Archives – draft

The email archives that people accumulate are a dense, complex, and highly personal record of their past interactions. As email becomes increasingly ubiquitous, these include not only their work interactions, but also their relationships with family members, friends, doctors, teachers, etc. However, current mail clients do little to support these archives beyond providing a basic filing and searching system: the interface to this very social environment renders it cold and impersonal. The problem we are addressing in this paper is how to transform this interface into a sociable space; a place where one feels surrounded by friends, conversations, and memories. In the first section of the paper we establish the need for visualizing and interacting with email archives. We start by looking at email archive use today – how they are used, the problems that exist. In the second section we then look at what information exists within an email archive, and what part of it is amenable to computational analysis. We also look at the information that is contributed by the user and discuss alternatives to filing that would allow the user to organize email in a more intuitive and informative way. In the third section we describe and critique a series of existing prototypes, discussing the social patterns that are embedded in email, such as the rhythm of interactions, the flow of relationships and the social networks that connect individuals, and examine the significance of making these perceivable. In the fourth section we explore the metaphor of email as habitat: if email is a virtual home, how can it help its inhabitant thrive? Here, we look at how personal beliefs and histories are reified by objects in the real world and at how this can be accomplished with email visualizations in the virtual world; we examine the role of display in establishing social role and in creating personal narratives; we discuss how email visualizations can function as both a private record and public display. 1 Why visualize email archives? Email has become a habitat[8]: it is one’s virtual home away from home. People check email all day, from work, from home, and on the road. It is used to set up appointments, discuss politics, and begin romances. Although it can be used to send to-do items to oneself, etc., email is primarily social: it is a communication medium for connecting people. A great deal of social information exists within one’s personal email archives, though little of it is perceivable though the interfaces of today’s email clients. Hidden among its cc’s and bcc’s is a depiction of one’s social network, big changes is one’s contacts demarcate moves to new jobs and new cities, and varying linguistic features and exchange rhythms create a portrait of one’s acquaintances and their relationships. This paper is about designing interfaces that reveal the social information contained within email archives. More deeply, it is about transforming these archives into a personal habitat, about using them to create a unique and meaningful virtual home. Email is one of the most ubiquitous computer applications, and is arguably the most basic and fundamental form of computer mediated communication. It was invented in 1972 – and within a year three-quarters of all traffic on the ARPANET was email. Over the next few years the mail header protocols were established, and the basic functions found in today’s mail clients were developed. It is useful to note that features that today seem fundamental and essential had to be invented – and were not obvious from the start. RD, a mail client written in 1973, was the first to separate messages for reading and deleting before that, one had one long inbox file. In 1975, MSG, which has been called the first modern email program, was created by John Vittal. It could automatically address replies, forward mail, and file messages into folders. MSG was designed to emulate physical mail; it included features such as cc and bcc, used today by millions who may have never seen an actual piece of carbon paper [16]. Over a quarter century later, today’s email programs are not fundamentally different from Vittal’s pioneering design. Yet much has changed around them. Email is no longer read at teletype terminals, but on personal computers capable of rendering complex graphics in realtime and of storing gigabytes of data. Computer and network access is no longer the limited to a tiny group of American computer scientists but is a popular medium for millions of people worldwide. In 2001, an estimated 9.6 billion emails were sent each day [15]. One interpretation of this data is that email is perfect. It’s been extraordinarily successful – why change it? Email is easy to write, easy to send. The basic ASCII format at has proved quite versatile, usable for formal declarations and hasty notes, purchase orders and love letters. Its simplicity is its strength. Studies of how people use email suggest that there is, however, room for improvement. In particular, problems with managing one’s growing archive of emails are consistently cited as a concern. People have difficulty coming up with good categories for folders, they find it time consuming to file messages and once filed, often cannot find them easily when needed. Beyond the practical problems of email management, however, are some deeper issues about what the email archives can become. In the words of Hallnäs and Redström: “When computer systems change from being tools for specific use to everyday things present in our lives, we have to change focus from design for efficient use to design for meaningful presence”. Email has certainly become an everyday presence in our lives, but our tools are still only focussed on efficient use. The goal of this paper is to explore future interfaces for email archives that use this data to create a meaningful environment. 1.1 Revealing ties and context In the physical world, we are accustomed to an environment replete with detail. We read a great deal of information in our companions’ facial expressions, clothing choices, tone of voice. Our recollection of social patterns, such as who knows whom, is aided by our memory of places and events. We live amidst an interlocking series of temporal patterns such as the changing of the seasons, the cycle of weekly management meetings, our daily routine; our sense of social patterns may be measured against these rhythms. Email, by contrast, is quite spare. There is little visual detail to distinguish messages or contacts, and many socially relevant details, such as co-location on a cc list, are easily overlooked. Email flattens space and time. An email from a colleague across the hall appears no different than one from across the world. It is asynchronous – I can write to you whenever I wish to, and you will read and respond as you can and wish. Online, it can be hard to keep track of who has been privy to the same conversation, to remember when you last spoke with someone, or to have a clear sense of how frequently you communicate with particular person. Today’s email clients may provide us with an automatically generated list of contacts, but such a rolodex is missing the rich contextual information that helps us sort through and comprehend our social environment. There are significant temporal rhythms in email [37], but they can be difficult to perceive. Visualizations can provide some of the missing context by revealing the data and patterns that are hidden within the email archive. They can show when you met someone, what are the connections among your acquaintances, major shifts in your contacts. At the organizational level, they can show who works together, who is the link between groups of people, how do new employees become integrated into the company. Such visualizations are practical and useful. They help people manage their expanding network of correspondents and growing collection of messages. Beyond the immediately pragmatic, they allow people to observe and reflect upon their online social world. Email has radically changed the scale of our social world. We are likely to be in personal (though virtual) contact with hundreds or even thousands of people, a far higher number than was previously common, or indeed possible. The numbers alone indicate a need for help in keeping track of these contacts and interactions; the need is exacerbated by visually sparse and abstract nature of online interaction. Robin Dunbar has proposed that language evolved because grooming was an inefficient way to maintain social ties. Now it appears that email may make face to face communication seem quite inefficient. A recent Pew Internet study shows the most popular online social activity of college students was to forward email to friends. Such notes are a form of social grooming: their pre-made content, like that of postcards, is not the point of the message, which is simply to stay in touch; the pre-made content lets one do so quickly and easily. The Pew study showed that today’s college students were staying in touch with more of their hometown friends than did their predecessors. Before email, the effort to stay in touch was greater and when faced with the big new social circles of college, the time and effort to keep up a larger circle of friends was too great. But today, with the efficient mechanism of email, maintaining social ties is easier and less time-consuming. This widening of the social circle, in conjunction with its maintenance via the sparse medium of email, underscores the need for tools to help keep track of and understand one’s social environment. .Dunbar also predicted that electronic mail would not enlarge people’s social networks. The human mind, he said, has a limited ability to handle information about other people. Yet tools have always extended humankind’s reach beyond our native abilities. Here, we will be looking at the tools that we need to thrive in today’s highly networked and connected world. 1.2 How do people manage email archives today? There have been numerous studies of how people use email ([7][8][25][27][37][41][42][43]). They paint a picture of a communications medium of increasing ubiquity, one that is used not only for communications, but for task management, personal archiving, and contact maintenance. Broadly speaking, email seems to function quite well as a communication medium, but current tools poorly support the management of email archives MacKay [25], Whittaker and Sidner [43]and Duchenaut and Bellotti [8] observed that there is a wide range of approaches to handling large amounts of email. Some people scrupulously delete unnecessary mail, while others keep almost everything they’ve received. Some file email into folders, typically organized by topics such as sender, project or hobby, in numbers ranging up into the hundreds [8], while others do not file at all, keeping a single large inbox, sometimes archiving this inbox by date when it gets too large [25],. People said they kept email messages for a variety of reasons: as reminders of tasks to be done, as records of conversations and decision-making processes, and a history and list of contacts. It is important to note that most of these studies of email use have been conducted in work settings: the email accounts that were analyzed were work accounts and the subjects were interviewed in their offices (in one study [43], the subjects’ company was also the developer of the email client). It is likely that this setting caused the respondents to emphasize the pragmatic reasons for their behavior, to say that they are saving vast quantities of correspondence for its potential information-retrieval uses rather than for a perhaps inchoate desire to collect and save their history, no matter how mundane. Whether practical or sentimental, the function of email archives to remind was frequently mentioned, not only the immediate reminding of a task to be done, but of a relationship to be maintained, a thought to be continued. In the course of their daily work, people often reviewed portions of their inbox (which might have hundreds or thousands of emails) as they scanned for a particular piece of mail, or simply looked over the list of contacts. Although all email clients provided search tools, people preferred to use sort functions to find particular pieces of email [8], a strategy that, deliberately or not, helps email work as a memory tool by refreshing the user’s familiarity with the archive. Today’s email clients are all designed to manage email archives by filing messages into folders. Yet filing is cognitively difficult and almost all users expressed difficulty with this approach People had difficulty with generating appropriate folder labels and with remembering these labels later for filing and retrieval [43].. If one has a folder for sender X and for project Y, where should a message from X about Y be filed? People had redundant folders, folder that were too big or too small, and folders that had long since ceased to be used. People worried that filed emails would get lost, and that filed junk would be kept forever. Once an email is filed, it no longer serves the purpose of reminding. We address the problems of categorizing and sorting email with a proposal to eliminate folders for email archives, and instead marking emails with user defined attributes (discussed in detail in section 2.3). While the focus of this paper is on visualizations, the process of filing or marking messages is the primary way that users annotate their email, providing data that is quite useful for creating meaningful visualizations. A key question that these studies raise is why do people save so much email? Although some people scrupulously delete unneeded mail, many save quite a lot of mail, including some who archive almost everything. People’s responses when asked why they saved so much often indicated a fear of deleting something that someday they would want. When pressed, the responses were pragmatic – email might contain some facts or commitments that would need to be referenced later. Yet that does not seem to address why so much seemingly mundane correspondence is saved, e.g. messages planning meetings or lunches, casual exchanges, etc. Our hypothesis is that people want to save the history of their experiences, achievements, and relationships. They value the accumulation of exchanges, in which the individual messages may not be very significant, but when considered as a whole, creates a meaningful pattern. An important caveat is that these studies have been conducted on email as it exists today. So while some people are meticulous about discarding old email, that behavior is occurring in a context in which there is little reason to keep lots of old email, and many reasons to throw it away (for instance, to keep the mailboxes from becoming too unwieldy). If this mail contributed to the detail and accuracy of a personal email visualization, they may well choose to save them instead. 2 What information is in email? Any visualization is constrained and determined by the data it is given. With email, the data can come from three sources. The header information is highly standardized and computationally accessible. The message content is free-form text: linguistic analysis could derive some information from it, but it is very contextdependent. Finally, there is data supplied by the user. This is likely to be easily accessible by a computer, the question here being the user’s motivation to annotate. Here, we look at these three types of data in greater depth.