, change other preferences, and even make a specific message be a direct message instead of a public tweet, with the syntax “d username message”—“d” being the command for “make the rest of this a direct message.”

  So, even when Twitter was mainly SMS based, there were plenty of opportunities for misdirected private messages. If you receive a message from a friend or loved one on your cell phone, the natural, learned reaction is to just text her a response. It’s a conversation: we’re so used to having them without thinking about using special commands, it’s hard to learn the new habit of checking what sort of message it is you’ve received and then remembering to add “d username” before your response. Adding to the problem, the phone’s environment doesn’t provide enough information to differentiate the context. In the physical environment, we have to make a major change in our bodily position (or the volume of our voice) to go from a private conversation to a public proclamation. But, in a digitally generated environment, it takes only the difference of a typed character or two.

  Even when using a more recent smartphone and a Twitter client, it can be easy to make the mistake of responding in the wrong way. On an iPhone, for example, my Twitter client’s direct-message interface is almost identical to my phone’s native SMS interface (Figure 18-2). Recall that a digital user interface is basically graphical semantic information role-playing as physical information—a simulated structural environment.

  Figure 18-2. A direct-message exchange, on Twitter (via Tweetbot) and SMS

  If you’re explicitly attending to the details of these views as interfaces, it’s not hard to spot their differences. In Heidegger’s terms from Chapter 6, that would be the much less natural and efficient “unreadiness-to-hand.” However, when we’re having a conversation, we’re not paying much attention to the interface; we’re treating it as “ready-to-hand” by default by attending to the communication, not its container. This is another example of the tacit way in which we engage our environment, satisficing in the moment. For the act of replying to someone, my perceptual system concerns itself with only a few bits of information—the words written by the other person, the field where I can put words to answer him, and the Send button. These are the invariant elements of the environment that matter to me in the midst of the conversation. The rest of it becomes clutter, unless I’ve worked to train myself to think twice and look at the other clues. In this case, I have to check if Twitter sent me the message, which makes the difference between a public conveyance or private communication.

  The way the object is designed for interaction can have a disproportionate effect on how the place is experienced as architecture. Twitter provides an environment for communication, structured by its architecture of labels and rules, and instantiated by various client interfaces, as illustrated in Figure 18-3.

  Figure 18-3. The layers involved in a Twitter direct-message exchange

  The client application is actually an important factor; rules embodied in the client can change the context of what Twitter is to a particular user. Twitter has historically been more of a service that informs disparate objects and places rather than a place itself. It has its own website and app, but even then, many people never use the official Twitter interfaces. Back when Twitter would show up to only a hundred or so posts in a feed, I recall how confused I would be when people I followed would tweet so much it kept me from seeing what others were saying. I later discovered that they were using client applications that cached thousands of tweets from their feed—removing this environmental limitation. Spamming Twitter didn’t look like a problem from their point of view.

  The client also changes what our actions mean for one another. Twitter’s API provides the ability to mark tweets as “favorites.” Early in my own Twitter usage, I didn’t use this feature much, other than to mark tweets that contained links I wanted to read later. (Unfortunately, this ended up being a dustbin I never managed to revisit.)

  But, after I began using a smartphone app as my Twitter client, I could set the app to give me immediate notification when one of my tweets was “favorited” by someone else. I then realized that I began using favorites differently, as if everyone else also had their client set like mine. Tapping the little “favorite” star became a way to give a sort of “nod” to the tweet’s writer. I set a mode that created a new invariant, and it sets my expectation that everyone else is in the same umwelt, or environmental perspective, that I’ve created for myself. I might be having a half-imagined moment of conversation with the other person and never know it.

  The rules-architecture of a social platform has a huge impact on determining what kind of places it instantiates and what sorts of conversations and actions happen there. The rules we make manifest in social software require our explicit, careful attention. Even a simple Listserv mailing list has rule structures that have been shown to encourage and allow trolling and “flame wars” because they lack the social environmental feedback mechanisms available to us in nondigital conversation.[347]

  In his book Here Comes Everybody: The Power of Organizing Without Organizations (Penguin), Clay Shirky uses an ecological analogy that nicely speaks to the environmental perspective we’ve been exploring:

  When we change the way we communicate, we change society. The tools that a society uses to create and maintain itself are as central to human life as a hive is to a bee. Though the hive is not part of any individual bee, it is part of the colony, both shaped by and shaping the lives of its inhabitants. The hive is a social device, a piece of bee information technology that provides a platform, literally, for the communication and coordination that keeps the colony viable. Individual bees can’t be understood separately from the colony or from their shared, cocreated environment. So it is with human networks; bees make hives, we make mobile phones.[348]

  Mobile phones certainly are a factor. Just as the invention of elevators helped make skyscrapers a possibility, a whole new networked-device category enables new structures and rules that accommodate new sorts of activity.

  Regardless of device, though, place structures can still result in different sorts of conversations. An often-mentioned difference between Facebook and Twitter is that Facebook requires symmetrical friending—mutual agreement to be “friends.” But Twitter employs the looser asymmetrical following: I can follow someone, but that someone doesn’t have to follow me in return.

  That doesn’t keep many people from obsessing about who is following them or not, but it does create a contextual expectation that differs from Facebook or other symmetrical connection platforms. Of course, on Facebook, this anxiety is only increased after users catch onto the fact that even its “Most Recent” mode for its News Feed is being filtered by a hidden, undisclosed algorithm; nobody really knows who is seeing their posts, or if they’re seeing all of them from their friends. When a label that sounds so solidly invariant turns out to be procedurally generated, it tends to make us question every other supposedly solid element.

  Extending Shirky’s analogy, we do not inhabit just one hive with one set of structures and rules. Although we’ve always, in a sense, lived in multiple “hives,” each with their own
sets of rules (home, work, school, club, and so on), we now need to be simultaneously present in them all. It’s like trying to play baseball, football, and tennis, at the same time while making breakfast, while having intelligent conversations about each activity.

  Throughout any given day, a regular person might be “present” and attending to conversations in any number of places:

  A conversation about a project in a workplace email thread

  A management discussion on an employer’s intranet

  Keeping up with the news feed from friends and family on Facebook

  Discussing professional topics with other practitioners on LinkedIn

  Bantering with other fans of a sports team, under a Twitter hashtag

  Engaging in photo-sharing on Instagram

  Texting with family via phone SMS

  This isn’t an extreme example; the chances are that if you’re reading this book, you’re simultaneously plugged into roughly this many places. Each has its own interaction design conventions, and each has its own architectural “rules of order” and structural constraints as well as its own cultural norms and nuances of expression. For example, you can’t assume a “favorite” in Twitter will be interpreted the same way as a “like” on Facebook, and your sense of humor on Instagram might come across as unprofessional on LinkedIn.

  It’s important to remember, though: conversation isn’t only about “social software” or “social media.” Conversation has been the main activity of social context since long before the invention of cities, much less the arrival of the Internet. Software and its capabilities are new arrivals, but they affect the nature of the conversations we have as well as their content.[349]

  “Proxemics” as a Structural Model

  Architectures are generally social environments, so they build on the behavior patterns humans embody in social life. These are tacitly constructed, but identifiable, and they can inform how we should design all sorts of places. In the 1960s, the cultural anthropologist Edward T. Hall described how physical distance affects social communication and coined the term proxemics. Hall believed that by studying the way people interact spatially, we can learn things not only about social activity but about how to create environments that better accommodate or even encourage different sorts of sociality, such as “the organization of space in...houses and buildings, and ultimately the layout of...towns.”[350]

  Hall developed a model for distinguishing different levels of intimacy based on physical space. He stressed that the distances are not necessarily the same in all cultures, only that the emotional and social engagement between people can be correlated with physical distance, as illustrated in Figure 18-4.[351]

  Hall’s model works nicely with an embodiment perspective because the distinctions emerge from what sorts of communication are physically afforded by various proximities. Additionally, the structure is spatially nested, which fits with how we perceive the environment. Here are the communication “levels” by layer:

  Intimate

  Allows whispering, embracing

  Personal

  Face-to-face and high-touch conversations with close friends and family

  Social

  Interactions with acquaintances or friends-of-friends, such as a handshake, and the ability to hear one another clearly at a conversational volume

  Public

  Performing or speaking in public, which requires a louder voice and larger gestures

  Figure 18-4. Proxemics model of personal distance[352]

  Because none of these interactions happen without motion and activity, Hall also broke each of these into “close” and “far” phases to indicate the transitional stages we move through as we socialize through our day.

  This model isn’t all there is to proxemics, but it provides us a helpful structure for thinking through the way we create architectures online, both for social media and for other interactions, such as the overtures of a retail website to gather personal information, or make product recommendations.

  We might learn from this that we should make structures that let users gradually build trust and intimacy with one another and the environment. For example, don’t expose personal information to social-platform “friends” by default; instead, let users establish distinctions over time. Or, don’t allow a digital agent to be too helpful too soon, because it can feel like presumptuous overfamiliarity (and the assumptions can often be flat-out wrong).

  The way we define categories of for social relationships determines fundamental structures of social architecture. Flickr has always had a simple, three-tier construct that roughly aligns with proxemic distance: Contact, Friend, Family (Figure 18-5). I assume others, like me, use these with some interpretive flexibility—for example, keeping some family members as contacts and some close lifelong friends as family. But Flickr lets us interpret this for ourselves.

  Figure 18-5. Flickr’s dialog box for choosing among contact categories

  When creating a Flickr account and becoming a new user of the platform, it’s pretty clear that these three simple tiers are nested: Family sees everything; Friends see a subset of that; Contacts see a subset of what Friends see. But, it would be even better if that were reflected here as a reminder.

  LinkedIn doesn’t have an especially clear proxemic progression (Figure 18-6) represented in its relationship categories.

  The interface says, “Only invite people you know well and who know you,” but we know LinkedIn is a professional-profile-based site—so what does “know well” really mean in this context? And what if categories overlap? For example, what if we are both friends and colleagues? There are multiple dimensions at work here and not much context for why we’d choose one over another.

  Figure 18-6. LinkedIn’s dialog box for categorizing a new contact

  Hall also developed a model that explained two different ways that cultures tend to communicate and understand information: “high context” and “low context.” High-context cultures rely more on the surrounding signifiers and tacit information that informs a communication. These cultures count on consistent, invariant meanings that they assume everyone will “get” even if not spelled out explicitly. By contrast, low-context cultures need much more explicit communication; they can’t rely as much on tacit information to clarify meaning.[353]

  Any social environment’s architecture can constrain its users to being more on the high- or low-context scale, in terms of how well they understand one another. Architectures that demand all communication and context-setting be only through highly structured, predefined limits can, perhaps ironically, manage to strain out all the “noise” that would otherwise add meaningful nuance to communication. The limits of these structures can stifle social connection in the name of engineered purity. At the same time, allowing a lot of high-context communication can open up room for chaos and cruft, making it difficult to monetize or manage a social software platform. These are tough challenges, either way, but the dynamics should be understood and worked through purposefully, not ignored.

  Identity

  Each digital place has environmental elements that nudge or outright constrain us into being different sides of ourselves—our personalities, interests, and modes of expression. Not unlike the Luna Blue Hotel example in Chapter 17—in which vacancy disappeared on the Web even while it had empty rooms—our identities are also partially constructed from the databases across the Internet. For everyone who knows us only by that information, we are what it says we are.

  There are arguments and evidence from many disciplines, from philosophy to psychology, contending that a person’s individual identity is not a singular, stable thing. Whether one believes there is some immutable core to the human self, it is clear that such a core—if it exists—is not all that matters for defining who we are. Indeed, we don’t exist in a vacuum. We’re part of an environment, including many other people who experience us and have an impression of who we are. And at the same time, they influence us and our a
ctions, beliefs, and preferences in ways we’re only recently fully coming to understand.

  Back in the early days of the public Internet, researcher and theorist Sherry Turkle described how living online, in digital information structures, has made the facets of our identities explicitly delineated. This is in contrast to our offline selves, which have historically been defined more tacitly and subtly. In her 1997 book Life on the Screen: Identity in the Age of the Internet (Simon & Schuster), she explains how the Internet has brought us to a literal culmination of what postmodern theorists had been saying about identity for the prior several decades: the self is a “multiple, distributed system...a decentered self that exists in many worlds and plays many roles at the same time.” Turkle then goes on to show how these “worlds” and “roles” have gone from being tacitly emergent entities that don’t necessarily have clear boundaries, to being things we can enact in an explicitly defined way in specific, clearly bounded contexts. Using Multi-User Domains (MUDs) for her ethnographic field work, she shows how people explore different sides of themselves, acting out facets of their personalities in separate, clearly defined places—in one, a hyper-masculine and demanding warrior, while in another a fey, androgynous elf. She makes the point that this provides many options for habitation and life experience, leaving “real life” as just “one more window.”[354]