I love this post on Unpredictable Work posted by Keith Swenson, VP for R&D at Fujitsu America. I read it three times. Keith’s post is actually an abridged email thread between himself and Jean-Jaques Dubray. To summarize and GREATLY oversimplify – one thinks some work can’t be predicted, the other thinks it can. One thinks process and activity modeling is of value, the other thinks states and transitions are what matters. Ironically – the synthesis for their opposing viewpoints may be represented by the banner image on Mr. Swneson’s own blog. My kitchen table critique of the two is that they are both essentially correct. The difference between them is not so much a lack of truth in either’s argument. As I read through the discussion it dawned on me that true understanding of any business process is analogous to the measurement of a coastline.
Complexity and predictability of a process seem to be function of an undefined measurement representing the distance of the process from an observer/modeler and the unit of measure employed in the design. The “distance” could be linear time or state transition or some other arbitrary. Using their patient example to illustrate. If I want to understand a hospital accepting a patient – exactly how much of that process do I want to understand.
Do I want to know how to get from “arriving person” to “admitted patient” or do I want to get from “injured person” to “well person.” Both are overlapping state transitions affecting the same resources but the measurement of the two is very different. One is essentially confirmed with something as simple as a database record being created, the other involves a complete understanding of the patients physiological condition. To under stand the later as clearly as the former one would necessarily need to model much more “closely.”
This example illustrates how predictability can be described as possible but remain rationally impractical. I can predictively model a process for patient admission. Living or dying as a matter of process modeling is quite another thing.
This brings me back to Mandelbrot. I am NOT a mathemation and most of my work in theoretical physics is a direct result of watching the Discovery Channel. Fractals have been applied to the stock market to be sure but I wonder if anyone specifically examined BPM for the same priniciples of self-similarity within and organizations. What do you think?
Lee Dallas
Marko Sillanpää
Yes, absolutely. I just generated that fractal (it is a Julia Set, not Mandelbrot, but who cares?) this past weekend because it symbolizes the self-similar nature of organizations. Really interesting point that a process has a fractal quality depending upon the level of detail you want to go to.
A few years ago I was in a lot of discussions about organizations and “autpoesis” which is the way that a fractal organism grows. For example a tree is a fractal: a small tree has a certain characteristic, and you can see similar characteristics in a big tree. Human organizations are the same. Trees are this way because they grow from many many nearly identical cells: human organizations grow from the work of individual humans, who organize the organization around them.
And the concept of “collaborative Planning” is the same: you distribute the job of planning, and it grows like a fractal as well.
Good discussion!
http://kswenson.wordpress.com/tag/human-process/
as you can tell I really enjoyed the thread – I didn’t even know Julia had a set 🙂 – not often that I discover an idea that makes me look at something in a totally new way anymore – very refreshing
Yes you are right, it is a matter of detail. The issue is what is the business perspective – managing only patient admissions, or managing the whole process. I think that the hospital would like to manage the whole process – but in order to that it will need tools that allow it to manage both rigorously defined processes and loosely defined processes. The problem is that many view current BPM technology as able to do both, which just isn’t correct.
I like your coastline analogy. I used a similar one – maps. Given your starting location, I can draw you a map to my house – but that really is a set of directions, not a map. I could hand you a true map, and give you my address – letting you figure out things for yourself. For me – BPM handles the first example well, the second is an unpredictable, unstructured hyman process where current BPM technology doesn’t apply.
http://blog.actionbase.com/the-difference-between-modeling-mapping-and-directions
I’m gutted. I’ve been meaning to write a blog post on the butterfly effect in project management and never found the time — although I did find a nice creative commons image of a mandelbrot. Now I find someone else has done something else better before I could even get started.
I think that while processes aren’t entirely predictable to a great level of detail, there are nevertheless issues which at first appear small but actually create an ever-increasing shift in behaviour that throws out planning completely. Project (and process) management is about trying to identify and manage these issues so that they don’t adversely affect your end goals. I’m still some way off working out how you do this though.
Any ideas?