Friday, November 13, 2009

Pick a Number

#systemsthinking #deming @dpjoyce asks "Anyone know where exactly Deming's 95% came from? Was is actual data? Vanguard also quote this, what did they study?"

Deming and his followers use a figure of 95% in at least two different contexts. 
  • 95% of problems are system-related (whatever that means)
  • 95% of the so-called improvement initiatives are futile because people don’t know the profound knowledge or competency
Vanguard repeats the first claim. For example "They found that 95% of the answers were down to a problem with the system rather than the worker." (via David Joyce). But this only makes sense if you draw the boundaries of the system to exclude the worker, which many systems thinkers would find puzzling or perverse.

I have found two sources that credit the second observation to Peter Scholtes.

I tend to read this kind of claim as more rhetorical than scientific. Not only is it hard to find any empirical study that might support this kind of claim, it is difficult to see what kind of evidence might be adduced.

As I pointed out in my paper Reasoning about Systems and their Properties, to decide that one intervention is successful and another is futile is an act of interpretation, and assumes we know which system we are talking about, from whose perspective, with what timescale, and so on. In order to have any credible basis for dismissing an intervention as "futile" or "meddling" or "tampering", you actually have to do a full systems analysis on the case, and the results could still be disputed. You can't jump to conclusions - "Oh, this didn't work, so the people who attempted it obviously didn't know what they were doing" - because success and failure and their causes are extremely complex, and well-chosen interventions based on a deep and thorough systems analysis may also sometimes fail.

If this 95% were both meaningful and true, what would be the consequences for action, and what would be a reasonable target for improvement? Writing books and articles complaining about a general lack of management profundity, or the folly of "The Regime", looks suspiciously like an ineffectual meddle rather than a well-chosen intervention with a well-designed outcome. Will Deming's followers take their own medicine? See my post Easier Seddon Done.

See also Paul Hebert on Deming and Systems in Today's Business World - No Answers Just Questions, pondering whether the 95% still applies to knowledge/software or whether a 50%/50% split would fit better (via @baob)


  1. Richard:

    Sorry to rain on your parade. Page 315 of Out of the Crisis.

    I believe it would be important for you to understand variation to understand the significance of the 95/5 or 94/6. We are talking about common and special causes. The above reading can give you some background. Both Shewharts and Deming's background in statistics is important to grasp.


  2. In the preface of his book, The Leaders Handbook, Peter Scholtes says, ‘ … 95% of your organisations problems derive from your systems, processes and methods, not from your individual workers. Your people are doing their best, but their efforts cannot compensate for your inadequate and dysfunctional systems’. I think that it is important to quote Scholtes accurately before considering the validity of his message.

    For me, the above quote gives a much better context within which to have this discussion. We seem to live in a world where much of the leadership/management practice is framed by reference to the theories of Taylor (scientific management). Quoting again, this time from Delavigne & Robertson’s book Deming’s Profound Changes, neo-Taylorism leads to ‘ … a belief that best efforts will achieve optimal results’. When you say that it is ‘hard to find any empirical study that might support this support this kind of claim (95% etc.)’ what is your belief regarding the elements that combine to create overall organisational performance? Do you believe that that performance derives from best efforts?

    You say that David Joyce’s tweet can ‘only make sense if you draw the boundaries of the system to exclude the worker, which many systems thinkers would find puzzling or perverse’. Deming would agree with you, ‘A leader must understand that the system is composed of people, not mere machinery, nor activities, nor organisation charts’. Those of us who facilitate the learning from Deming’s Red Bead Experiment refer to the interaction between equipment (the paddle), environment (which includes management style), materials (the presence of the red beads), method (the angle of the paddle) and people (the willing workers).

    Using the Red Beads Experiment as an example to which many reading this post will be able to relate, we learn that the variable results stem from the interaction of equipment, environment, materials, method and people. We learn that this is all common cause variation (Shewhart, Wheeler) attributable to the system. If one worker decides to tip all of the beads back into the box and offer the inspector a £5 note to cover up his actions, the outcome is a special cause attributable to that person. [I am not making this example up!]. So overall organisational performance can be down to the actions of people. Whether that is 5% of the time and the system accounts for the other 95%, I am not too bothered. The important learning is surely that organisational performance is primarily down to an interaction and interdependence between equipment, environment, materials, method and people. This is my experience of working in organisations and that of Deming too. (Thanks Tripp for the Out of the Crisis reference)

    I feel that we do the systems-thinking/Deming community a disservice if we are seen to focus our intellect on discussions about whether 95% is supported by empirical evidence. My concern is for those significant numbers of leaders/managers who focus their effort on attempts to ‘motivate’ their employees by carrot or stick to achieve best efforts and thus seek to improve organisational performance.

    Finally, I had the privilege to know Peter Scholtes before his untimely death in July this year. To seemingly dismiss his contribution to management thinking as coming from a ‘Catholic priest’ is, in my opinion, disingenuous and suggests a failure to research his background – see The great systems thinker, Russell Ackoff (who also sadly passed away this year), in his foreword to The Leaders Handbook says, “Peter Scholtes is an educator, not a guru … ”.

    Glyn Lumley

  3. I agree that Peter Scholtes being a Catholic priest is not relevant and I have removed this remark.

  4. I have enormous respect for Deming and I think a lot of his insights are extremely valuable.

    What I'm objecting to is Deming's followers latching onto arbitrary numbers like 95% as if they really meant something. I am happy to accept that lots of managers don't understand systems very well, but I don't think it is useful or meaningful to try and count the number who are this or that side of some subjective threshold of understanding.

    I also object to the false precision of how many problems are inside or outside "The System". There are many ways of understanding sociotechnical systems, including the workers as part of the system, as Deming himself clearly understood, but Vanguard training materials (as quoted by David Joyce) seem to assume that the worker is outside "the system", which kind of misses the point.

  5. Richard:

    You make a good point about the worker. The worker is part of the system (boundary wise). That actually is our point, the worker can not be separated from the system, unless they are a statistical outlier. If they are, they may need help.

    Empirical evidence is definitely not there, the way you would like to see it. As systems are different, I would not draw individual system conclusions without empirical evidence. However, from the data I have collected from over the years (from all types of organizations) I have data that supports the 94/6, 95/5 or something in that arena enough to know that treating a special causes as common cause or vice versa makes things worse. To blame a person with out understanding the insight of variation and understanding systems is to work on the wrong problem.


    We have something in common, I met Peter in Indianapolis on several occasions. He wasn't a Guru (no doubt), but probably one of the nicest men you'd want to meet.

    Also, thanks for taking the time to write all that, I am trying to pack.

    Regards, Tripp

  6. Presumably the purpose of talking about (and teaching) systems thinking is to encourage and enable people to manage systems change better - in other words we all want make an improvement in a very broad system called Managing-Systems.

    Glyn asks what I believe about the elements that combine to create overall organization performance. What I believe is that these elements combine in complex and context-specific ways, and I tend to be suspicious of quick and easy answers to this kind of question. My question back is what do we know about organizational performance, and how do we know what we know.

    Tripp says that there isn't any empirical evidence, at least of the kind he thinks I'm looking for. But why not, after all this time? Some people (let say 5% for the sake of argument, I just pulled that number out of the air) may be persuaded to adopt "Systems Thinking" on the basis of the authority of a guru, some plausible rhetoric, and/or a few simple anecdotes. But a lot of people (let's say 95%) are not willing to adopt difficult techniques without some concrete evidence for its efficacy.

    A lot of people in the systems thinking world are simply preaching to the converted, encouraging would-be systems thinkers to use their "best efforts" to follow systems thinking guidelines, and publishing despairing notes about the folly of the remaining 95%. What effect is this kind of activity likely to have on the system called Managing-Systems?

  7. "But a lot of people (let's say 95%) are not willing to adopt difficult techniques without some concrete evidence for its efficacy."

    Actually, my experience tells me that a lot of people (let's say somewhere near 90% of the 95%) are not willing (or even able) to adopt difficult techniques EVEN WITH EXTENSIVE concrete evidence for its efficacy. This is the world we live in.

    - @flowchainsensei

  8. As Bob says, even concrete evidence won't convince everyone. But it would help convince some - especially those who pay allegiance to a certain form of rationality. (Some people in the systems thinking world sneer at rationality, which is why rational people don't take them seriously.)

    Tripp sent me a tweet saying if I wanted the data I could wade through Deming's papers myself. No thank you, that's a task for the Deming community itself.

    Any community of practice that disdains empirical analysis of its results is not only cutting itself off from people and organizations that are more likely to be persuaded by solid empirical data than by anecdote and rhetoric, but also cutting itself off from an important source of collective learning and self-improvement.

    So it shocks me if people who claim to be systems thinkers are willing to rely on decades-old rhetoric rather than systems science for the cornerstones of their practices.

  9. By the way, Tripp thinks I'm only arguing because I don't understand something about variation and its causes.

    As a matter of fact, I do understand the distinction between common-cause variation and special-cause variation, and I can see that it can be useful in some systems engineering situations, but the distinction is not as simple as many Demingites appear to believe.

    JR Lucas' analysis of probability is relevant here. In his Reason and Reality (Chapter 5, Projection and Probability, pdf) he uses the example of cot death to argue against the concept of coincidence. Translating this example into Demingite language, we could say that the experts thought that there was no "common cause" of these deaths, and so they were attributed to a "special cause", namely murder by the parent. However, as new knowledge is discovered, it can cause a shift between the categories of "common cause" and "special cause". As Lucas explained in his earlier book The Concept of Probability, "coincidence is a dialectical concept".

  10. With due respects to all the above views and comments i would like to share my views about competency and systems

    Perfect system is important as well competency, a perfect system with incompetency would create a perfect incompetency. I would also like a philosophical aspect to the above

    A perfect system with competency, values and spirit is important For eg a thief with a system would become a perfect thief.



  11. Richard:

    As systems thinkers shouldn't each system stand on its own? Studied and empirical evidence gathered with each system. I am sharing the data from organizations I have worked with and what Dr. Deming and others have shared. You have choices: (1) Believe what I have found ( or at least be curious), (2) collect data on your own system (suggested always) or (3) Deny what you have heard and move on . . . and usually keep doing what you have been doing. My aim is to create curiosity and for organizations to "see for themselves" how they perform based on their own empirical evidence. Where I would start anyway with any new customer is get knowledge (understand the what and why of current performance).

    Part of systems thinking is engaging the brain so strange coincidences and wrong conclusions can be lessened by thinking rather than assumptions. Data can help, but only if we have a thinking person (and not some computer spewing data). We won't avoid all missing interpretations, but context with people discerning the data and events is much better than anecdotal world we live in.

    Theory is theory and no theory is ever proven as one event counter will prove otherwise. Scientists are always careful to explain the nuances of their theories. We should be no less diligent.

    Regards, Tripp

  12. Dear Mr.Richard

    It is a very good view point, it happens that people who gains experience some times go by data and experiences of others. But i agree that there should be self evaluation so that we are not misled by these. We dont ascertain the environment and cultures , hence our own thinking is very important. I agree


  13. Tripp

    I mostly agree with your most recent comment. Debate about systems thinking often degenerates into political strife, which I regret as much as anyone, and I can see a lot of common ground in what you are saying, as well as some detailed points of difference we should probably discuss over a Beer.

    Many thanks for a good discussion, R