Statement for the 21st Century Engineering Consortium Workshop

Michael Lowry
NASA Ames
February 1998

[Responding to the statement by P. Flener]

     I think that the position of most at ASE'97 is that the essence of SE,
now and in the future, is to transform informal requirements into formal
artifacts. Today, the formal artifact is usually just the code - which has
a precise operational semantics. Unfortunately, in developing code, one has
to do a lot more than just develop a formal representation of the
(informal) problem one is trying to solve. The situation is like that of,
say, a control engineer -what if they immediately went from a control
problem to an analog circuit for controlling a device? (Ignore for the
moment that most control is now digital). This is how engineering was
practiced in ancient times. In modern engineering, between the informal
problems and the system, there is a hierarchy of formal abstractions
mediated by mathematics, usually supported by automated tools. We are
trying to do for SE what has already been done in most other engineering
disciplines. While a priori you might think that the major benefits will
only be compilation and consistency checking, I believe the results could
be much more profound, matching if not exceeding what Prof. Newton
indicated has been done for digital hardware design.