Deming’s Great Lapse in Logic
I’m going to start out by saying that I fully expect to be blasted for this article. Challenging anything that Deming or Ohno or Shingo or Juran says is walking on thin ice in the Lean community.
Despite that, I am going to do it anyway. Lately, I have been more aware of the numerous references to Deming in the online Lean community. Forums, blogs, and newsletters frequently make a reference to something that Deming said, or a belief that he held.
The reason I have been paying more attention to it is that I am reading Out of the Crisis cover to cover for the first time. I am actually embarrassed to admit that. Deming is one of the most highly regarded thinkers in the Lean world, and I have only read extracts and summaries of his most famous work. For about the last 10 or 12 years I have always been in the middle of a couple of Lean works at any given time, but just never put Deming’s book into the mix. I know a lot about him, and have read chunks of his works and summaries of his biggest ideas, but just never sat back and dove into his book until now.
But before I go further into this discussion, I’d like to ask you this question:
Loading ...Now, despite the title of this article, I agree with nearly all of what Deming says. He’s got a brilliant mind, and was way ahead of his time in his thinking. I just get concerned when I see value being put into who said something, rather than what was said. A few of the things Deming said does not sit well with me.
Specifically, I have recently seen several references to the second part of his 11th point, “Eliminate numerical goals for people in management.” The fact that Deming made this point is used to bolster the argument against using incentives to increase performance.
What I have a problem with in this point is a single paragraph at the end of his discussion.
“The only number that is permissible for a manger to dangle in front of his people is a plain statement of fact with respect to survival. Examples: (1) Unless our sales improve 10 per cent next year, we shall be out of business. (2) The average level of carbon monoxide in an area, over an 8-hour period, must not exceed 8 parts per million. Reason: 9 or more parts per million has been declared injurious to health.” – W. Edwards Deming, from Out of the Crisis
9 Comments
Chris Paulsen · December 22, 2010 at 1:19 pm
Jeff,
I’ve had a hard time buying into this one as well. I’m re-reading a book by Rafael Aguayo who was one of Deming’s students. The book goes through the 14 points and I continue to struggle with the one about goals. What John Hunter states in his comments makes sense though. Some organizations clearly are trying to solve everything today by measuring everything which is ineffective.
Thanks for sharing.
Chris
Jeff Hajek · December 24, 2010 at 5:10 pm
Chris,
That’s the funny thing about metrics. They work because they make people take action to impove on what is being measured.
But sometimes they don’t work because they get people to take action only on what is being measured.
Or, as John said (and you pointed out), because you measure something that is going to happen anyway.
Thanks for the comment.
Jeff
Lawrence M. Miller · December 21, 2010 at 10:26 am
Jeff, I think you right to begin with.
I knew Dr. Deming and have great admiration for the work that he did, as well as the other leaders of the quality movement. However, Dr. Deming strayed into areas about which he had little knowledge and did not know the science (the data, that he always insisted we look at.)
This is particularly true in regard to the use of positive reinforcement or incentives. While he was correct that most piece rate incentives were detrimental to continuous improvement, he mistakenly generalized that to all incentives. There is a lot of data on the use of positive reinforcement, under which conditions it works, how it works, etc. and he did not familiarize himself with this data, but instead relied on “slogans” about how rewards don’t work.
Nevertheless, he made a great contribution.
John Hunter · December 20, 2010 at 6:03 pm
My understanding of what Deming meant by that is better expressed in the phrasing “arbitrary goals” which he used more. http://curiouscat.com/deming/management_by_target.cfm He distinguished facts of life from targets. It isn’t usually that you target breathing enough oxygen or drinking enough water those are just fact of life that are necessary but hardly goals (unless you are in really bad shape).
It is good for people to question Deming, Ohno and everyone else. They also just have to realize that when they question people who general have pretty good ideas you have a better chance for being wrong then if you question Dilbert’s boss.
Jeff Hajek · December 20, 2010 at 7:41 pm
John,
You make a lot of good points in your article, but I don’t think it echoes what Deming said–at least not in the 14 points. Deming did not make his 11th point say to eliminate arbitrary goals for managers. He said eliminate numerical goals for people in management.
In point 11b’s write-up, he makes a brief mention of goals without methods being a problem, but he doesn’t advocate using goals with methods. He says eliminate them. I think if Deming had wanted people to use goals with methods he would have specifically said to do so.
By the way, thanks for the link. I think it helps the discussion to have that point of view added in.
Jeff
Evan Durant · December 20, 2010 at 10:33 am
Hi Jeff,
Interesting and thought-provoking post. My interpretation of this, and reinforced by my own experiences, is that having numerical goals belongs to management alone. Having them is fine, but they ought not be communicated to the people doing the work. Goals and work standards should be no more than an indication that standard work is being followed and is effective. If goals are not met it is an indication that the system has broken, and management should find out what’s wrong with the system.
Too often I’ve seen well-intentioned goal setting for operators drive exactly the wrong activity. If the expectation is set that operators follow standard work and only react to situations where they cannot, rather than to failure to achieve a numerical goal, then I think you get much more consistent quality and happier operators.
So I think goals are an important indicator of system performance and should be used to root out problems in the system. But if it were up to me I would never tell an operator what they were.
Jeff Hajek · December 20, 2010 at 11:04 am
Evan,
But isn’t Standard Work, by definition, a goal? It is work balanced to a takt time. Otherwise, it is just a sequence of operations.
As far as communication of goals, I’ve always found that keeping a team in the loop raises morale and gets them involved. This is a part where I strongly agree with Deming. He says you can’t magically expect people to improve by setting a goal. But I’ve always found that people respond well when they are empowered to make changes to help reach what they see as a reasonable goal they had a voice in establishing. And of course they need to have the training and resources to make good decisions. And they have to trust that they won’t get in trouble when they make the inevitable mistakes.
I’m not talking about daily production goals here. I’m talking about KPI types of goals–reducing stock outs, improving satisfaction with calls in a call center, lowering lead times, etc.
I think the problem is poorly set and managed goals, not that goals are inherently bad.
Always great to hear you thoughts, Evan.
Jeff
Steven Leuschel · December 20, 2010 at 9:37 am
If you’re trying to keep the level of CO below 8 parts per million do you monitor the level of in order to keep it below 8 parts per million in order to achieve the target?
Or-
In order to keep the level of CO below 8 parts per million do you:
a. identify the activities needed to keep the CO levels down
b. establish visual management, PDCA, and roles/responsibilities around those activities
c. check/adjust the action plan (based on the results)
If you have a perfect action plan – do you need to focus or even pay attention to the numbers/results?
Jeff Hajek · December 20, 2010 at 9:44 am
Steven,
Yes, I think you do need to measure the results.
For point ‘A’, how can you be sure the activities are actually bringing down CO levels if you don’t measure?
For point ‘B’, how do you know what a normal condition is for visual management, or what are you checking against in PDCA?
For point ‘C’, how do you adjust without knowing if something worked?
My point is that you can’t only pay attention to results, but you can’t ignore them either.
Thanks for the comment.
Jeff