Electronic Design

What's All This Spreadsheet Stuff, Anyhow?

Sometimes people ask for my opinions on spreadsheets. I usually reply that they are perfectly fine on beds. If the people persist in further inquiry, I ask them to show me a definition of the word in a dictionary. Of course, that slows them down a little, because that’s a word too new to be in most dictionaries. But when they try to explain what a spreadsheet is, I say, “Oh, yeah, that’s one of those computer programs that makes mistakes, and lies a lot. We used to use those, but they made too many errors.” Which, alas, is partly the truth.

At one time we had a procedure for computing Return On Investment, using paper and pencil and maybe a calculator. It worked fine. It actually took you 3 or 4 minutes to crank through a whole set of numbers, and if you changed one input, you could modify the rest of the computations in 2 or 3 minutes. Then a new guy showed up one day, and he said he had computerized the whole thing. Just plug in the n umbers, push a key, and out come the results. If you wanted to try a different set of numbers, just plug them in and, Bingo, you get the new results. You could print out a dozen different versions on a stack of paper that weighed only 2 or 3 pounds…

About a year later, one of the managers, a singularly suspicious fellow, asked why the Speadsheet’s outputs did not match with the numbers he had calculated at home at midnight. After a good bit of checking, he found that the Spreadsheet was doing it wrong—some kind of systematic error had been programmed in, and all of the ROIs for the last year had this error, and nobody had ever questioned it. Sigh… So much for people who always trust a computer, because nothing can go wrong..?!? Fortunately, we did not have to go back and correct all of those errors, because those ROI numbers don’t mean anything, anyhow.…

Even last month, a guy showed me some figures on a frequency synthesizer. He said he had optimized the numerators and denominators for the circuit, so as to get the output frequency as close as possible to 60 Hz. I told him, “That’s funny, how come my optimization gives better optimization than yours did, with different numbers? I used a slide rule, and when I checked it, I used a hand-held calculator. We can’t both be wrong. Go back and check your stuff.”

He went back and checked—he told me, “no errors in my work.” I got mad and did it by LONG DIVISON. I told him he had better throw out his computer because I was in no mood to put up with any more lies. He went back and discovered that his spreadsheet was doing some rounding off—obviously a misapplication, because a spreadsheet should not be asked to do scientific or engineering computations if the round-off errors can cause false answers. People shouldn’t use computers when a simple calculator does a better job. But it’s the same as saying that people should not drive a 3400-lb. car down the block just to pick up a loaf of bread. Now, if the calculator is broken (or if the rain is pouring down), the spreadsheet (or the car) may make some sense. But just remember where to bring in a sanity check.

The worst thing about spreadsheets is that they get you out of the habit of thinking. Just as I said back in my column on cost accounting—if you let an over-simplified formula or computer program tell you what to do, and that defines your policy, then what are you going to say when your policy forces you to do something truly stupid? You will then find that you have been doing stupid things for a long time, but not stupid enough to notice. Some people can have a spreadsheet multiply 2.99 X 3.01 and they’ll never complain when the answer pops up “6.” Unless, of course, it apples to their paycheck—and even then they won’t notice it for a year.

I’m supported in my wrath by Michael Schrage, who writes about technology’s impact on our lives for the Los Angeles Times: “The most dangerous, hideously misused and thought-annihilating piece of technology invented in the past 15 years has to be the electronic spreadsheet.”* And he’s barely warmed up.

He details many other aspects of foolishness in people’s trust of spreadsheets. He argues that planners would plan out the future in terms of scenarios—a little story that argues, “what if this happened?” In the past, planning by scenario was used by wise managers to imagine some unpredictable situations.

However, “Companies are more comfortable in the formal sense with simple quantification because it reduces the apparent complexity of the problem,” says Peter Schwartz, president of Global Business Network, an Emeryville-based think tank and consulting firm. “I use the word ‘apparent’ deliberately… Most managers believe that if they get the right spreadsheet, the right formula, the right methodology, all that complexity will disappear.” But, of course, it will not go away.

People who style themselves as “executives” can always pretend that complexity and risk do not exist. But when they do so, they’re just fooling themselves.

Schrage explains, “They’re just increasing the risk and turning themselves into gamblers, not businessmen. Because real businessmen understand risk.” But you won’t understand the risk (not to mention, manage or reduce it) by manipulating it within the framework of a spreadsheet. “You need techniques that let you creatively explore risk and vivisect it,” he says.

Asking a spreadsheet to deliver a best-case scenario is hopeless, because the spreadsheet is inherently capable only of linear thinking. Now, I like operational amplifiers because they can be very linear when you want them to be. I enjoy building and studying linear circuits.

But the business world is not a linear place. You can’t predict the future by extrapolating linearly from the past nor the present. So whenever you get help from a spreadsheet, you’re envisioning a world that has no more imagination than an old history book. Can a history book tell a child how to guess and predict what will happen in the future? Probably not, and neither can a spreadsheet—even if it doesn’t make any numerical errors. That’s because the spreadsheet doesn’t help you understand the significant aspects of the problem.

You don’t develop a feel for what’s important if you let machines do all the work. Spreadsheets, especially. But, I say the same thing about SPICE.… What’s new?

All for now. / Comments invited! / RAP / Robert A. Pease / Engineer

*”Spreadsheets Endanger Corporate Health,” by Michael Schrage, San Jose Mercury News, April 15, 1991, page 4F.

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish