GDPA  
Interview with Prof. Lehman  

Prof. Lehman (London)

Revolution in the Software Evolution?

As the plate on the door signs, Meir M. (Manny) Lehman is Senior Research Fellow and an active Emeritus Professor (stand March 2002) at Imperial College in London.

Prof. Lehman
As a recognition for his outstanding research on process evolution he was granted with the IEEE Harlan D. Mills Award which was announced in the ICSE 2001 Conference in Toronto, Canada. The award was established in "Harlan D. Mill's name to recognise researchers and practitioners who have demonstrated long-standing, sustained, and meaningful contributions to the theory and practice of the information sciences, focusing on contributions to the practice of software engineering through the application of sound theory".

In 1969, Manny Lehman wrote the report The Programming Process which pioneered the concepts of systems evolution and feedback . (See cronological approach of the papers on software process).

As he said in an interview with William Aspray to the IEEE History Center: "Every word of that report is as true today as it was when I wrote it over twenty-five years ago... I myself didn't understand everything I was seeing until many years later. I spent nine months, doing the work, talking to hundreds of people, thinking a lot, writing it all up, producing a report... At the same time, on the basis of my findings, during the Programming Process study, I started working first on my own , and then together with Les Belady on a discovery which we called "Programming Growth Dynamics" which came directly out of my observations and data on the growth of the IBM OS360 operating system. It showed that the growth of a piece of software seemed to have a dynamics of its own which could be disciplined, analysed, and used to plan". (Lehman, 23 Sept 1993).

The computing expanding universe, starting out of the computer, then getting into the computer until it becomes a part of the system
(M. M. Lehman, 17 Dec 2001)
   

In 1971, based on this study, he made the observation that software processes must be seen and treated as feedback systems which later led to the formulation of the Laws of Software Evolution. These laws apply, in the first instance, to E-type programs. The E stands for "Evolutionary". "E-type programs address a problem in the real world. As the real world is infinite and continuous, you cannot come up with an exact or complete specification of the problem. So, you develop a system which is predestined to change and evolutionate. There will always exist some new enhancement to be done. In the other hand, S-type programs are programs which are correct with respect to a fixed and consistent specification. But, is the specification correct? The criterion of success in creating an S-type program is that it is correct in a strict mathematical sense." (Lehman, 17 Dec 2001)

Does every software have feedback? Is software development an open loop by definition?

Yes, the eighth of the laws of software evolution concerns "Feedback Systems" (1996). It states that E-type evolution process are multi-level, multi-loop, multi-agent feedback systems and must, in general, be treated as such to achieve major process improvement.

From 1996 to 2001 he was the principal investigator for the FEAST - Feedback, Evolution And Software Technology projects: FEAST/1 from 1996 to 1998 and FEAST/2 from 1999 to 2001.

FEAST Hypothesis:
(1) The software evolution process for E-type systems, which includes both software development and maintenance, constitutes a complex, feedback, learning system.
(2) The feedback nature of the evolution process explains, at least in part, the failure of forward path innovations, such as those introduced over the last decades to produce impact of the order of magnitude anticipated at the global process level.

The FEAST projects study software evolution as a phenomenon in its own right. As evidenced by the observed regularities in patterns and trends of evolutionary attributes, the FEAST observations support the hypothesis that process behaviour is constrained by its system dynamics. "We define the inverse square model to define the growth trend of a system, but we don't have any sausage machine, where you put systems parameters in on side and you get the predicted system growth in the other!"(Lehman, 17 Dec 2001)

"SETh - Approach to a theory of software evolution" is the proposed research for giving continuation and profundization to the prior results of FEAST.

The FEAST projects contributed significantly to increased understanding of the software evolution phenomenon and to spreading interest in the topic as illustrated by active research groups in the UK, USA, Europe, Australasia and Japan.

Four workshops were done:

Many Lehman also helped to initiate the first International Software Process Workshop. The sucessfull workshop set the basis to 9 further meetings.

Will a Revolution in the software Evolution take place someday?

The answer is NO, because

Which technology could bring a longterm smooth in the "infinite software evolution"?

Knowledge based systems could bring some more knowledge in the understanding of the systems. It could bring some progress in modelling the infinite real world into the finite computer world. Anyway, little knowledge is a dangerous thing!
(M. M. Lehman, 17 Dec 2001)
   

by Carla Freericks.

This page online  •  GDPA Online  •  Last Updated 12.Sep.2002 by C. Freericks