Craft of Software - Home
Why Lean?
Dr. Tore Dybå
AUG 31, 2012 14:25 PM
A+ A A-


With more or less regular intervals we see new software engineering fads like maturity models, RUP, XP, and scrum. This is nothing new, and many other disciplines experience the same things. Often, however, such fads only fulfill symbolic functions such as signaling innovativeness, but do little to boost organizational performances.

One of the latest fads in software development is lean. The origin of lean is car assembly and its production system, which runs repetitively, day after day, week after week in an identical manner.

Fundamentally, lean is built around the core idea of eliminating waste through reduction in activities, staffing, and working-time buffers to improve performance. This is incompatible with software development. Eliminating waste inhibits the flexibility and improvisation that are needed to gain a crucial understanding of both the problem and its solution. Software problems are only partly specifiable and predictable. Problems tend to be unique and difficult to formulate and solutions tend to evolve continually as both developers and users gain a greater appreciation of what must be solved.

In contrast to the simplistic business environment that gave birth to the idea of lean, software development is in constant flux. It is not possible for software managers and practitioners to keep track of every feature or every technological possibility. Rather, management in software development should focus more on creating opportunities for learning and innovation rather than shortsighted cost cutting. Elimination of waste is not an effective principle to accommodate this need.

The constant pressure to eliminate waste might easily put too much emphasis on the present, with too little time and opportunity for reflection and experimentation to anticipate future challenges and to make necessary changes to respond to such challenges.

Progress depends on a deeper understanding of the fundamental differences between production and development; not a blindfolded embracement of methods for the assembly line. In my and Helen Sharp’s Voice of Evidence column in IEEE Software (Sept./Oct. 2012) we go back to the original evidence behind lean showing that evidence isn’t always what it seems to be and that popular interpretations aren’t necessarily the most accurate ones.

So, the bottom-line question is: why lean? Why embrace a fad that isn’t targeted to development and that has questionable evidence even in its original context of car assembly?


Dr. Tore Dybå is chief scientist and research manager at SINTEF and professor at the University of Oslo. He received his doctoral degree in computer and information science from the Norwegian University of Science and Technology. His research interests include empirical software engineering, software process improvement, and agile software development. His research has been extensively published in major software engineering conferences and journals including IEEE Transactions on Software Engineering, IEEE Software, Information and Software Technology, Empirical Software Engineering and Software Process: Improvement and Practice.

[%= name %]
[%= createDate %]
[%= comment %]
Share this:
Please login to enter a comment: