A System Beyond Their Control

Deming proposed his famous “Red Bed” experiment more than half a century ago. These days, videos and descriptions circulate freely via the web, and there are many books and other publications that describe the experiment. But even for those who are familiar with its lessons, the applicability of the experiment and what it teaches are as striking today as they must have been the first time it was run.

If you aren’t familiar with the Red Bead experiment, there’s a pretty good overview here. Briefly, the Red Bead experiment can be summarized like this…

Workers are asked to “produce” red beads by dipping a dimpled paddle into a large container full of beads. Management has set up “the system” such that the container is filled with a mixture of mostly red beads, but also a small fraction of white beads. Thus, when workers pull out their paddle, they inevitably pull out some white beads along with the red ones. Regardless of how workers try (and if you’ve ever done this experiment live, you’ll know that they do try), their paddles always pick up some white beads. In fact, the red bead experiment is set up such that there is very little that can be done by the worker to influence the results. The point, to paraphrase Deming, is that all workers perform within a system that is beyond their control.

Beyond that fundamental message, there are many, many things to be learned from the Red Bead experiment. Deming, for example, famously tracked the performance of various paddles over time, noting that even paddles that were “the same” regressed to different averages and standard deviations over time. Thus different workers in the same system who are ranked according to the defects they produce are being ranked on random differences attributable to the system, rather than on their own individual performance. This is just one example – Deming and others have taken the basic lessons of the Red Bead experiment in scores of directions to illuminate all sorts of lessons.

In my experience, the most common reaction to seeing, playing, or reading about the Red Bead experiment is this: so what – isn’t that obvious? And it is, of course. The genius of Deming’s set up is that it is completely, blindingly obvious what will happen. The genius is that is strips away all the smoke and mirrors of real life situations and makes the conclusions obvious. But even so, the lessons of Red Bead still haven’t sunk into general consciousness. Even for those of us who study it and ruminate on it, the lessons are easy to forget and hard to implement. This must be the case, because the experiment keeps repeating itself over and over again in real life, and we keep trying to blame the workers for the failings of the system.

Consider the recent foibles of trader Jérôme Kerviel and French bank Société Générale, described here in an account by the New York Times.

SocGen and Kerviel’s story has been smothered n coverage – a $7 billion USD loss will do that – and virtually all of the articles (including the one cited above) describe Kerviel as a “rogue trader”. In fact, a Google search combing the terms “Kerviel” and “rogue trader” turns up no less than 700-800 results.

But was Kerviel’s behavior really “rogue”, as in aberrant, different, or going against the usual behavior at SocGen? To be perfectly honest, I don’t have any sort of informed opinion of the answer to that question. I’m not well versed in the general area, and I had never heard of Société Générale before this story broke. But I do have a hunch. I can tell you that all the accounts and interviews I have read, including comments by other employees, indicate that the far from being rogue, Kerviel’s behavior and practices were encouraged and expected. My reading is that he was a classic manifestation of a system carefully crafted and maintained over time by SocGen. All of which makes the a classic case of the Red Bead experiment.

Handpicked Content:   Understanding Permutations and Combinations

Let me be clear that this hypothesisdid not require and special cleverness on my part. In fact, the New York Times article makes the same point:

While management depicts the 31-year-old Mr. Kerviel as a lone operator who spiraled out of control, interviews with current and former Société Générale employees suggest that he was also the product of an environment where risk taking was embraced, as long as it made money for the bank.

To put it in Red Bead terms, Kerviel was doing nothing more than sticking his paddle into the container and pulling it out. For a long time, he had seen a normal number of white beads come out. One day early this year, he stuck in his paddle like he had been taught to do (heavily rewarded for doing, in fact) and got a few more white beads than normal. Random variation is like that. But for Kerviel on this day, voila, he became an instant pariah. SocGen built the container, added the red and white beads, designed the paddles, and taught Kerviel how to put his in and draw it out. Kerviel what he was expected to do. In December he was up $2 billion. In January he was down $7 billion. Like I said, random variation is like that. So who should be made the pariah?

If you don’t like Red Bead, you can think of it in control chart terms. Standard six sigma control limits mean that normal variation will fall within the control limits 99.99967% of the time, right? Which means that one out of every 300,000 will fall out of the control limits with no attributable cause. Now, are there 300,000 folks like Kerviel out there? Or maybe 3000 who perform strings of trades 100 times in a year? If there is, then sooner or later one of them is going have results that fall outside the limits, just like Kerviel did. If that happens to go in the right direction, they get a huge bonus (like Kerviel probably did in years past). If it goes in the wrong direction, they get to be the subject of an uncomfortable article in the New York Times. Even though it is all normal variation, even though it is all the Red Bead experiment, playing itself out again and again.

Handpicked Content:   Retail Industry - Six Sigma

Now, I certainly don’t mean to absolve Kerviel of guilt. What he did was clearly wrong; it threw up a number of warning flags and violated all sorts of rules. But it can’t be called unexpected in any way. It was a logical output of the system that SocGen built. Punishing Kerviel isn’t going to do a thing about that. Red Bead.

Comments 5

  1. Sue Kozlowski

    Thank you, thank you, thank you for your post.

    Whenever I say in training, "It’s the process, not the people" I always get a big push back. Intuitively, it’s easy to think about outcomes as based on the individuals involved – after all they were the ones who "caused" something to happen, right?!?

    I haven’t thought about the red bead experiment for quite a while. Thanks for bringing it back to the forefront for me.
    –Sue K.

  2. Pamela Coates

    As a healthcare professional in the lab who is frequently blamed for "hold-ups" in patient care, I must reply. Many times we are the scapegoat for the delays caused by other healthcate personnel. Was the delay due to the order not being placed in a timely manner? Was the incorrect order placed? Was there a true emergency elsewhere in the hospital that took priority?
    Is a pneumatic tube system used that sent the medication to the wrong destination? Was it sent and someone from the ER took it out and placed it somewhere without letting the person who needed it know? Sometimes the weakest link is really the person telling you the story.

  3. Andrew Downard


    I think that’s exactly what Deming was getting at. You and those around you are likely judged according to "your" wait times, either implicitly or explicity. That’s just like judging the relative merits of Red Bead workers by the number of white beads they pull.

    I think the problem is that we are seduced by what is *easy* to measure, instead of what *should* be measured to give a true inication of performance. It’s like the old story of the guy looking for the diamond ring he dropped on the street, late at night under a streetlight. The punchline is that he dropped the ring in the middle of a field, but he’s looking for it under the streetlight because the light is better. He’s getting lots of data, but none of it is going to help him with his problem. I see this happening a lot in the world of process improvement – we gravitate towards what is easy to measure, and hope for the best. As with many things, Deming realized this a long time ago and left us with Red Bead to faciliate the conversation.


  4. david

    A "rouge" trader — he dealt with cosmetics? :-)

  5. Andrew Downard

    Hmmm…I spelled it wrong 2 out of 4 times. What sigma level is that?

    Thanks for the correction. I’m so embarrassed! Should be fixed now though.

    Thanks again,


Leave a Reply