OR/MS Today - February 1998

February 1998

Oracle:
The Watchmaker's Parable

By Douglas A. Samuelson

"There! All fixed," the watchmaker beamed. White-haired but still erect and enthusiastic, he reminded many of his customers of their favorite college professors. The cozy shop, reminiscent of a small bookstore from the days before shopping malls and discount chains, added to the impression.

The OR/MS analyst handed the watchmaker his credit card and cheerfully inspected his newly cleaned and adjusted watch. He, too, liked the comfortable feel of the shop and the pleasant company of the watchmaker; but he also knew that the old fellow actually had made watches in Europe as a young man and knew more than just about anyone when it came to fixing them. "Competent and enjoyable to be with, " he thought to himself. "Now there's a contrast to a lot of the people I work for."

"It's great, as usual," the analyst told the watchmaker. "Thanks again."

"You're welcome," the watchmaker replied with a smile. "Can I interest you in anything else today?"

"Not unless you have a way to put more hours in the day," the analyst said wistfully.

"Buy another watch and set it to run slow," the watchmaker suggested with a wry grin.

"But then I'd be late all the time," the analyst protested, chuckling.

The watchmaker's eyes twinkled a bit more. "Use a different watch to keep you on time for the appointments you really care about," he proposed. "In fact, a busy fellow like you might need four or five of them."

The analyst's jaw dropped, as a sudden thought struck him. "Of course you're kidding," he murmured, half to himself, "but I think you've just described what my management is actually doing!"

The watchmaker eyed him curiously but said nothing.

"You see," the analyst continued, "I do studies that support decision-making. I know the higher-ups are often looking for a particular answer — they have their minds mostly made up, for political reasons. But I've never had anyone try to influence me to give the answer they wanted. They know most people would see that as unethical.

"But they often commission several different analysts, mostly outside consultants, to study the same problem," the analyst went on. "And until now I never understood why they did that, or how they decided what to use from all the reports they got. Now I get it: they cover the range of recommendations and then find some reason to pick the one they wanted in the first place!"

"Sounds like politics as usual to me," the watchmaker shrugged. "Or corporate politics, or academic politics. People want support, not advice."

"That's about it," the analyst laughed. "But this goes further. When lobbyists come in on an issue, we expect them to bring in experts to support their position, and we just assume they hired experts who would support their views. But maybe these guys are actually playing dirty, pretending to want objective analysis but then not following it as the analysts — and the public — expected. I hear some of my colleagues complain sometimes about how we need more 'scientific consciences' in our agencies, but it seems those few really good people can't change things much even when we have them. Now I see why."

"If that's so in your organization," the watchmaker pointed out, "then there's a way to tell."

"What's that?" the analyst asked.

"Look at the analysts they keep around," the watchmaker said. "If I really had five watches, when I wanted to be early, I'd always look at the one that ran fast. A watch that sometimes runs fast and sometimes runs slow wouldn't do me any good. So, do they get the best people they can, or do they hire people who aren't so good but are very consistent on some set of issues?"

"I see," the analyst nodded. "They can develop a supply of analysts whose opinions are predictable, then just ask the ones whose expected answer is what they want to hear, so they support their case without having to explain away opinions they didn't like. Sneaky!"

"I think you've got it," the watchmaker smiled approvingly. "Now you know a way to spot a corrupt organization. Another way is, watch for contradictions over time — not just policy changes as people come and go, but giving two different answers out of the same studies. Like telling people they're short of money, then giving all the executives big raises. Or like one decision about a power plant I heard about years ago, where there was a big argument over whether it was needed, and this famous think tank turned out to have done the studies for both sides."

"Or insisting that they're pro-quality, but repeatedly cutting their budget for quality management," the analyst exclaimed, excited by his new insight. "Or always making grand pronouncements about the analytical support they've arranged to get, but never telling anyone how they used the analysis in decision-making."

"Right," the watchmaker agreed. "As you see, the signs are there if you know how to read them."

"You sound as if you've been talking with a lot of policy analysts," the analyst marveled.

"I've seen my share, and heard their frustrations," the watchmaker acknowledged, "but none of this is new. One of my other customers told me a quote from some famous statistician - I don't remember who. He said, 'Politicians use statistics the way a drunk uses a lamppost: for support rather than illumination.' "

AUTHOR'S NOTE: After some digging and asking around, I learned something interesting. Apparently the original author of the quote about the lamppost was Mark Twain, but he said "Most people," not just politicians. Thanks to Tom Jabine for tracking this down.

Table of Contents

OR/MS Today Home Page

Lionheart Publishing, Inc.
506 Roswell Street, Suite 220, Marietta, GA 30060, USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com