The University Record, October 26, 1992

Group editing by computer produces good results

By Deborah Gilbert
News and Information Services

Imagine three employees sitting beside each other at three computers, all of them writing and editing the same document at the same time. It may sound like “computers from hell,” but U-M researchers have created a software program, dubbed “ShrEdit,” and found that the parallel editing program results in excellent products.

“Groups working with ShrEdit generate substantially higher quality products and do so more efficiently than conventional committees or teams working with paper, pencil and whiteboard,” says Gary M. Olson, professor of psychology and director of the Cognitive Science and Machine Intelligence Laboratory (CSMIL).

ShrEdit, which allows several users to create and edit the same document to within one character of each other in the same sentence, was tested on groups of M.B.A. students in the School of Business Administration. Each group included three students or, occasionally, students and one or two of their colleagues from work.

The ShrEdit software allows users to set up a number of windows to use in the session. People typically use about three: one window shows the document that all users are working on; the second window is for shared notes they all can see; and the third window is for private notes for each user.

“While they are working, users chat among themselves about the changes and additions they are considering as well as the ones they are making in the document,” Olson says.

One-half of the 38 groups used ShrEdit at workstations face-to-face in a conference room. The other half relied on paper, pencils and a whiteboard. Each session was videotaped.

Each group was asked to draft a preliminary report designing an Automatic Post Office—a collection of postal services in a stand-alone device similar to an Automatic Teller Machine. The report had to include the core services to be offered, such as stamps, express mail and package scales; some of the required equipment; a rough cost-benefit analysis; and a list of questions still to be explored. The groups had 90 minutes to complete the task.

“The final designs created by the ShrEdit groups were better than those by the conventional groups,” Olson says.

The ShrEdit groups were more likely to complete the entire task, offer ideas that were fully fleshed out, explain ideas in clear and coherent sentences, and arrive quickly at designs that were practical and marketable.

Unexpectedly, the ShrEdit groups discussed fewer alternative possibilities, criteria and basic issues than the conventional groups. “We were very surprised by that finding, because we had always assumed that the more alternatives considered, the better the meeting and the better the product,” Olson says.

“However, since the quality of the final product was higher in the ShrEdit groups, we can only assume that the software provided a subtle vehicle for silent design production and evaluation.

“The ShrEdit groups had less to say orally about alternatives and issues but perhaps more to say by typing and reading each other’s output. The software seemed to blend the separate processes of generating, selecting and reporting ideas. It also seemed to cut the number of digressions and harebrained ideas. The ShrEdit groups stayed focused on the critical issues,” Olson adds.

If the person who “holds the pen” has the most influence over the final document, ShrEdit groups spread that influence around. “The shared influence built into ShrEdit seems to culminate in better final designs because it enhances focus, productivity and quality,” Olson says.

This study is one of a series of CSMIL studies of how technology can support group work. Future projects will include the study of “remote work” in which group members use ShrEdit in different offices, holding a “virtual meeting” that includes audio and/or visual technology.

Olson’s colleagues in the research projects include Judith S. Reitman Olson, professor of computer and information systems and of psychology; computer systems consultant Mark R. Carter; and graduate student Marianne Storrosten.

Their report on ShrEdit will appear in the Proceedings of the 1992 Conference on Computer Supported Cooperative Work.