Answer Garden

Voice Card  -  Volume 12  -  Paul Card Number 5  -  Thu, Jan 4, 1990 11:37 PM

From: DECWRL::"rws@expo.lcs.mit.edu"
To: advisory@expo.lcs.mit.edu
Subj: The Answer Garden

At the last Advisory meeting I agreed to send out proposals for funding once they were approved. Here is a proposal that has just been approved:


Thomas W. Malone
Center for Coordination Science
Massachusetts Institute of Technology

Mark S. Ackerman
Center for Coordination Science
Project Athena
Massachusetts Institute of Technology


Two of the considerable problems in using X are finding a starting point when learning X initially and understanding the site dependencies in using X. Net news has an early continuous flow of similar, naive questions. Users, even in research and development laboratories, have difficulty in distinguishing MIT releases from vendor releases, locating source materials at their site, knowing about bug fixes, and finding the proper people to ask.

The objective of this project is to design and develop an information tool that will alleviate these problems. It is our conjecture that the same 100 or 500 or 1000 questions are asked by most naive users. A semi-structured database of these questions and answers would be of considerable help.

Furthermore, this tool, which we have called the Answer Garden, would provide the platform for substantial research into group and organizational information gathering and dissemination. The questions about X cannot be answered merely at the Consortium level; they also involve dependencies upon the particular organization and site. The Answer Garden will provide an ability to collect and organize information at the Consortium, organizational, group, and even individual level. An important consideration in this project is how people might organize their information using a mechanism that allows easy change and addition of materials.


Most current systems for online information sharing provide only a fairly simple kind of category structure into which items are grouped linearly. For instance, the Athena On-Line Consultant (OLC) has a long list of commonly asked questions and answers, and bulletin board systems (like the net news) provide a set of topics, but within each topic all messages appear in the order in which they are entered. This means, for example, that the answers to a question might be scattered erratically through a long series of messages that come after the original question.

Answer Garden will provide an easy way of constructing and viewing branching networks of semistructured text items. In some cases, the branching network will be created automatically as the text items are entered. For instance, when someone enters an answer to a question in a bulletin board discussion, the answer could be automatically be linked to the question, and later viewers can easily see all the answers to a given question.

In other cases, a more efficient branching network can be constructed retrospectively by someone familiar with an area. Users would benefit from having an expert organize the information for them. As the information needs of a group become more obvious, a well-constructed branching network of multiple choice questions that would guide the user quickly to the specific answer to his question could be constructed.

Since any node can be active, if the question has not previously been answered, the node can ask a few questions of the user and point them towards the person who specializes in that kind of question. In many groups, just knowing the right person of whom to ask questions is very useful.

In a similar manner, nodes in the network can be constructed actively, from files or structured databases. Organizations can use this mechanism to publish source availabilities, source locations, bug fixes, toolkit widget descriptions, and the like. Site administrators have expressed considerable interest in this feature.

One of the appealing properties of such a system is that it could grow "organically" with well-honed branching networks being developed for heavily used and well-understood parts of the system and ad hoc networks growing up as needed in lightly used or "frontier" areas.

If the user does not find the answer to his question, he can tell the system so at any point. The user will be able to send his question by e-mail to the proper expert for the inquiry. We plan to automate the reverse procedure as well. The expert should just be able to mail back his response, and the system should place the node correctly. In general, asking questions and answering them must be easy, allowing the system grow easily.


Answer Garden will be constructed in C, using the Athena widgets and subclasses. The standard Unix mail and file facilities will be utilized; no special or additional databases or systems will be required.

Answer Garden is not part of the Object Lens system. The Object Lens system currently works only on Xerox Lisp machines; we are just beginning the port of Object Lens to Unix. We expect, however, that Answer Garden will greatly influence many architectural decisions for the Unix version of Object Lens.

New widgets either constructed or under construction for Answer Garden include:

Knedit The text layout widget. This knows a superset of a subset of Scribe, a popular markup language. It can handle multiple fonts and command button layout. Additional features will include layout of other widgets, Scribe @style commands, and resizeable text. The Knedit widget has callbacks for button activation and embedded processes.
Grapher The tree layout widget. This knows how to layout a tree in linked list format or from a file. Each node is a command button. Additional features will include separable interior nodes and leaves, and insertion and deletion of nodes. The Grapher widget has a callback list for node activation.
Stretch This is a special composite that does the specialized placement of the control buttons at the bottom of the Knedit widget. Its semantics are that some buttons lie flush left, some in the center, and some lie flush right.


Direct Expenditures
Salaries (1)
RA (2) $15,855.00
UROP (3) $10,138.35
Operating Expenses
Travel (4) $5,000.00
Materials and Services (5) $3,600.00
Indirect Expenditures (6) $15,406.65
Total $50,000.00

  1. Work will be supervised by Thomas Malone. Costs of supervision charged to another research grant.
  2. One full year of a PhD student, Mark Ackerman. Funds include employee benefits.
  3. Two UROPs full time in January and 10 hours/week during the school year. One UROP for one month in the summer.
  4. Travel as required to visit beta sites and present research reports.
  5. Includes books, office supplies, xeroxing, postage, mailing and shipping, telephone, and other miscellaneous expenses.
  6. Indirect expenditures calculated on all MIT expenditures except for UROP expenses.


Thomas W. Malone

Thomas W. Malone is the Patrick J. McGovern Professor of Information Systems at the Sloan School of Management, MIT, and will be the Director of the Center for Coordination Science at MIT. He received a B.A. from Rice University in Mathematical Sciences, and an M.S. in Psychology, an M.S. in Engineering - Economic Systems, and a Ph.D. in Psychology, all from Stanford University. Before joining the MIT faculty in 1983, Professor Malone was a research scientist at the Xerox Palo Alto Research Center (PARC).

Professor Malone's current research focuses on the interdisciplinary study of coordination, including: (1) how artificial intelligence and other computer technology can help people work together in groups and organizations, and (2) how organizations can be designed to take advantage of the new capabilities provided by information technology. He is the principal designer of the Information Lens, an early example of coordination technology that uses artificial intelligence techniques to help people share information and coordinate their activities.

In addition to his research and teaching activities, Professor Malone is a member of a number of editorial boards (Human-Computer Interaction, MIS Quarterly, and Information Systems Research) and program committees (ACM Conferences on Computer-Human Interaction (CHI), ACM Conference on Office Information Systems, International Conference on Information Systems). He has been a leader in forming both the computer-supported cooperative work community (as a program committee member for both of the first two conferences) and the coordination theory community (as chairman of the planning committee for both of the first two workshops).

Professor Malone has published over 30 technical articles and book chapters, including articles in Communications of the ACM, ACM Transactions on Office Information Systems, Human-Computer Interactions, Management Science, and Operations Research.

Mark S. Ackerman

Mark S. Ackerman is a Ph.D. candidate in Information Technology at MIT. He is a research assistant with the Lens Group and a research associate with the Visual Computing Group at Project Athena.

Prior, Mr. Ackerman was a senior staff engineer at Project Athena, where he participated in the development of the Athena widget set. He has also developed a number of other X-based tools, including a PostScript previewer for Scribe output and a hypertext video browser. The Illustrated Neuroanatomy Glossary was presented as a video at the ACM Computer-Human Interaction '89 conference. Previous to being at Project Athena, Mr. Ackerman was a co-architect and implementor of a number of software products, including Atari 2600 videogames (Ms. Pacman, Galaxian, and Moon Patrol) and the first videotext system in the United States.

Mr. Ackerman has published in IEEE Software and Dr. Dobbs. He was the co-author, along with Ralph Swick, of a paper on the Athena widgets at the Usenix conference. He presented tutorials on simple widget writing at Xhibition '88 and Xhibition '89 and on advanced widget writing, with Ralph Swick, at the 1989 X Conference and at Xhibition '89. He has also trained both industry engineers and MIT staff in programming with the X Window System.