Viewpoint: Not Now, Not Like This

Two "Viewpoint" columns in the December 1999 issue of Communications present arguments in favor of the licensing of software engineers (Amr El-Kadi’s "Stop That Divorce!," and Dennis J. Frailey’s "Licensing Software Engineers"). Readers may be interested to learn the ACM Council firmly decided last May that it would not endorse licensing software engineers at this time, and that ACM would withdraw from any activity that gave the appearance of condoning the licensing of software engineers. The event that triggered this stand was the presentation of the report of the ACM Advisory Panel on Professional Licensing in Software Engineering. (The full statement of the Council motions and a copy of the panel’s report can be found at www.acm.org/serving/se_policy/report.html).

As is the case with many human activities, the licensing of software engineers has several different aspects: the scientific, the political, and the philosophical. As part of the Advisory Panel’s investigation into whether or not ACM should support the licensing of software engineers, the panel interviewed representatives of the Texas Board of Professional Engineers, one of several groups developing such licensing programs. They were asked, "Why do you want to license software engineers?" They replied, "To assure the public safety." Therefore, the science aspect of the licensing discussion is the question: Is there a test that will assure the person who passes the test will be qualified to write programs that would never endanger the public? Will that person be qualified to sign off on program designs to assure they are sound, just as building designs are signed by structural architects to assure the building is sound? The panel agreed there is no form of licensing that can be instituted today assuring the public safety. Indeed, no one knows how to do that. We do not have building codes for programs. We do not have a vocabulary of program design rich enough to discuss structural integrity. Much more research is needed before such a test can be devised.

Concern about public safety notwithstanding, the Texas Board of Professional Engineers’ licensing scheme applies only to consultants and not to the software engineers who are employed by the companies producing safety-critical systems. Therefore, it would likely have little or no impact on safety. But it could have a significant impact on readers who are consultants.

On May 16, when the Council voted, the lawsuit of the Association of Professional Engineers and Geoscientists of Newfoundland (APEGN) and the Canadian Council of Professional Engineers (CCPE) was still pending in the Canadian courts. This lawsuit is the essential political point: Who owns the word "engineer," and how shall the use of that term be regulated? AGPEN and CCPE sued the Memorial University of Newfoundland because the University began a software engineering program (under that name) that was not within the college of engineering (see www.mun.ca/ muse/vol50/iss3/news/lawsuit.html for details). This lawsuit is on hold, the parties having agreed a mediation panel will decide the issue. But the point is nonetheless quite real. As traditional engineering disciplines attract relatively fewer people than the software construction/programming disciplines attract, those groups who make their money from fees derived from licensing and accrediting engineers will look upon software engineers as a good growth area, and will attempt to assert their control. The licensing of software engineers is a part of that control. This was worrisome to several Council members.

There are two philosophical points. The first is, "Do you cooperate with something you don’t approve of?" The panel was not unanimous in their decision. Some contend it would be better to help with the licensing efforts, since it is inevitable that agencies will decide to license software engineers. The majority of panelists, however, contend that ACM’s involvement in licensing efforts would be perceived as an endorsement of those efforts, and would lull people into thinking we do know how to assure the public safety when we don’t. The Council held the latter position.

The second philosophical point is "do you begin licensing at the end of a process, or at the beginning?" It was the hope of some of the panelists that by beginning the licensing process, the field itself would mature. The Council determined that the possible harm of licensing—that is, how and whom to license—outweighs any possible good.

ACM has now withdrawn from active, direct participation in the development of a licensing examination for software engineers in the state of Texas. ACM, however, will continue to work with the IEEE Computer Society through the Software Engineering Coordinating Committee (SWECC) on the definition of an appropriate "Body of Knowledge" for software engineering and on the identification of standards of practice. While ACM expects this work will contribute and add value to the licensing efforts in Texas and elsewhere, the association has requested its name be removed from the Texas Board’s Web site and that ACM not be identified as being directly involved in the Texas Board’s licensing effort.

ACM is also taking proactive steps to address the very significant issues implicit in the licensing debate by creating two task forces (see www.acm.org/serving/se_policy/taskforce.html#bok for details). The first task force, chaired by SIGSOFT Chair David Notkin, is trying to assess the efforts to define a core body of knowledge for software engineering, with the goal of ensuring that ACM is helping the field make effective progress in this important area. The second, co-chaired by Nancy Leveson and John Knight, is studying the issue of safety critical software by evaluating all the options, not simply licensing, for ensuring public safety. They will also examine the differences among industries and countries in their approaches to safety.

The position taken by the ACM Council defines an informed and appropriate role for ACM. Regardless of how desirable licensing of software engineers might appear to be, the ACM Council opposes licensing at this time since there is no form of such licensing that will ensure public safety. However, the ACM is committed to solving the software quality problem by promoting R&D, by developing a core body of knowledge for software engineering, and by identifying standards of practice.

Back to Top