Late Night Discussions
Number 4

By Eliyahu M. Goldratt

How to throw out the baby with the bath water - a discussion on automation

"Fashions, it's mainly fashions," I moan. "Sometimes I'm getting the distinct impression that it's almost voodoo. And the resulting waste is mind boggling. I wouldn't be surprised if we're dealing with many zillions down the drain."

"Alex, it's quite apparent that you are in a combat mood tonight, but what on earth are you talking about?"

"Oh, just the general way we conduct business. It's fashions, just fashions I'm telling you. Every few years another gizmo that everybody is religiously following and then nothing. Nobody even mentions it any more, it's as if nothing happened. As if all efforts, time, money were not ever spent. As if all lectures, presentations, conferences had never taken place. I wonder in what graveyard they bury all related publications, it must be a very big one and in a secluded place."

"You certainly made your point," Jonah is amused. "But if you want to discuss it, not just to bitch and moan, you'd better be much more specific. What I'm suggesting is that out of all the fashions that you condemn, please pick one. Preferably the one that bothers you the most and then we can be in a position to logically debate it."

Jonah with his rational approach to everything. Why must he be so dry, so analytical. Doesn't he realize the necessity to pour out your emotions, to just let go? I guess not. Quite annoyed I'm looking desperately for an example to demonstrate this syndrome of overnight fashions. Not that there is a lack of examples but I want to find a juicy one - one that will provide me with a very solid case. As angry and emotional as I am tonight, I'm not careless to the extent that I forget with whom I'm debating.

"Automation," I finally say. "Automation in its most glorified form - the lightless plants. You know, that pie-in-the-sky that everybody was talking about just five years ago. The fully automatic plant (or department). Automated to the extent that you could even put out the lights on the second and third shifts and full production would continue in the dark. Automatic machines, robots, automatic guided vehicles, empire-strikes-again stuff."

"Carry on," Jonah encourages me, but tonight I don't need any encouragement.

"The hysteria was so big that almost every substantial company rushed to build a lightless plant. And let me remind you, each such baby cost many millions of hard dollars. It was the 'in thing', the way that every company must embark on if it desired to stay competitive in the future. Now we are in the future but where are the lightless plants? Buried under the carpet."

"Fine Alex. Cool down. You definitely presented your case." He puffs once or twice on his cigar and through the cloud he says, "do you mind switching chairs?"

Looking at my surprised face he bursts into laughter. "Only metaphorically," he explains.

Since I still don't get it, he carries on, "Alex, fashions do not stem from a vacuum. People are not dumb and certainly they are not irresponsible. Fashions start spreading because they do have merit. In order to understand what really happened I'm suggesting that you switch places with those executives that gambled on automation and present their case. Can you do that?"

"I guess I should at least try," I sigh.

It takes me a minute or two to complete the metamorphosis but at last I'm ready.

'With passion," Jonah encourages me, "give it your all."

"Automation definitely has a great appeal," I slowly start. "Machines don't need rest, their performance doesn't fluctuate with moods, they don't even take pee breaks. In short they are very consistent and reliable. If we can build a plant that does not contain the human factor, we'll certainly be in a much better position."

"You are doing fine," Jonah interrupts. "Now what is needed is to be a little bit less vague. 'We'll certainly be in a much better position', may sound very appropriate for a political election speech but it cannot be a base for an analysis. What will the impact be on Throughput, Inventory and Operating Expenses?"

I knew it was coming, that's Jonah. "As for Operating Expense, it's quite hard to estimate off-hand," I answer. "On the one hand, almost no workers are needed, maybe two or three to load materials on pallets, maybe not even that. But on the other hand, relative to a conventional plant, many more professionals will be needed for maintenance and constant reprogramming the automated machines, robots and vehicles. Likewise the depreciation will be considerably higher. My guess is that it will vary case by case, but for our discussion we can assume that the major benefits are not in that avenue."

"As for material inventory," I confidently continue, "we are going to enjoy considerable gains. Constructed and controlled properly the leadtime through such a plant will be dazzlingly short compared to conventional plants. A situation with almost no Murphy, queue or waiting for a worker, can be expected to shrink lead time - and thus inventory - by as much as ninety percent. But the real gains will come from increased Throughput. Very short quoted lead times, very high reliability, very high quality, these three factors combined, guarantee a powerful competitive edge. Thus, if we have chosen a viable product, the market itself will not be a constraint to increasing sales. The fact that our plant is capable of producing continuously, twenty four hours a day, will enable us to capitalize on that wide-open market." Enthusiastically I summarize, Throughput will be astonishingly higher than for a comparable conventional plant."

"Alex, you are something," Jonah cannot control his laughter and then, more seriously he concludes, "No wonder that so many executives decided to experiment with the concept of lightless plants."

"On paper, it definitely had a lot of merit," I agree.

"You see Alex, the problem with fashions is not the way they start, it's in the way they end. When the initial expectations are not fulfilled we forget the initial analysis of the expected benefits. In most cases, like in this one, the analysis of benefits was correct, the lightless plant is a very beneficial and viable avenue. Unfortunately, rather then analyzing the reasons for the disappointment, we just drop the entire issue and everybody flees to find shelter. To cover the rear parts."

"And all efforts and investments go down the tube," I sadly add.

"Not just the investments, something much more important is lost. A very good avenue to improve is now almost hermetically sealed. Today, only a very courageous president will risk experimenting with lightless plants. That is the biggest loss of all."

"Granted," I concur.

We sit quietly for a while.

"Jonah, what do you thing went wrong?" I finally ask.

"Don't you realize?" Jonah is surprised. 'The Achilles heal was, as always, the simple words that you yourself used: 'constructed and controlled properly'. This word 'properly' is just a cover up for ignorance. Haven't you noticed that the theoretical basis for constructing such a plant is lacking?"

"Frankly, I haven't."

"Fine," Jonah doesn't lose his good mood. "Let's examine the simplest form of a fully automated area, where we do have a tremendous amount of experience, the production lines. In this case, is the theoretical background satisfactory?"

He caught me off guard. "In every university we teach how to build such lines," I answer, knowing too well that this does not prove a thing.

"We do teach it," Jonah is on the war path. "We even gave it a very respectable title - 'line balancing'. A title which actually exposes the underlying fallacy. Alex, you know very well that, to be efficient, lines are not supposed to be capacity balanced."

Jonah is right. More then once we have discussed in depth the concept of a chain. "Let me remind you, the impact of statistical fluctuations coupled with dependent resources," Jonah continues mercilessly.

"Don't bother, I remember," I gloomily say. The fact is that I haven't remembered, or to be more precise, I didn't connect. In a line the resources are strongly dependent on each other. If one resource is down, it doesn't take long before the entire line is down. In such an environment the negative deviations, and not the positive ones, tend to accumulate.

"But Jonah, we have proven that in order to balance the flow, the capacity of a line should not be balanced. Granted. But we haven't discussed how the line should be built."

"And if we haven't discussed it, does it mean that the proper rules can not be easily determined? As a matter of fact, it's quite simple to find the rules, the difficulty is the realization that we didn't bother to find them."

"This is quite amazing," I reply. "We design and build lines as if they should be balanced. Of course, the predicted Throughput is not realized. Then we start to debug the lines, putting all the blame on some technical problems rather than on the fact that the line was ill-designed to start with. Through trial and error, we adjust here, we add there, until at last the line starts to produce the expected Throughput. But look on every functional line in reality, all of them are far from being capacity balanced. Why don't we design them properly to start with?"

"Good question," Jonah joins in. "Now you see why the lightless plants failed. Not like the simple one dimensional lines, in the case of a lightless plant, the various parts have to go through different paths through the machines."

"Moreover," I add my two cents, "the equivalent of the conveyer belt, the automatic guided vehicles, were running all over the place. In an environment that resembles spaghetti, you can look until you're blue in the face and still you will not be able to tell a head from a tail. No, in the lightless plant if it is not designed right to start with, trying to correct things through observation simply doesn't work."

"You see Alex," he concludes, "we should not be mad that fashions do exist. The thing that should irritate us is that lack of analysis caused the spill of the baby with the bath water."

This "Late Night Discussion" is Copyright © 1991 Dr. Eliyahu M. Goldratt