Friday, June 11, 2010

Reflections on a Couple of Others’ Thoughts on Scrum Compared to Kanban

A couple of guys both longer in the tooth than me, who have authored more books than me, have also published their experienced thoughts on Scrum and Kanban this past week.

First Alan Shalloway, The Real Differences Between Scrum and Kanban.

Alan lists a number of practices that he sees as different between the two approaches: some of them are core fundamental elements of the Kanban method - the seed properties that generate the emergent Lean behavior in the organization.

Kanban differs from Scrum in the following ways…

1. Make process policies explicit
2. Visibility of process, not merely results
3. Inclusion of Management
4. Flow as an option instead of time-boxed iterations
5. Controlled change management

You need to read Alan’s full post. I feel it is fair and balanced and represents an alternative perspective that broadly follows mine

Secondly, Ken Schwaber posted his own thoughts. Ken had evidently been attending the Microsoft TechEd conference and had seen a presentation on Scrum and Kanban by Joel Semeniuk and Steven Borg. This prompted him to post, Waterfall, Lean/Kanban, and Scrum

This article talks a lot about complexity theory and understanding processes as simple, complicated, and chaotic, and how Ken’s motivation for creating Scrum was based on providing something that worked in the chaotic space. This is all fair enough and interesting in itself. However, the post seems biased by Ken’s assumption that we’re doing manufacturing kanban systems on software development. Ken has no actual experience using Kanban. His post also highlights a fundamental principle of Scrum that the underlying process for software development cannot be mapped. This point is reinforced in an older post by Tobias Mayer, Scrum and Kanban - different animals, that was referenced by Alan Shalloway.

This reveals a truly fundamental difference between Scrum and Kanban that I missed in my post yesterday or more accurately between the Scrum community and the Kanban community. It’s not an accident that Tobias’ blog is called Agile Anarchy. The underlying assumption is that trying to map the creative process that produces software is futile and there is no point in trying. It simply has to be contained in a black box. Ken refers to this as a “container” in his post and that container is a Sprint. The Kanban community believes in a scientific approach and the use of models to understand the natural philosophy of what is at work in our universe. While we recognize that all models are wrong, we recognize the value in some models. A value-stream map is a model of a process, it is probably not an entirely accurate model, more an approximation. The question is whether it is useful or not?

Once we have that model, and can observe actual behavior against it, we can then use other models such as the Theory of Constraints, Theory of Profound Knowledge, Systems Thinking and Lean’s Waste (or the economic models of transaction and coordination costs), to analyze what is happening and suggest changes. We accept that these models are incomplete and that any suggested changes will not work in all circumstances. The question is whether they work often enough to be useful?

It seems to me that we have an overwhelming quantity of evidence that has been reported from firms large and small from all corners of the world at the Lean & Kanban 2009, UK Lean 2009, Lean & Kanban Exchange, Lean Software & Systems 2010 and much more to come at Lean & Kanban Europe 2010, and LESS 2010 later this year, to demonstrate that this scientific approach is working and genuinely helping organizations and teams improve both their economic performance and also the social capital of their organizations - resulting in a better sociological outcome for everyone involved.

[I’ve decided to update this post on June 12th following a private email from a Scrummaster who has been a lifelong believer in the anarchist ideal]

It appears that few practicing Scrum followers realize the underlying anarchist ideals held by the leadership of the community and designed into Scrum as a method. It appears that this Scrummaster sees Scrum as very prescriptive…

“I think of scrum as being a very strict system of prescribed practices that are imposed whether they fit your needs or not.”

and then provides an explanation of true anarchy as

“Anarchy means “no government”, and as practiced by modern anarchists,  is a system of total self regulation, living a life of personal responsibility, and the abstention from imposition of one’s own will onto others as happens with violence, oppressive social behavior (even aggressive driving) and voting.  It is closer to a religion in this way and it is more like perfect order than chaos.”

It appears that these ideals are not obvious in Scrum but I believe it is a fair assessment to characterize at least some of the leadership in the Scrum community as true anarchists. They believe in “no government”, “total self regulation”, “a life of personal responsibility”, and “abstention from imposition of one’s own will onto others.” This particular Scrummaster went on to say that…

“I can’t even tolerate a queue that is force ranked. Too command and control.”

And this highlights an incongruence in how Scrum is articulated, taught and practiced. Both Scrummaster and Product Owner training contain advice, guidance, and practices (belonging to the role) that are considered antithetical by a true anarchist.

The second issue that became obvious to me from this email exchange is that few people appear to understand my linking of the “use of models” and “science.” Perhaps this is a reflection of how poorly science is taught in our schools? So let me explain. Newton’s Law of Gravity is a model for a small part of the natural philosophy of the universe. It explains, in part, how things are attracted to each other and stick together, and why objects fall to earth. Newton’s Law of Gravity is known to be in inaccurate model. It only works at relatively large scale and has needed modified several times since Newton created it to accommodate concepts like distance from the center of gravity, varying levels of gravitational pull, and the other core forces of the Universe that were not known to Newton in his day. However, we need to ask whether Newton’s Law is useful or not despite its lack of completeness? The answer to this is surely “Yes!” for many common everyday commercial, scientific and military applications around planet Earth Newton’s Law is perfectly adequate. It is this pragmatic approach that is commonly held in the Kanban community. Models can be useful. They can allow us to make improvements even when they do not work in all circumstances. There are many in the Agile/Scrum communities who do not share this pragmatism and believe that in the absence of a complete model that defines the full natural philosophy of the domain that only an emergent solution based on empirical observation, experimentation and feedback is possible. The concept of predicting an outcome based on a scientific model is not acceptable nor does it exist within their paradigm. This can lead to discussions where individuals talk past each other. Kanban community folks using a scientific paradigm talking with Scrum advocates with an Edge of Chaos paradigm fail to comprehend what the other is saying. This pattern has happened frequently in the Kanbandev Yahoo! group when new folks arriving with a strong Scrum background try to understand Kanban by force fitting it to their existing mindset.

It appears that Scrum leaders like Ken and Tobias openly position Scrum as as a solution for Chaos, that values Anarchy and Revolution. While Lean and Kanban leaders like me, Alan Shalloway and Donald Reinertsen position Kanban and Lean as a scientific approach that approximates chaos as merely complicated or complex and delivers an Economic, Scientific and Evolutionary approach. This juxtaposition seems useful as a selection criteria for organizations. Do you wish to use a framework for change that tames chaos through anarchy and revolution, or a framework for change that models complexity incompletely but usefully and uses science to generate significant evolutionary economic and sociological improvement? (even if those improvements are theoretically less than optimal?) Choice is good! It is only right that one approach should not fit all and that you should have the best information available in order for you to make an informed choice that is appropriate for your business and organization. It’s been refreshing that folks like Ken and Tobias have published their thoughts on how Kanban differs from Scrum and we’ve been able to raise the quantity and quality of information available to the market.