The Rubber Band Demo

[From Fred Nickols (2006.08.26.1528)] --

I will be making a presentation in late September to a group of plant managers attending a Plant Managers' Forum. The "plants" in question are coal-fired electricity-generating plants.

The title of my session is "The Plant Manager as Interventionist." Basically, I'm going to introduce them to (1) PCT, with an emphasis on their actions as "interventions" aimed at achieving their goals of controlling certain conditions in and in relation to their plants and (b) what I call "performance architecture" (i.e., three related and connected domains of performance: financial, operational and behavioral). My hope is that they might benefit from looking at their world via these somewhat different lenses.

I plan on using the rubber-band demo as a concrete representation of PCT in action (and we'll talk about some others). Many of these plant managers are engineers so I'm curious as to any pitfalls awaiting me with that kind of audience. I'm also open to suggestions regarding any better way of introducing them to PCT (and themselves as "living control systems").

The audience size is small (20-30). Ninety minutes has been allotted to my session and I hope to save most of the last 30 minutes for a substantive Q&A session. Right now, I don't plan on devoting any more than 10-15 minutes to the rubber band demo.

Anyone care to toss out a caution, an idea, a suggestion?

···

--
Regards,

Fred Nickols
Senior Consultant
Distance Consulting
nickols@att.net
www.nickols.us

"Assistance at A Distance"

[From Bill Powers (2006.08.26.1350 MDT)]

Fred Nickols (2006.08.26.1528) --

I plan on using the rubber-band demo as a concrete representation of PCT in action (and we'll talk about some others). Many of these plant managers are engineers so I'm curious as to any pitfalls awaiting me with that kind of audience. I'm also open to suggestions regarding any better way of introducing them to PCT (and themselves as "living control systems").

Since there are engineers involved, and engineers like models that work, how about asking them to predict what the controller will do when there are two disturbances?

Start with one disturbance. Ask them to predict which way the controller will move when the disturber (experimenter) pulls back six inches. Then add a rotation -- the experimenter swings that end 45 degrees from the target spot.What will the controller do? Use identical rubber bands that are about 6 inches long, so the movements will be big enough to see in the rear of the audience.

I suggest putting the dot on a whiteboard, and labelling up and down as North and South, and right and left as East and West, as on a map.

That should take about five minutes (demos always take longer than you think they will).

Finally, the final exam. Add a third rubber band attached at the original knot between the other two. Start with one experimenter holding his end south of the dot so the controller holds his end an equal distance north of the dot.

Tell the audience that two experimenters will now move their ends: One will first move straight south by some amount, then the other will move straight west by the same distance. The controller is to keep the knot over the dot all the time this is happening. Question: in what direction and how far will the controller move his/her end of the rubber bands?

I strongly advise practicing these demos so you know exactly what will happen in every case.

Answer to final exam: the controller will first move north by the same amount that the first experimenter moves south. Then when the second experimenter moves west by the same distance from the position where the slack is just taken up, the controller will move northeast 1.414 (square root of two) times as far as either experimenter has moved. If you want to get fancy you can have someone with a ruler measuring the distances moved.

The point: if you know that control is involved, you can predict what the response will be to any disturbance of the controlled variable. It looks like stimulus and response, but it isn't.

Best,

Bill P.

Okay; that's a variation I hadn't thought of and I'll give it some thought. I was initially thinking of breaking them into pairs and having them do the demo "in the blind" so to speak; that is, neither would know what the other is up to.

···

--
Regards,

Fred Nickols
Senior Consultant
Distance Consulting
nickols@att.net
www.nickols.us

"Assistance at A Distance"
      
-------------- Original message ----------------------
From: Bill Powers <powers_w@FRONTIER.NET>

[From Bill Powers (2006.08.26.1350 MDT)]

Fred Nickols (2006.08.26.1528) --

>I plan on using the rubber-band demo as a concrete representation of
>PCT in action (and we'll talk about some others). Many of these
>plant managers are engineers so I'm curious as to any pitfalls
>awaiting me with that kind of audience. I'm also open to
>suggestions regarding any better way of introducing them to PCT (and
>themselves as "living control systems").

Since there are engineers involved, and engineers like models that
work, how about asking them to predict what the controller will do
when there are two disturbances?

Start with one disturbance. Ask them to predict which way the
controller will move when the disturber (experimenter) pulls back six
inches. Then add a rotation -- the experimenter swings that end 45
degrees from the target spot.What will the controller do? Use
identical rubber bands that are about 6 inches long, so the movements
will be big enough to see in the rear of the audience.

I suggest putting the dot on a whiteboard, and labelling up and down
as North and South, and right and left as East and West, as on a map.

That should take about five minutes (demos always take longer than
you think they will).

Finally, the final exam. Add a third rubber band attached at the
original knot between the other two. Start with one experimenter
holding his end south of the dot so the controller holds his end an
equal distance north of the dot.

Tell the audience that two experimenters will now move their ends:
One will first move straight south by some amount, then the other
will move straight west by the same distance. The controller is to
keep the knot over the dot all the time this is happening. Question:
in what direction and how far will the controller move his/her end of
the rubber bands?

I strongly advise practicing these demos so you know exactly what
will happen in every case.

Answer to final exam: the controller will first move north by the
same amount that the first experimenter moves south. Then when the
second experimenter moves west by the same distance from the position
where the slack is just taken up, the controller will move northeast
1.414 (square root of two) times as far as either experimenter has
moved. If you want to get fancy you can have someone with a ruler
measuring the distances moved.

The point: if you know that control is involved, you can predict what
the response will be to any disturbance of the controlled variable.
It looks like stimulus and response, but it isn't.

Best,

Bill P.

[From Bill Powers (2006.08.27.0020 MDT)]

Fred Nickols (2006.08.26) --

Okay; that's a variation I hadn't thought of and I'll give it some thought. I was initially thinking of breaking them into pairs and having them do the demo "in the blind" so to speak; that is, neither would know what the other is up to.

Using the rubber band demo that way (pairing up) is very time-consuming. People get so excited about what's going on that it's hard to get a simple message across. If you leave them to discover what is going on you'll get all sorts of theories (He's moving his end opposite to my end). And how do you tell one person in each pair to keep the knot over the dot without the other person hearing?

I think it's best to put on a demo in front of the group, so you can at least whisper to your volunteer what to do (or show instructions on a card), then ask the audience what they see the volunteer doing. One trick Dag and I worked out was to take a largish piece of cardboard with a notch in one edge, and hold it edgewise to the whiteboard so the notch is over the dot and the rubber band goes through it. That way the volunteer can't see what your hand is doing -- and controls anyway. You do that after the first round of guessing about what's going on.

If you can do a brief demo with one volunteer, it's then easier to get the audience, in pairs, to repeat what they saw. I do advise trying this out with a "volunteer" before doing it in front of an audience. Remember to make your disturbances very slow, so controlling is easy, and keep reminding the volunteer to "be as accurate as you can" (without giving away the secret).

Best,

Bill P.