58
1 MessyBoard: Improving Collaboration by Decreasing Costs and Increasing Enjoyment Thesis Proposal Adam M. Fass [email protected] School of Computer Science Carnegie Mellon University Wednesday, August 6 th , 1:30 pm Newell Simon Hall, room 3305 Committee: Randy Pausch (chair) Jodi Forlizzi Jessica K. Hodgins Terry Winograd (Stanford University) Abstract Large projects require multiple people to work together. Often, these people do not communicate as much as they should. One reason is that communication takes time and effort and causes interruptions. Another reason is that work related communication can be unpleasant. The goals of this research are to reduce the costs of communication and to make it more enjoyable in order to improve collaboration. In order to achieve these goals I have created MessyBoard, a communication medium based on the metaphor of a 2D bulletin board. This medium allows people to easily use mixed media and spatial relationships to communicate their ideas. I will extend this medium to solve common communication problems, like scheduling a meeting or voting. I will also integrate the medium into peoples’ lives using large public displays, screen savers, and a Java applet that runs in any web browser so that people will see it naturally without being interrupted unexpectedly or having to explicitly set aside time.

MessyBoard: Eliminating the Costs of Collaborationafass/proposal/FassThesisProposal.doc · Web viewA group of people shares multiple “rooms”, and each room is a persistent shared

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

MessyBoard: Eliminating the Costs of Collaboration

1

MessyBoard: Improving Collaboration by Decreasing Costs and Increasing Enjoyment

Thesis Proposal

Adam M. Fass

[email protected]

School of Computer Science

Carnegie Mellon University

Wednesday, August 6th, 1:30 pm

Newell Simon Hall, room 3305

Committee:

Randy Pausch (chair)

Jodi Forlizzi

Jessica K. Hodgins

Terry Winograd (Stanford University)

Abstract

Large projects require multiple people to work together. Often, these people do not communicate as much as they should. One reason is that communication takes time and effort and causes interruptions. Another reason is that work related communication can be unpleasant. The goals of this research are to reduce the costs of communication and to make it more enjoyable in order to improve collaboration.

In order to achieve these goals I have created MessyBoard, a communication medium based on the metaphor of a 2D bulletin board. This medium allows people to easily use mixed media and spatial relationships to communicate their ideas. I will extend this medium to solve common communication problems, like scheduling a meeting or voting. I will also integrate the medium into peoples’ lives using large public displays, screen savers, and a Java applet that runs in any web browser so that people will see it naturally without being interrupted unexpectedly or having to explicitly set aside time.

To evaluate this approach, I will conduct laboratory studies and ethnographic observations of a few groups of users, and I will augment this data with surveys and activity logs from a wider audience of users.

Introduction

Projects often demand the resources and attention of multiple people. Some projects, like the development of a large software system, require too much work for one person to do in the allotted time. Others might be too complex for a single person to understand every detail, such as planning a large event. Finally, some projects require a diverse set of skills that a single person is not likely to have. For example, the creation of an interactive video game requires both technical skills and artistic talent.

People who are working together on a project must communicate with each other, and the act of communication incurs a cost. Talking face-to-face or on the phone, writing an e-mail message, using an instant messaging program, and leaving a note on someone's desk all take time and effort. People need to set aside time every day to read e-mail messages and listen to voice mail messages. If people want to meet in person, they need to find a time when all of them are free and perhaps reserve a room. These costs can be extremely high for groups that are geographically distributed, but even when a group of people works in the same location the cost can be substantial.

Many people enjoy communicating in a variety of situations, but work-related communication is often boring, inefficient and distracting. Many workers complain about having to read e-mail and attend meetings [1].

When the cost of communication is high and the activity is not enjoyable, generally people do not do it unless they are forced to, and even then they do not do it as often as they should. For example, research findings suggest that the cost of intentional communication via telephone or e-mail is much higher than the cost of a chance face-to-face meeting in the hallway, and this extra cost makes long distance collaborations more difficult [2]. If a home computer is located in an out-of-the-way room, users may be less likely to send e-mail or check for new messages [3]. Greenberg and Roseman argue that the need to move data between applications on the same computer can be a high enough cost to discourage communication [4]. Taken together, these results and arguments suggest that there are many times when a worker feels a need to communicate but does not because the cost is too high.

The goal of the present research is to lower the cost of communication and make it more fun. Lowering the cost of communication is important because if the cost is lower, people will be more likely to communicate when they feel the need, and this will lead to more effective collaboration.

Making communication more fun is important for three reasons. First, if it is more enjoyable then people are likely to do it when they feel the need. Second, people are more likely to communicate with each other even when they don't sense any work-related need, and research has shown that this can lead to increased productivity, so long as the communication is not overly distracting [2]. Third, enjoyment is important in its own right. The main focus of this work is on improving collaboration, and it is not my goal to encourage enjoyment at the expense of productivity. However, all else being equal, more pleasure in a person's life is desirable. A study by Morkes et al suggests that humor can make a collaborative task more enjoyable without a loss in efficiency [5].

This research will focus on improving communication between relatively small groups of people (less than 30) who work together and trust one another. This work does not address communication for entire organizations, nor does it address issues brought about by malicious users such as security and privacy.

My approach to lowering the cost of communication and making it more fun has three main components:

1. I will build a software communication medium that provides a networked, 2D, freeform, finite, What-You-See-Is-What-I-See (WYSIWIS) space with a simple user interface.

2. I will search for specific solutions to common communication tasks and embed them in the networked 2D medium.

3. I will look for ways to integrate the medium and specific solutions into people's lives in a natural way, such as projecting it on the wall or displaying it as a screen saver on idle workstations. The medium will run in a web browser with no software installation so that people will be able to start using it with a minimal investment of time and effort.

1. The Medium

The communication medium will be a networked 2D bulletin board. I built the initial prototype as a part of the “Information Cockpits” project [6] with a different goal in mind: my colleagues and I wanted an easy way to share pictures and project them on the wall of our lab so that they could act as memory cues in the future. For instance, if I wanted my colleague to remember a conversation that happened in the lab several months ago, I might say to her: "remember, there were pictures of fighter planes projected on the wall." Psychology studies have shown that these kinds of "context" cues can be helpful [7].

It takes a lot of time and effort for one person to find many suitable pictures to project on the wall, but I thought that if it were easy enough then everyone in the lab would contribute pictures, perhaps just for fun. I built a networked bulletin board system called MessyBoard in order to allow anyone to decorate the wall instantly [8]. The MessyBoard client runs in a window on any user's computer, and the same client runs on a computer that is connected to a projector. Putting pictures or text on MessyBoard is as simple as dragging and dropping or cutting and pasting from any other Windows application.

When I deployed MessyBoard, I observed some interesting collaborative behaviors, such as long-running asynchronous exchanges of text and pictures (Figure 1), and the playful creation of visual humor (Figure 2). I believe that these behaviors are enabled by specific properties of a 2D mixed-media shared space. Specifically, persistent notes and pictures can be used to keep a conversation grounded, spatial proximity and overlap can be used to quickly establish that a new object refers to an existing note or picture, and the ability to simply put anything anywhere is conducive to humor and creative expression.

Figure 1: A mixed media conversation on MessyBoard about the design of a drag and drop programming interface.

Figure 2: A collaborative humorous collage on MessyBoard.

2. Specific solutions to common communications tasks

Certain communications problems are both common and extremely costly, given the communication tools that people commonly use and the ways in which they are inclined to use them. For example, people often attempt to schedule meetings by sending e-mail messages back and forth. For people with busy schedules, this can be extremely inefficient. Shared calendar systems like Microsoft Outlook [9] are designed to solve this problem, but they suffer from two important drawbacks. First, a group must change its culture and practices, and some people are unwilling to share their schedules or to do the work necessary to keep them up to date [10]. Second, in some circumstances scheduling will still revert to e-mail because a calendar only allows limited forms of expression and communication.

I intend to look for common problems that people solve in costly ways, like scheduling meetings, and integrate task-specific solutions to these problems into a freeform 2D medium. A calendar widget on MessyBoard could allow people to quickly mark the times that they are not available. This solution is far more efficient than exchanging multiple e-mail messages. No cultural change is required, since users still get to choose what information they share and when, and they only do a little bit of work at the time when they are scheduling the meeting. Users can express arbitrary information like preferences or dependencies between constraints using notes that overlap the calendar or putting notes next to the calendar with arrows pointing to the relevant days and times.

Voting and sending announcements are two other common communications problems for which we may build task-specific solutions. A voting widget on MessyBoard could assure anonymity, tally the votes automatically, and make sure that everyone votes before displaying the results. An announcement widget could allow people to quickly indicate that they have read it just by clicking on it, it could show the sender who has not read it yet, and it could have an expiration date after which it would automatically delete itself. As with the calendar, the voting and announcement widgets could also benefit from being embedded in a freeform medium that allows the expression of arbitrary information. For example, people could post arguments next to the vote and questions and answers near the announcement.

3. Natural, Low-Cost Integration Into People's Lives

Some communications media, like the telephone or instant messaging, allow a sender to interrupt a receiver. Other communications media require that the users set aside time every day to receive their messages, such as e-mail or voice mail. Both of these costs are high, and people are not likely to adopt yet another communication tool that incurs one or both of these costs.

In order to eliminate the costs of interruption and setting aside time to receive messages, I intend to display my medium in places where users will observe it naturally, unavoidably and without much effort. My colleagues and I have already experimented with projecting the MessyBoard prototype on the walls of our lab and displaying it as a screen saver. I intend to look for other places where the medium can be displayed so that observing it is neither a chore nor an interruption.

Thesis Statement:

A communication medium based on the metaphor of a bulletin board with freeform layout of text and pictures can lower the cost of communication and make it more enjoyable. Integrated special-purpose solutions to common communication problems, such as scheduling a meeting, will encourage people to use the medium. Projecting the medium on the wall, displaying it as a screen saver, or otherwise integrating it into people's lives so that they see it often and without much effort will further decrease the cost of communication and make it more enjoyable. A combination of iterative software design and testing, ethnographic observations, surveys, activity logs, and laboratory studies will provide evidence to support these claims and suggest design guidelines for similar communication media.

Building a compelling communications medium

In this section, I describe ideal characteristics of a 2D networked bulletin board and argue that these characteristics serve to lower communication costs and make communication more enjoyable. I have implemented and deployed a prototype system called MessyBoard that has some of these characteristics, and some of the arguments will be based on anecdotal observations of how people are using MessyBoard. MessyBoard is far from an ideal implementation and we believe that users' behavior will change in some respects when we release a more robust and full-featured program.

2D Space, Freeform Layout

MessyBoard is a shared 2D space that allows the users to place objects wherever they want, delete objects at will, or leave objects on the board for as long as they like. These characteristics lower the cost of two important communication activities: referring to an object and grounding a conversation [11]. Figure 1 is an excellent example of both of these ideas. The green note was posted first, and another user posted a snapshot in response to the note. The fact that the snapshot refers to the note is established simply by putting the former on top of the latter. The entire conversation is grounded because the objects are persistent. In other words, it is always clear what the conversation is about because the relevant objects are near the discussion.

A freeform 2D space may also encourage certain kinds of playful and creative behavior [12]. For example, the Stage3 lab at Carnegie Mellon University and Professor Dennis Proffitt’s Perception Lab at the University of Virginia played a game of Pong (Figure 3). The paddles are actually small notes, and a bitmap of a white circle serves as the ball. There were no strict rules, since the ball did not move on its own, any user was able to move either of the paddles or the ball, and anyone could update the scoreboard. Nevertheless, people seemed to enjoy this game for a short while, and they did a lot of work to create it.

Figure 3: A game of Pong on MessyBoard. The "paddles" are notes and the "ball" is a picture of a circle. Players move the paddles and ball and update the scoreboard manually.

A freeform 2D space allows work and play to occur in the same medium, right next to each other, and closely intertwined [12]. Playful behavior such as the pong game may encourage people to pay more attention to MessyBoard and use it more often. If this is the case, they may also take part in some work-related exchanges that they otherwise would have ignored. One could also imagine a tighter integration of work and play. For example, MessyBoard could be an excellent way to play buzzword bingo [13]. This game requires participants to listen closely to what their coworkers are saying in order to pick out key words that they can use in the game.

Finite, WYSIWIS

MessyBoard has a finite amount of space with no scrolling or zooming and it enforces a strict What-You-See-Is-What-I-See (WYSIWIS) paradigm [14]. The choice of a finite space is a conscious design decision, not a technological limitation.

Infinite spaces with scrolling or zooming have obvious benefits, and several infinite collaborative spaces have been built [15, 16]. An infinite space allows an arbitrary amount of information in the medium. Zooming can allow users to see particular pieces of information or zoom out and get the “big picture”.

A finite medium with no scrolling or zooming also has some compelling benefits. When a user posts information, he can be confident that anyone else who looks at the medium will see it. When a user looks at the medium to find new information, he can be sure that he has seen everything that the last user intended him to see.

One might argue that a large 2D space could have all of the benefits of a finite space if it were augmented with a combination of visualization techniques that highlight recent changes and explicit notices. However, no such mechanism could allow a user to view an arbitrary number of changes on a single screen without expending time and effort. An infinite space encourages authors to post large volumes of content and never delete or consolidate anything, and each receiver must spend time to view all of the information and make sense of it.

A limited space with a strict WYSIWIS paradigm forces the authors of the information to think about exactly what the receivers are going to see and make sure that a quick glance will be enough. This is not as much work as it sounds: WYSIWIS is part of our natural physical environment and we make use of this unconsciously all the time. With a digital WYSIWIS medium, as on a physical wall or bulletin board, people are automatically aware that they need to put certain items in prominent places, make it obvious if something has changed, and use the space efficiently by consolidating and summarizing existing information. An alternate way to think of this is that by limiting the amount of space, we are creating a scarce resource this scarcity makes authors perceive it as valuable.

Mixed Media, Easy Import

Some kinds of collaboration require users to share information in a variety of forms: text, pictures, sounds, etc. How much they share can depend on the amount of time and effort that it takes: if the cost is too high, they will not share as much as they should. MessyBoard currently allows users to share text notes, pictures, and hyperlinks, and arrange them all arbitrarily in the same medium. This greatly reduces the cost of a mixed-media exchange like the interface design conversation like that shown in figure Figure 1, where someone responds to a text note with a screen shot and a third party responds with a mock-up annotated with a text note.

MessyBoard users have requested some additional media types, and we are planning to add support for arbitrary files, allowing MessyBoard users to upload and download anything even if MessyBoard cannot understand the file format. In addition, we may add support for sounds and movies.

Support for multiple media types in MessyBoard would be almost useless if it were mechanically difficult to get the media onto the board. If sharing an image required the user to have an image file in a known directory and use an "Open File" dialog box, at best he might occasionally share important pictures that he wants everyone to see. Many users do not keep their files organized, relying instead on their applications to keep track of which files they have been manipulating recently. These users may not be able to find a file using the “Open File” dialog box, even though they can easily edit the file with a specific application. MessyBoard must allow users to share media without knowing details about where their files are stored, what format they are in, etc.

MessyBoard allows users to post content by dragging and dropping or cutting and pasting, either from other applications or by dragging files directly onto MessyBoard. Reducing this cost may encourage behavior like that shown in Figure 4, where several artists share preliminary renderings of 3D models to be used in an animation.

Figure 4: Artists collaborate by sharing preliminary renderings of 3D models to be used in an animation.

The combination of mixed media and easy import may also make communication more enjoyable by encouraging playful behavior. For example, Figure 2 shows a collaborative collage with text, pictures, and hypertext links. The mix of images, text, and links is likely to be important here, and it is doubtful that the participants would have had as much fun if they were limited to text only. It is also hard to imagine them taking the time to do this at all if it involved more work than simply dragging the images from a web browser into the MessyBoard window.

Visual Appeal

Some users have complained that the MessyBoard prototype is not visually appealing. If the medium were more attractive, it might encourage more people to use it. This may even be the deciding factor for some users.

Complaints about the appearance of MessyBoard fall into two categories. First, some users complain that the objects in MessyBoard are simply ugly. They want notes that look more like real notes and less like flat rectangles, objects with rounded corners and drop shadows, etc. Second, a small number of users want MessyBoard to give them more creative freedom. They seem willing to invest time in order to create interesting content on MessyBoard, and they ask for features like vector drawing, painting, and support for images with transparency.

To address complaints in the first category, I intend to create stylized backgrounds and objects for MessyBoard and allow the users some choice in which objects and backgrounds they use. The challenge is in finding objects and backgrounds that make the board more attractive while still allowing the users sufficient freedom to make content distinct. As a first step, I have been working with two artists in the Stage3 research group and they have produced some concept sketches, as shown in Figure 5.

Figure 5: Concept sketches for the new MessyBoard appearance.

Addressing the second category of complaints is much simpler from a design standpoint: we simply give the users the features that they want. The new version of MessyBoard, currently in development, already supports images with transparency and it allows users to draw freeform pen strokes, and I will consider adding more advanced tools for drawing and painting. Though such features may only be utilized by a small group of expert users, these users are willing to invest substantial amounts of time and effort in order to create interesting and attractive content on MessyBoard. All users will benefit from this.

Content History

One of the biggest problems with the MessyBoard prototype is that people have no way to recover deleted material. This manifests itself in two ways. First, users are often reluctant to delete old material, even if they suspect that it is no longer useful. This leads to a lot of clutter and/or additional coordination to get everyone to agree on what can be deleted. Second, people might be reluctant to put things on MessyBoard at all, since it could be deleted at any time. Even if MessyBoard is better-suited to a given activity or piece of information, people will use a different medium to ensure that their data is safe.

To solve this problem, I intend to store all of the objects ever placed on MessyBoard and allow the user to browse through the entire history. Users will be able to see what MessyBoard looked like at any particular time, and we will add search mechanisms to make it easy to recover specific objects with queries.

A history mechanism could radically change users' behavior. People may be more aggressive about deleting old content, resulting in less clutter. People might also put new and different kinds of information on MessyBoard, secure in the fact that if it gets changed or deleted they will always have a way to retrieve it.

Solving specific communications problems

Certain communications problems are both common and extremely costly, given the communication tools that people use and the ways in which they are inclined to use them. For example, a small group of people will often attempt to schedule a meeting by sending e-mail messages back and forth. Typically, one person starts by suggesting the time that is most convenient for him. This is a natural thing to do, since it requires very little work to suggest a single time. If the rest of the people have busy schedules, it is likely that one person will not be able to meet at that time. That person typically suggests her most convenient time, and the process repeats itself. After a few rounds of suggestions and replies, all of the users will have several e-mail messages and they will have to read all of them in order to get all of the information. Even after all this, they still may not have all the information about when people are available, since most of the messages suggest a single time.

One might argue that people are simply using e-mail inefficiently. For people with busy schedules, it would be far more efficient to have everyone send all of their availability information to one person. That person could sort through all of it and select a time. However, this would require someone to step up and do this extra work, and everyone would have to send all of his or her availability information to that person. In the moment, this seems to take a lot more time and effort than simply suggesting one's most convenient time.

One could also argue that it would be more efficient to use a shared calendar system such as Microsoft Outlook [9]. However, shared calendar systems suffer from two important drawbacks. First, a group must change its culture and practices, and some people are unwilling to share their schedules or to do the work necessary to keep them up to date [10]. Getting all of the users in a workgroup to standardize on one tool is difficult, especially in circumstances where they cannot be forced. Second, in some circumstances scheduling will still revert to e-mail because a calendar only allows limited forms of expression and communication.

My colleagues and I have found an efficient way to schedule meetings on MessyBoard. First, someone puts a note on the board and creates a table of people and possible meeting times, as shown in Figure 6. One by one, people fill in the rows next to their name. It is easy to see who has replied so far and what the current status is: users simply glance at the table instead of having to read several messages. Everyone in our group is willing to use this solution because it does not require any change in our existing culture: users still choose what information they want to share and when. Since MessyBoard is a freeform medium, users can express arbitrary constraints, preferences, etc. right next to the scheduling table so the process never needs to revert to e-mail. Even if one or two users refuse to use MessyBoard, they can send e-mail and someone else can easily fill in their information. The public display of the current status also serves to remind people who have not yet shared their information and to provide some social pressure for them to do so.

Figure 6: Users use a MessyBoard note as a scheduling table in order to find a time for an upcoming meeting.

We have used this method several times and it works extremely well. However, we sometimes have to explicitly assign someone the task of creating the table, which is still more work than simply suggesting a convenient time via e-mail. Editing a text note to make it look like a table can also be tedious, since each user has to use tabs and spaces to make sure that the columns stay lined up.

The ideal way to schedule meetings may be to create a scheduling table, embedded in MessyBoard and specifically designed just for scheduling meetings. The object could automatically set up a grid and allow people to mark available times just by clicking. It could be so little work to create the object and enter personal information that people would be more inclined to use it than to suggest a time via e-mail.

By embedding a scheduling table in a freeform medium, we may also encourage additional collaboration unrelated to the scheduling task. When a user starts the application to enter his scheduling information, perhaps he will be reminded that he has some images to share with his group. It seems much easier once the application is already open, and he knows that other people will be paying attention because they all want to see when the meeting is going to occur.

Scheduling a meeting is one example of a common communication task with a high cost. We believe that there are more, including:

· Sending out an announcement and making sure that everyone reads it

· Taking a quick opinion poll (anonymously or not)

· Scheduling use of a limited resource, such as time on a shared computer or use of a conference room

All of these tasks may have simple solutions using existing media, but for various reasons people are more inclined to solve them inefficiently or simply not to do them at all because the cost is too high. I plan to look for tasks like these, implement task-specific solutions, and integrate those solutions into a 2D freeform medium. In all cases, the freeform medium will allow additional arbitrary communication that may be necessary to complete the task at hand. In addition, these objects may encourage additional collaboration by getting people to pay more attention to the medium.

I also intend to explore a class of objects that are not completely task-specific, but are useful for certain kinds of tasks. For instance, a general-purpose grid object could facilitate scheduling, making lists, or voting. Such objects may encourage groups to find creative solutions to unforeseen or idiosyncratic communication tasks.

Ideally, the combination of 2D freeform medium and task-specific objects will increase the chances that groups will use both aspects of the system for a long period of time. A group might start using MessyBoard because one user decides to use it to schedule a meeting, and then discover that they can share pictures and comment on them. Or a group might be attracted to the freeform medium and then discover that they can use it to solve certain problems very efficiently.

Integrating the medium into peoples' lives

We cannot expect users to voluntarily tolerate more interruptions or set aside more time to use a new communication medium. This section discusses how we can eliminate these costs by naturally integrating a communication medium into people's lives. By observing people’s habits and behaviors, I intend to find places where we can display a communication medium so that people will see new information when they want it, when it is most convenient, and with minimal extra work. I discuss two such places: the walls of the workplace and idle workstation displays. I plan to look for additional places as well.

Aside from reducing costs, I also argue that displaying a medium in prominent places where it can attract attention can make using it more enjoyable. I also argue that a 2D freeform WYSIWIS medium is a good way for users to contribute content that will be shown on such public displays, since people often know exactly what they want others to see.

Another cost that can prevent people from using a new communication tool is the one-time cost of installing the software. I discuss my plans to eliminate the startup cost by creating a Java version of MessyBoard that runs in most web browsers with no software installation or configuration.

Reducing the Cost

Thus far, I have argued that we can reduce the cost of communication and make it more fun by integrating a freeform 2D collaborative space with objects to solve specific communication problems. However, such a tool will be of no use to anyone if people do not pay attention to it. People already complain that the costs of reading e-mail, answering the phone and attending meetings are too high. It is likely that if users perceive any cost at all, they will not use a new tool even if they have compelling reasons to use it.

The table in Figure 7 lists several methods of communication and some costs associated with them. From this table, we see that most of these media require one of two costs for the receiver: interruption or setting aside time. Either of these costs can be deadly to a fledgling communication medium. Interruptions are distracting, and many users will turn such a medium off immediately if they can. Setting aside time to deal with incoming messages takes planning and discipline, and until people form a strong habit they will need to be reminded or they will forget. My goal is to get people to pay attention to a 2D freeform medium without incurring the costs of interruption or setting aside time.

InterruptionSetting aside

time

E-mailX

Go to office to talkX

TelephoneX

Instant MessageX

Voice MessageX

Figure 7: Several methods of communication and their costs to the person receiving the message.

An alternate approach to eliminating the costs of interruption and setting aside time is to sense the user’s attentional state in real time in order to deliver messages only when the benefits outweigh the costs of disruption [17]. This is a very difficult problem and the research is still in its early stages.

Projection on the Wall and Screen Savers

My approach is to display the medium in places where people will see it naturally, without the perception of having to do any extra work, and where they can not avoid seeing it a few times each day. Several researchers have suggested that this can be done with large projected or wall-mounted displays [18-24], and both the Stage3 research group at Carnegie Mellon University and the Denny Proffitt’s lab at the University of Virginia have informally observed that using a projector to display MessyBoard on the wall is extremely effective. We have been projecting two separate MessyBoards in the Stage3 lab, as shown in Figure 8.

Figure 8: Two MessyBoards are projected on the walls of the Stage3 lab. The one on the left is used just for Stage3, and the one on the right is shared with Dennis Proffitt’s lab at the University of Virginia.

We have also built a prototype screen saver that displays MessyBoard when a computer is idle. This is different from a large display, since the text is unreadable from a distance. However, large displays of any kind are expensive and require maintenance, and it will be difficult to convince many organizations to purchase and maintain such a display just to use a new communication medium. The screen saver makes use of existing displays when people are not using them, so the only cost is the time and effort involved in a one-time installation. If the screen saver is running on many computers in a lab or office, people are likely to glance at it several times every day. The screen saver is also fully interactive, allowing passers-by to post a note using someone else's idle computer instead of having to go back to their own. Other researchers have experimented with using the screen saver as a communication medium, and these efforts are described in the related work section.

Additional Benefits

Aside from minimizing the costs, projections and screen savers offer some more subtle benefits. Seeing a communication medium on a wall or on an idle workstation is a subtle reminder that everyone else is also paying attention to it. A person may even see others looking at it. This gives authors an extra incentive to use the medium when they have something to share. In addition, a large display on the wall can facilitate spontaneous conversations or become a meeting location where people can refer to objects by pointing and gesturing [12, 18-26].

Another benefit of projections and screen savers is that visitors to the workplace can, in a quick glance, get some information about what is going on there. Depending on how the group uses MessyBoard, visitors may learn about the work that they are doing or the attitudes of the group members. The group may even choose to put some information on MessyBoard specifically for visitors to look at.

Other Times and Places to Display the Medium

Large displays in the office or lab and screen savers are two ways to display a communications medium so that people will pay attention to it, and I believe there may be more. In general, we want to display MessyBoard so that:

· People see it with minimal effort

· People can’t avoid seeing it a few times every day

· It is not annoying or disruptive

I intend to observe users in their work environments in order to figure out where MessyBoard should be displayed. Here is a list of possible places:

· On the refrigerator door

· In the coffee room

· By the printer

· Next to the mirror

· Above the urinal or on the inside of the door of the bathroom stall

· Next to the doorknob on an office or lab door

In looking for places to display the medium, it is also useful to think about the times that people are most likely to look at it: when they are tired and taking a break, when they come back from lunch or from the bathroom, etc. A location is only good if people are there at the right times. Here are some times that people are likely to glance at MessyBoard:

· Arriving in the morning

· Leaving at the end of the day

· Switching tasks

· Taking a break

· Going to the bathroom

· Going to lunch

· Waiting for a meeting to start

It is natural to ask the question: "If users are not setting aside time to deal with messages and they are not being interrupted, how can a sender be sure that anyone will ever see his message? If he doesn't think that anyone will see it, why would he use this medium to send the message at all?" Indeed, tools like MessyBoard may in some instances need to leverage other more established tools and practices. For instance, a user may decide to start scheduling a meeting using MessyBoard and then send e-mail messages to his colleagues to make sure that they check the board. I intend to explore the idea of integrating MessyBoard with existing tools like e-mail so that the cost of using both is very low. For instance, MessyBoard could include a way to e-mail the contents of a note to all of the users. Or users could include a special address in their e-mail announcements so that their e-mail is automatically placed on the board. With the right kind of integration, the cost of using MessyBoard and e-mail together could be less than that of using e-mail alone.

My willingness to rely on established tools should not be taken as an admission of defeat, since it is common for a new medium to reach widespread acceptance by leveraging an existing medium. People often find out about web pages via Usenet newsgroups [27] and e-mail from their friends and coworkers, and many web pages now include an "e-mail this page to a friend" button. Several researchers have integrated new collaborative applications with e-mail or have considered doing so [4, 23, 24, 28-30].

Over time, if a new medium leverages an established medium and it is useful enough in its own right, users may eventually build habits around it. Integrating the medium naturally into people's lives will help to support these habits over the long term: A habit that only requires me to glance at the wall is probably more sustainable than one that requires me to log into my computer and run an application.

Integrating to make communication fun

Displaying a communication medium in a prominent place where people will notice it can make the medium more enjoyable. Consider what happens when you send a joke to a group of people via e-mail: it disappears into the ether, and you have no idea who is reading it and when. Some of them may appreciate your antics and laugh out loud. However, unless one of these people takes the time to come up with a witty reply, (which can also be perceived as more work,) you'll never know how people reacted. Someone might even laugh while you are in the room, but he is hunched over his terminal and you aren't sure whether he is reading your message or a humorous web page.

Now consider what could happen if you post the same joke on a projected display in a public space. You see someone pause to look at the projected display and he starts to giggle. Others notice this and join him at the display, pointing to it and laughing. Thus, the primary benefit of sharing a joke is realized: you get to see how other people react to it.

Projecting a communication medium on a wall can make the medium more enjoyable for another reason: people can use the medium to decorate the room. People enjoy decorating and personalizing their environments in order to make them more pleasing and to make a statement about themselves and their group. A digital medium takes much of the cost (time, effort, and money) out of this activity.

Interaction between the medium and the place where it is displayed

One might argue that large public displays and screen savers are beneficial in getting people to pay attention to any communication medium, and that a web page that everyone can edit would be just as good as a freeform 2D WYSIWIS medium. A web page, and any other medium that allows infinitely large collections of information, encourages the accumulation of large amounts of content over time with minimal thought about how easy it will be to understand the whole document. As I have argued previously, visualization techniques and explicit notices about recent changes may help in the short term, but they do not scale to allow a user to see an arbitrary number of changes in a quick glance.

A medium with a strict WYSIWIS paradigm makes users aware that others will see precisely what they put on the board and nothing else. Massive accumulation is implicitly discouraged, and people who contribute information are automatically focused on getting people's attention and making it easy for others to understand the content quickly. A freeform 2D medium gives users the power to achieve these goals: They can arrange content however they like in order to attract attention and communicate efficiently.

Startup Cost

An important cost that keeps many users from trying any new tool is the startup cost. For a software tool, this means downloading and installing the software. It may not matter how easy this is. If it is any extra work at all, a person is far more likely to put it off or forget about it than to start using it right away. People are also afraid of installing new software, and this is justifiable. Installing new software can break existing programs and infect a computer with viruses.

Software installation is sometimes a one-time cost, but not always. Workers often travel to visit family, attend conferences, etc. and they may need to use someone else's computer to communicate. A traveler may be unwilling or unable to install software on a computer that she does not own. For this reason alone, many people use web-based e-mail clients instead of more full-featured programs, since the web-based clients can be used on any computer that has a web browser.

In order to eliminate the startup cost, my colleagues and I are building a new version of MessyBoard in Java 1.1. Many computers purchased in the last few years have a web browser installed that is capable of running Java 1.1 applets with no software installation whatsoever. A user will be able to start using MessyBoard simply by entering a URL into a web browser, or perhaps simply clicking on a URL received via e-mail. Likewise, a traveler will be able to access MessyBoard on a public terminal or a relative's computer. Each MessyBoard server will have a simple URL that is easy to remember. For example, the Stage3 MessyBoard will reside at “www.messyboard.com/stage3”. Users will also be able to view a list of all boards if they forget their board’s URL.

Since Java 1.1 does not support some important features, most notably drag-and-drop from other applications, we also intend to build a stand-alone application using the latest version of Java. People who use MessyBoard frequently will probably install the stand-alone version eventually, incurring the costs mentioned above. The important point here is that they incurring these costs after they have seen the benefits of the new tool. People are much more likely to install the software after they have used it for a while than to endure the cost up front before they know what benefits they will derive from the new tool.

One of my goals is to get many people to use MessyBoard, and the importance of eliminating the startup cost cannot be emphasized enough. By eliminating the start-up cost, we not only make it easy for one person to start using it, we also make it easy for them to get their friends and co-workers to use it. All they have to do is send a URL, and anyone who clicks on it is using the tool right away. If the URL is easy to remember they may visit their coworkers' offices and tell them to type it in. Perhaps someone will ask a presenter to use it during a meeting while her display is projected on a screen for all to see. These scenarios may be merely plausible for a tool that only requires a web browser, but they are almost impossible for a tool that requires software installation.

Related WorkSimilar systems

A few systems have several features in common with MessyBoard. This section gives a brief description of each of those systems and explains how MessyBoard will differ. Unless stated otherwise, these systems are displayed on a traditional desktop workstation with a standard monitor.

TeamWave Workplace

TeamWave Workplace (formerly known as TeamRooms) is a shared freeform 2D WYSIWIS space designed for synchronous and asynchronous collaboration [4, 31]. A group of people shares multiple “rooms”, and each room is a persistent shared space in which users can draw, chat, and manipulate special-purpose “applets”. The available applets include simple tools like notes and pictures, as well as more complex tools such as the hierarchical outliner, a database interface, a shared web browser and several games. The system facilitates real-time collaboration by providing awareness information (a list of users that are currently in the room) and shared telepointers. The TeamWave Workplace server can store an arbitrary number of versions of the shared space, allowing users to access the history of a room.

TeamWave Workplace allows users to spread information among multiple rooms, while MessyBoard users must put all of their information in a small space so that the entire state can be seen in a glance, without any interaction. This design choice may make MessyBoard better suited for public displays and screen savers. TeamWave Workplace has been used on large public displays, but never as a screen saver (Mark Roseman, electronic mail, March 13, 2003).

MessyBoard will have a more advanced interface for browsing and visualizing the history of activity on the board. Though TeamWave has the capacity to save an arbitrary number of versions, the authors have not developed an interface that allows users to browse a large history quickly and easily (Mark Roseman, electronic mail, May 2, 2003).

MuSwikis

A MuSwiki is a collection of 2D freeform pages designed for asynchronous collaborative learning [32]. Users can modify existing pages or create new pages, and MuSwiki pages are hyperlinked to each other much like HTML pages. Each page is a freeform WYSIWIS 2D space that supports text boxes, bitmap images, freeform drawings, and “computational elements” that support arbitrary kinds of computation and interaction. Users do not see each others’ modifications in real time: an author makes some changes and then explicitly asks for them to be broadcast.

MessyBoard differs from MuSwikis in that it has no concept of multiple pages. As noted above, this makes MessyBoard better suited to display as a screen saver or on a large public display. In addition, a MuSwiki does not provide the user with any sort of history mechanism.

Designers’ Outpost

Designers’ Outpost is a tangible interface for supporting synchronous collaboration for the early stages of web site design [33]. The system runs on a large electronic whiteboard with rear projection. Users add information by writing on a physical note with an ordinary pen and sticking it to the board, and a camera tracks the location of the notes and stores high resolution images of them. A stylus allows users to draw links between notes with digital ink. The entire set of notes and links can be saved, and when the physical notes are gone the system can show the notes digitally. A branching history allows designers to go back to previous states and try different alternatives [34]. The system supports remote collaboration between two sites and provides presence information in the form of “shadows” and “transient ink” [35].

Designers’ Outpost is designed for synchronous collaboration on a specific task, whereas MessyBoard is designed primarily for asynchronous collaboration with a much broader audience in mind. The evaluation of Designers’ Outpost has focuses exclusively on synchronous collaboration for short periods of time, and MessyBoard will be evaluated as a tool for asynchronous collaboration over weeks or months.

Notification Collage

Notification Collage is a collaborative 2D bulletin board system for sharing notes, pictures, live video streams, and other media [36]. Notification Collage is not a strict WYSIWIS system: each user can have a unique arrangement of items on his screen, and users can individually show and hide elements without affecting other users’ displays. The authors have created a robust toolkit to facilitate the development of new media elements and modification of the medium itself, and they are currently working on a system to record the history and visualize it [37].

MessyBoard and Notification Collage are designed for similar audiences and to solve a similar set of problems, but the two systems are designed differently. The strict WYSIWIS nature of MessyBoard allows users to use spatial relationships to communicate information, while the relaxed nature of Notification Collage allows users to personalize their own displays and filter information as they see fit.

Notification Collage and MessyBoard are both shown on large public displays in addition to personal workstations, and largely for the same reasons. Notification Collage currently does not run as a screen saver.

iCom

The iCom system is designed to foster awareness of activities, synchronous conferencing, and asynchronous messaging between different physical locations [28, 38]. Each iCom station has a large projected display and a seating area with a trackball. The display contains video streams (one broadcasted from each station) and a community bulletin board. The video windows and the bulletin board window can be moved with the trackball and each station shows the same configuration (strict WYSIWIS). Users post content on the bulletin board by sending e-mail messages to a designated address. Each station broadcasts video and audio, facilitating cross-site meetings and casual interaction.

The design of iCom is driven by the desire for a subjective sense of “connection” between different physical locations, whereas MessyBoard is designed more for explicit communication and collaboration. iCom’s always-on audio and video encourages synchronous interactions, but iCom does not allow freeform expression with a direct-manipulation interface. There is currently no way to add pictures or drawings of any kind except by placing them in front of the camera, and even then they are not persistent.

What’s Happening

The What’s Happening system is designed to foster community awareness for a large organization [39]. The system is made up of two components: a communication bar that displays text messages and a screen saver that presents collages of images. The text messages on the communication bar are automatically taken from a variety of sources including announcements and popular web sites, and users can also submit their own messages. The screen saver collages are created automatically using images from public web sites in the community. Occasional “value added” collages show satellite images from weather sites, maps with traffic conditions, or a random text message from the same set of sources as the communication bar. All collages are assembled and laid out automatically. A single server creates the collages, so all clients see the same collage at any given time. The authors have run the collage screen saver on their personal workstations and on large plasma displays in public areas.

MessyBoard and the What’s Happening screen saver are similar in that they both make use of large public displays and screen savers in order to get peoples’ attention. What’s Happening creates all the content automatically, without the users having to do any extra work. MessyBoard relies on the users to do all of the work involved in content creation, and the users have complete control.

ScanBoard

ScanBoard is a prototype system designed for communication between homes [40]. A single wall-mounted unit comprises a touch screen display and a scanner. The display shows a shared WYSIWIS 2D space. Users add pictures or notes to the space by placing pieces of paper in the scanner, and they can move the notes around using the touch screen. The system allows multiple shared spaces, and the user switches between spaces using buttons on the display. Two buttons (“Forward” and “Back”) allow the user to temporarily remove recently posted items in order to view the items underneath.

ScanBoard is designed for home use, whereas MessyBoard is designed primarily for collaboration in the workplace. ScanBoard’s integrated scanner makes it easy to share existing paper artifacts such as photos and comic strips, while MessyBoard is optimized for composing new information and for importing existing digital information. The concept of an integrated public display and scanner is interesting, and perhaps this would work well for MessyBoard.

Semi-Public Displays

Semi-Public Displays are intended to aid collaboration amongst small co-located groups [41]. The system displays four separate applications on a large public display in the authors’ lab, and each application occupies about a quarter of the space. The “Reminders” application displays a slide show of requests and reminders that are automatically parsed out of the group’s weekly status reports. The “Collaboration Space” application also cycles through the help requests, and it provides a freeform space where lab members can scribble comments in digital ink using a stylus on the shard display. The “Active Portrait” provides awareness information by displaying a group photo. Each user’s keyboard activity is logged, and when a user is not active his image in the group photo slowly fades. The “Attendance Panel” presents an abstract representation of upcoming events, in the form of flowers, and how many people will be attending, represented by the colors of the petals on the flower. Users touch the petals with the stylus to indicate that they are planning to attend, and that explicit action changes the color of the petal.

MessyBoard and Semi-Public Displays are designed for a similar purpose: to aid small group collaboration. One important difference is that Semi-Public displays run only on a single shared public display, and MessyBoard runs on any number of clients simultaneously. Another difference is that Semi-Public Displays are designed to facilitate specific aspects of collaboration (awareness, help requests, event attendance) and provide content automatically from existing sources, and MessyBoard is a freeform space that relies entirely on the users for content.

Groove Pinboard

The PinBoard tool [42] is a component of Groove Workspace [15], an all-purpose collaboration tool. Groove Workspace integrates asynchronous and synchronous text messaging, voice messaging, and a number of other collaboration tools in a single application. The PinBoard tool allows users to share and decorate an infinite 2D WYSIWIS space. Users create notes, which can have background images and attached files. Notes can also be turned into polls, allowing members to vote on issues. The viewport shows a portion of the whole space, and a miniature view shows the entire board and the current location of the viewport.

MessyBoard provides a small finite 2D space, and PinBoard provides an infinite space. The infinite amount of space in PinBoard may encourage users to spread notes out over a vast area, making it less suitable than MessyBoard for public displays.

Though an entire Groove workspace can be saved as an archive at any time, this is an explicit and fairly heavyweight operation. The PinBoard tool has no lightweight history mechanism for browsing and recovering old content.

Kansas

Kansas is a collaborative programming environment designed for education [16]. A very large shared 2D space allows students to work separately and collaboratively, both synchronously and asynchronously, sometimes under the supervision of an instructor. The fully programmable environment is capable of supporting a variety of tasks, including building physics simulations and writing collaborative notes on a shared video presentation.

As with other large 2D spaces, Kansas is not well suited to projection on a single public display. Kansas aims to provide a complete learning environment for students, and the large amount of space allows students to work on their own before merging their work into a finished product. MessyBoard is focused solely on collaboration, and the design assumes that users will continue to do most of their individual work in separate applications. Kansas also differs from MessyBoard in that it provides no interface for the user to browse through the history of activity.

Shared Drawing Applications

Shared drawing applications are networked programs designed primarily to support synchronous collaboration between either co-located or remote participants [14, 40, 43-54]. Typically, the tools provide a 2D WYSIWIS space and support some subset of these features: freeform sketching, bitmaps, text, shared telepointers, scrolling over a large area, multiple pages, and marking up an existing web page or document.

These tools are designed to support synchronous meetings where people are in the same room or communicating by voice using a telephone or the Internet. Their interfaces are optimized for rapid visual communication of ideas in real time, augmented by voice or text chat, though one could conceivably set up a dedicated server for one of these programs in order to allow asynchronous use. It is possible to run any of these programs on a public display, but none of them function as a screen saver. Only one of these programs (We-Met) allows users to browse a complete history of activity [43].

Integrating Specific Solutions with a Freeform Medium

A few software systems integrate solutions to specific communication problems with a 2D freeform networked medium. The PinBoard Tool [42] for Groove Workspace [15] provides a “quick poll” object for voting, and TeamWave [4, 31] provides many specialized objects including an outliner and a calendar. The Whiteboard tool in MERBoard [47] provides image manipulation and flowchart capabilities. Kansas [16] and MuSwikis [32] are fully programmable and allow embedded objects to perform arbitrary computations.

Integrating the medium into peoples’ lives

Public Displays

A few researchers have built systems that use large public displays to support focused, time-critical collaboration. The MERBoard system [47] is designed to help NASA scientists analyze data from a rover on Mars, and the eWhiteBoard system [55] supports scheduling in a cardiac catheterization center. Neither of these systems provides a networked freeform 2D space designed for asynchronous collaboration (MERBoard includes a shared drawing application).

Many researchers have attempted to design software systems that facilitate informal information sharing, scheduling, planning and casual. Two key issues are: 1) making sure that the users do not forget about the systems, and 2) making it very little work for users to view and author shared content. A number of systems use large displays (projectors, plasma screens or physical props) in public settings to accomplish these goals [18-25, 56]. Other systems use smaller displays in key locations, such as outside office doors or meeting rooms [18, 29, 57-59]. None of these systems provide a networked freeform 2D space.

Screen Savers

Researchers and companies have created several different kinds of screen savers that deliver interesting information to the user. Some screen savers support peer-to-peer collaboration, meaning that a user can contribute content that appears on another user’s screen saver [60-64]. Other screen savers are designed to broadcast information from a central source [63, 65-68]. Another category of screen savers acts as message boards, displaying away messages and idle times and in some cases allowing visiting coworkers to leave messages on a person’s machine when the person is not in his office [62, 68-70]. None of these screen savers provide a freeform 2D WYSIWIS shared space.

Shiozawa et al. have built a screen saver that displays a shared XWindows desktop [71]. Their architecture is based on VNC [72] and the shared desktop can run any UNIX application. This is a clever and powerful approach, but the XWindows desktop is clearly not designed as a lightweight interface for sharing information. In order to display a given picture in the shared space and post comments nearby, users need to run multiple applications and move the windows around. It is not possible to draw freeform ink strokes over existing content, and the system does not maintain a history of user activity.

Lightweight Asynchronous Collaboration Tools

Wiki Wiki Web [73] and CoWeb [74] are systems that allow multiple authors to edit a shared web page using only a web browser and no additional software.

A number of commercial and research systems aim to support collaboration over the World Wide Web [50-54, 73, 75-90]. These systems typically comprise a subset of the following features:

· Document repository with version control

· Asynchronous and synchronous messaging

· Real-time application sharing (like VNC [72])

· Support for votes or polls

· Shared calendar

· Conferencing (real-time voice chat, video, shared drawing program, shared presentation with real-time annotation)

Research Plan

The main goal of my research is to support the claims that MessyBoard reduces the costs of communication and that it makes communication more enjoyable. I will gather evidence by analyzing logs of MessyBoard activity, observing users directly, and conducting laboratory studies.

This work will put me in a position to answer additional questions about the effects of MessyBoard’s startup cost (installation, configuration, etc.), the usefulness of special-purpose objects, the kinds of people who benefit most from MessyBoard, and the kinds of activities that it best supports. I will look for opportunities to answer these questions, but they are less important than the main research questions.

Main Research Questions

I intend to focus my research efforts on answering questions about two important aspects of MessyBoard:

1. The medium: A 2D freeform finite WYSIWIS space

2. Integrating the medium naturally into peoples’ lives: Projection on the wall and screen savers

For each of these aspects, I would like to answer the following two questions:

1. Does this aspect reduce the costs of communication?

2. Does this aspect make communication more enjoyable?

The matrix in Figure 9 shows the relationships between the questions and the aspects of MessyBoard and indicates the methodologies that I will use to answer each question for each aspect.

Medium

Projection,

Screen Saver

Reduce Cost?

More Enjoyable?

Laboratory

study

Laboratory

study

Ethnographic

observation

with

manipulation

Ethnographic

observation

with

manipulation

Questions

Aspects

Medium

Projection,

Screen Saver

Reduce Cost?

More Enjoyable?

Laboratory

study

Laboratory

study

Ethnographic

observation

with

manipulation

Ethnographic

observation

with

manipulation

Questions

Aspects

Figure 9: Methodologies to be used to address questions about different aspects of MessyBoard

Research Strategy

Two research methodologies will be important in answering all of the questions above: logging MessyBoard activity and ethnographic observation. For each question, I will use additional methodologies that are well-suited to the particular question.

Ethnographic observation

In order to gain a qualitative and in-depth understanding of how groups of people use MessyBoard, I will give the software to a few (between 2 and 5) carefully selected groups at Carnegie Mellon University. I plan to use the methods described by Beyer and Holtzblatt in Contextual Design [91]. These methods include surveys, interviews, and first-hand observations. I will also log all MessyBoard activity and collect quantitative data about when people are using it, how often different kinds of objects are used, etc.

I will begin by observing each of these groups before they use MessyBoard, so as to establish a baseline and be able to compare their behaviors before and after they start using the software. I plan to conduct open-ended interviews with the group leaders and several members in order to learn about the nature of the work that each group does and the relationships between the members. I also plan to distribute a survey to all of the group members in order to learn about how they currently use e-mail, instant messaging, and other electronic collaboration tools. I plan to have them fill out the same survey again after they have used MessyBoard for several weeks to see if their use of these tools has changed.

After the groups have been using MessyBoard for several weeks, I intend to interview several members again and show them logs and screen shots of their MessyBoard activity in order to refresh their memories and refer to specific events. This process will be less intrusive than interviewing the users while they use MessyBoard, and it is likely to yield more accurate and detailed information than if we asked general questions without referring to anything concrete. (Users will still be prone to forgetting and unconscious reinterpretation, and we will need to be cautious in interpreting these results [92].)

Mechanics and support

Some amount of promotion and support will be necessary in order to get groups of people to adopt and continue using MessyBoard. Even if the software has clear benefits and the cost of using it is extremely low, it is still often difficult to persuade people to try anything new. Here are some of the things that I will need to do:

· Install software on people's computers

· Install and maintain projectors

· Set up and maintain MessyBoard servers

· Give public demonstrations

· Create and maintain a web site (www.messyboard.com)

· Add features that are important to particular users or groups

Logging MessyBoard activity

The few groups that I study closely may be typical of most MessyBoard users in some ways and idiosyncratic in others. I will distribute MessyBoard to many groups, at least 20 and possibly as many as 100, in order to get a wider view of how groups use it and to put my ethnographic observations in context. These groups may be located anywhere in the world. I will not observe these groups first-hand, but the server software will log all of their MessyBoard activities and I may ask them to fill out brief surveys. (The software will warn users that their activity is being logged and they will be able to use it anonymously if they wish.) The quantitative results from these logs and surveys may corroborate or contradict the qualitative understanding that I gain from the ethnographic observations.

Measuring the effect of the MessyBoard medium on the costs of communication

If groups of people choose to use MessyBoard for long periods of time, it might be because the MessyBoard medium reduces the costs of communication compared to the tools that people currently use, or it might be for other reasons. The server logs will clearly indicate how many groups use MessyBoard for more than a few weeks, and surveys and interviews with a few groups will indicate whether people feel that MessyBoard decreases the costs of communication or if they are using it for another reason.

In order to demonstrate that a persistent, freeform 2D WYSIWIS medium can reduce the cost of communication for a given task as compared to commonly available tools like e-mail and Instant Messenger, I also intend to conduct a laboratory experiment that compares MessyBoard to other communication tools for a specific task. Subjects will work in groups of 3 or 4 to accomplish a specific goal as quickly as possible. Subjects in the control condition will use a combination of e-mail, instant messenger, and other communication tools that are commonly available to most knowledge workers. Subjects in the experimental condition will use only MessyBoard. I will measure the time to task completion and the accuracy of the groups’ solutions in order to demonstrate a quantitative difference between MessyBoard and other media.

I have begun piloting an experiment in which a group of three subjects works together to compile a set of driving directions between two locations. One group member has a map showing all of the relevant streets, another member has a map showing that some of the streets are one-way, and the third member has some information about streets that are under construction. Early results suggest that the ability to arrange all of the maps and information in a persistent space and collaborate on a single set of instructions confers an advantage over an instant messaging tool, in which pictures and text scroll off the screen as new messages come in.

Measuring the effect of the MessyBoard medium on the enjoyableness of communication

Through interviews and surveys of a few groups of MessyBoard users, I will learn whether or not people think that communicating with MessyBoard is more enjoyable than communicating with other media. Interviews will allow me to understand specific events and transactions in depth, and thus provide a qualitative understanding of how the medium makes communication more enjoyable.

By looking at a reasonable sample of the log data for all of the MessyBoard servers, I can determine whether the groups that I am studying closely have unique idiosyncratic behaviors that make their communication more or less enjoyable. More specifically, I intend to code interactions on MessyBoard as being recreational or work-related, and then compare the frequency of different kinds of interactions across MessyBoards.

I will conduct a laboratory study in order to demonstrate that the MessyBoard medium makes some tasks more enjoyable as compared with other media such as e-mail and Instant Messaging. I will ask groups of subjects to complete a suite of tasks using different media. Each group of subjects will do all of the tasks, and different groups will use different media for each task. For example, one group might use MessyBoard for task 1 and e-mail for task 2 while another group uses e-mail for task 1 and MessyBoard for task 2. After completing all of the tasks, group members will be asked to rate their enjoyment of each task on a Likert scale, and I will look for correlations between the medium and the amount of enjoyment.

Measuring the effect of integrating MessyBoard into peoples’ lives on the costs of communication

In order to examine the effects of projecting MessyBoard on the wall and displaying it as a screen saver, I intend to control these factors for the groups that I am examining closely. I plan to lend projectors to one or two of these groups in order to have them project MessyBoard in a shared workspace, and then I will remove the projector after they have used it for six to eight weeks. I can similarly control access to the screen saver feature by enabling and disabling it from the server. I will interview the people in the group to determine whether or not they think that the loss of projection or the screen saver changes the perceived cost of checking MessyBoard, and I will examine the logs for changes in the number of work-related and recreational interactions.

To corroborate my ethnographic observations, I will survey the entire population of MessyBoard users in order to ask them if they display MessyBoard on a public display in their workspace, display it as a screen saver, or both, and determine how they feel about the costs of using the medium. I will attempt to get at least one user from each board to respond to the survey in order to get a representative sample, and I may ask users to agree to fill out the survey in advance as a condition for using the software. The server logs will indicate how many people actually use the screen saver, and how often it is displayed on idle workstations.

Measuring the effect of integrating MessyBoard into peoples’ lives on the enjoyableness of communication

As with the previous question, I plan to lend projectors to one or two of these groups and then remove them, and I can enable and disable the screen saver from the server. I will interview the people in the group to determine whether or not they think that the loss of projection or the screen saver makes communication more or less enjoyable, and I will examine the logs for changes in the number of work-related and recreational interactions. I can survey the entire population of MessyBoard users in order to look for correlations between the use of projection or the screen saver and peoples’ attitudes about how much they enjoy using MessyBoard.

Additional Research questions

The four questions above constitute the main research agenda of this thesis, but this work will put me in a position to answer additional questions as well. These questions are less important than the main questions listed above, and I am not planning specific studies or manipulations to address them, but the data that I am already planning to collect may shed some light on them.

What are the best ways to reduce the startup cost of a collaboration tool, and how does this affect adoption and long-term use?

In order to get many people to use MessyBoard, I plan to make the startup cost as low as possible. MessyBoard will run in a web browser without any software installation, and each board will have a URL so that people can view the board just by clicking on a link on a web page or in an e-mail message. Since the software will be working before some of these features are in place, I may have the opportunity to observe changes in peoples’ behavior once these features are added. Regardless of whether or not I am able to gather evidence, I believe that a very low startup cost is crucial to getting people to use MessyBoard, and the software will definitely have the features mentioned above.

Can concrete communication problems be solved with special-purpose objects embedded in a freeform medium?

I have identified some common communication problems, such as scheduling a meeting, sending out an announcement, taking a poll and scheduling use of a limited resource, that could be solved with special-purpose objects on MessyBoard. I intend to identify as many of these problems as possible, design objects to solve them, iteratively refine and test them, and examine how people use them through ethnography and logging.

What sorts of groups will benefit from MessyBoard the most?

Thus far I have observed several different groups using MessyBoard for a variety of different tasks. Some groups enjoy using MessyBoard and feel that they truly benefit from it, while other groups use it for a short while and then stop. As I observe more groups, I intend to look for group characteristics that predict how much they will benefit from MessyBoard and in what ways they will use it. My initial observations suggest that MessyBoard may be best for the following kinds of groups:

· Groups of 10 to 20 members with a shared physical workspace who are working on interrelated projects

· Groups of people who collaborate remotely on a specific project

One important group characteristic is the nature of the work that they do. I anticipate that MessyBoard will be best suited to groups of knowledge workers, as defined by Kidd [93]: "… the defining characteristic of knowledge workers is that they are themselves changed by the information they process." In other words, the job of the knowledge worker is to build an understanding of new information and inform others. MessyBoard may be especially useful to knowledge workers that typically work with imagery and visual representations. MessyBoard will probably be of little use to people who carry out routine procedures on a fixed schedule with little variation. It is most likely to help groups that are constantly making unanticipated decisions and dealing with unforeseen circumstances.

Here are some other potentially important group characteristics:

· Size of group: MessyBoard has been successful with groups of 10 to 20 people, and in some situations with groups of 2 or 3.

· Strength of social bonds: MessyBoard may more useful if the group has tight social relationships, shared experiences and inside jokes.

· Number of projects: The usefulness of MessyBoard may depend on the number of different projects that the group is working on and their degree of overlap.

· Remote or co-located: MessyBoard has worked well both for remote and co-located collaboration, but the kinds of collaboration that take place on the board may be different.

· Presence or absence of a shared physical work space: A shared physical workspace may eliminate the need for some asynchronous communication, but the close proximity may also foster more digital collaboration.

· Personalities and training: Artists and Engineers may use MessyBoard differently.

· Power structure: MessyBoard may be more helpful to a group where a leader imposes strict policies, or it may be more useful in “flat” groups where workers are mostly autonomous.

· Computer skill: MessyBoard may be more appealing to novices, or perhaps computer-savvy users may be quicker to adopt it.

Another potentially important factor in the adoption and long term use of a tool like MessyBoard is the roles that individual group members play. Anecdotally, we have seen that people step into a few different roles:

· Organizer: Someone who maintains a spatial structure and deletes old content.

· Content Creator: Someone who regularly finds existing content or creates their own content for the enjoyment of others.

· Lurker: Someone who observes but rarely contributes any content.

My ethnographic observations will allow me to observe many of these group characteristics and roles firsthand, and I can gain an in-depth understanding of how these factors affect peoples’ use of MessyBoard. I can survey of the entire user population to discover some of the group characteristics that are most common, and correlate those characteristics with the amount of activity on MessyBoard and the length of time that they continue to use it.

What kinds of behaviors does MB best support?

Though I have observed that different groups use MessyBoard in different ways, I anticipate that the medium will be most useful if it supports behaviors in two broad categories. The first category is sharing information. This includes the sharing of sketches, drafts and ideas. The second category is supporting group process. This means supporting group practices like brainstorming, voting, getting feedback, and scheduling.

Supporting particular behaviors or categories of behaviors does not mean that the tool should explicitly support every concrete practice in which a group engages. Attempting this is likely to be counterproductive, as group members may not even be consciously aware of all of their important practices, or of how their actual behavior differs from explicit rules [1, 10]. Further, the communication medium that a group uses is likely to affect their practices. My approach is to offer groups a freeform medium that has the potential to support a wide range of behaviors and to let the medium and the group’s practices co-evolve. These arguments are consistent with the theory of Adaptive Structuration, which states that group practices are not static, but rather they are defined and modified through their use [1].

Another potentially important use of MessyBoard is maintaining a persistent context for collaboration. MessyBoard can serve as a single “location” where group members can store content and organizational materials. A group that has no dedicated common workspace may rely heavily on MessyBoard to remind them what they are doing from meeting to meeting. Even for groups that do have a dedicated work area, MessyBoard may be a useful way for individuals to establish context when they are working in their homes or traveling.

If groups do in fact use MessyBoard to support these different kinds of behaviors, I will be able to observe this firsthand through my ethnographic observation. I can examine a sample the logs of all of the MessyBoard servers in order to determine how prevalent certain behaviors are.

Current Status

At this time, I have a working prototype system for Windows called MessyBoard 1. This system is a networked WYSIWIS 2D space that makes it easy to import images and text using drag-and-drop or cut-and-paste. It runs in a window, full screen (for large public displays), and as a screen saver. A logging program captures a bitmap image of the board every time a change is made, and I have archived all of these bitmaps.

MessyBoard 1 has been deployed to the Stage3 lab (my own research group), the Center for Arts Management and Technology at Carnegie Mellon University, the Proffitt Perception Lab at the University of Virginia, the Information Interfaces group at Georgia Tech and Innovative Technical Solutions, Inc. (NovaSol) in Hawaii.

MessyBoard 1 has been useful as a prototype, but the code base is not flexible enough to serve as a platform for further rapid prototyping. I and several members of the Stage3 research group are building a new version of MessyBoard from scratch called MessyBoard 2. The new version is designed for extensibility, and we will deploy it using Java Applet and WebStart technology so that the users will get new upgrades and features automatically.

Detailed Plan

Figure 10 shows the concrete steps that I will take to complete this research, along with their dependencies.

Deploy MB 1

Initial Exploration

Build And Test MB 2

Design MB User Interface

Complete MB 2 Architecture

Deploy MB 2

Contextual Inquiry

Laboratory Studies

Deploy MB 1

Initial Exploration

Build And Test MB 2

Design MB User Interface

Complete MB 2 Architecture

Deploy MB 2

Contextual Inquiry

Laboratory Studies

Figure 10: Research plan tasks and dependencies.

In reality the steps are not so clearly separated, so some of them can be carried out in parallel. Figure 11 shows when I plan to work on each of the tasks and for how long.

Deploy MB 1

Initial

Exploration

Build And Test MB 2

Design MB

User Interface

Complete MB 2

Architecture

Deploy MB 2

Contextual

Inquiry 2

Laboratory

Experiments

Write Thesis

May 2003

Sept. 2003

Jan 2003

May 2004

Deploy MB 1

Initial

Exploration

Build And Test MB 2

Design MB

User Interface

Complete MB 2

Architecture

Deploy MB 2

Contextual

Inquiry 2

Laboratory

Experiments

Write Thesis

May 2003

Sept. 2003

Jan 2003

May 2004

Figure 11: Research plan timeline.

The text below explains the steps in more detail.

Deploy MessyBoard 1

This step has been completed. I and other members of the Stage3 research group have installed projectors in the Stage3 lab to project MessyBoard all day long, and the Dennis Proffitt’s lab at the University of Virginia have created a similar setup in their lab. I have identified several groups that could make use of MessyBoard and I have given presentations and demonstrations, set up servers, and installed the client software for people.

Initial Exploration for MessyBoard 1

This step is nearly complete. I have spoken informally with members of the Stage3 research group and Dennis Proffitt’s lab about how they use MessyBoard and how to improve it. A staff member has conducted formal interviews with members of the Center for Arts Management and Technology group after they started using MessyBoard, and we intend to conduct follow-up interviews now that they have been using it for a long time. This step has provided most of the ideas and insight expressed in this document.

Though we have logged all of the activity on all of the MessyBoard servers, the logs are stored as bitmap snapshots of the board and these are difficult to analyze without going through each snapshot manually. MessyBoard 2 will correct this problem by storing detailed meta-data in a format that facilitates automatic analysis.

Design the MessyBoard 2 User Interface

Based on information from the previous step, I will collaborate with artists and designers to design a new user interface. Jodi Forlizzi and Peter Scupelli (a staff member in the Stage3 research group) have sketched several design ideas and features, and two artists in the Stage3 research group, Tiffany Pomarico and Mike Darga, have created concept art to help in finding an attractive look and feel. I plan to deploy the entire interface as early as possible and iteratively refine it.

Complete the MessyBoard 2 Architecture

The MessyBoard 2 architecture currently supports most of the features present in MessyBoard 1. I will extend the new architecture design to support the following features:

· A history mechanism that automatically records all activity and supports interactive browsing, searching, and visualization of activity

· Uploading large files to the MessyBoard server for quick and easy sharing of documents, code, etc.

· User identities, preferences, and lightweight security

Build and Test MessyBoard 2

The current MessyBoard 2 implementation works, but it cannot be deployed in its current state. With the help of research programmers in the Stage3 research group, I will extend the implementation of MessyBoard 2 in the following ways:

· Create a robust server that automatically creates backups

· Create a logging system that records all user activity for research purposes

· Use Java WebStart for automatic distribution of new features

· Create a Web interface for starting and managing MessyBoard servers

As I complete new features and fix bugs, I will deploy new versions to the Stage3 research group and Dennis Proffit’s lab.

Deploy MessyBoard 2

Once the major features are complete and the system is stable, I plan to release the system to more users and promote it through presentations and demonstrations. Besides the groups that already use MessyBoard 1, I am considering some additional groups, including service organizations, small companies, and students in project courses.

Contextual Inquiry for MessyBoard 2

This phase will be more rigorous than the initial exploration. I intend to study groups both before and after th