Dealing with IT Industry

We advocate the use of formal techniques in a limited application area: the construction of grammar-centric tools based on term rewriting. This is a modest goal compared to solving software engineering problems in general by means of formal techniques. The latter requires major paradigm shifts and re-education of the users of these formal techniques. The former limits the use of formal techniques to the tools that are being developed and these tools can be fitted in traditional software engineering frameworks. Even with this limited ambition, we have to face serious technology transfer problems.

Capers Jones states that the main cause of slow technology transfer appears to be based in the normal human psychology: the human mind establishes a set of paradigms about methods, tools, and technologies. Once such a paradigm is established, all evidence to the contrary tends to be discarded. As a result, the time from first creation of an idea to widespread deployment is usually no less than 14 years [34]. Perhaps it is human to fail to understand concepts when they are approached from a different viewpoint, in a different language. This phenomenon applies to using term rewriting in the IT industry as well. It is, however, interesting that the speed of adoption for new products seems to be increasing [47].

We give an indication of the time we measured from first contacts between CWI/University of Amsterdam and IT industry until the moment that a solid relation with the begin of technology transfer was established.

$\bullet$

Work on the domain-specific language Risla has lead to solid cooperation and use in IT industry after five years.

$\bullet$

Work on software renovation factories for COBOL has lead to intensive contacts between IT industry and CWI/University of Amsterdam after three years of research.

$\bullet$

Work on a code generator for UML took over four years before an amalgamation between CWI/University of Amsterdam and the involved industrial party was ensued.

$\bullet$

Work on railway safety regulations took also over three years before solid relations within the industry were established.

$\bullet$

Cooperation between companies whose core business is reengineering does take less time before serious relations are established. The time between first contact and formal cooperation varies from a few months to a year. We assume that the Year 2000 problem is a catalyst.

Adequate functionality, robustness and efficiency are, of course, important for industrializing new technologies. In our experience, however, the key inhibitor is the long time needed for technology transfer, industrialization and commercialization. The danger being that either researchers or industry (or both!) loose their interest in the transfer. This topic is extensively discussed in the excellent textbook [33].