The Silicon Jungle by David H. Rothman

the data so you needn’t repeat complicated procedures one by one. You might work out these files to simplify your secretary’s work—or he or she might do the same for you. Bowie and Grimes mastered dBASE II basics in eight hours of classes at Clinton Computer. “dBASE II was real easy,” he said. “The manual is in plain English. Even without all our records in the computer now, it saves me and Julie probably five or six hours a week. We were doing it manually before, and we’re on the verge of saving other people in the company lots of time with the loan processing by using dBASE II. I think we’d have to say with five hundred houses in our backlog there is the potential for this to save us $50,000 a year in carrying charges that may have accrued under the old system of keeping loan-processing records. “We are going to buy another machine,” Bowie said, “for the Northern Division.” The potential $50,000 annual savings from the two machines, by the way, would include just reductions in carrying charges—not in executive time or tasks besides loan processing. Consider the economies that would result simply from less paperwork. “When we first set up the computer,” Bowie said, “we set it up exactly like we were doing things manually. As a division manager, I’m in charge of marketing and production. And earlier we had (1) marketing reports, (2) production reports, and (3) reports combining highlights of each for me to determine whether to start houses and things like that. But now all three categories appear on one loan-processing form.” dBASE II (or the new dBASE III) may not work for you. But for Bowie it was a dream program through which he could store and retrieve records quickly and conveniently for any one of many purposes. He might set up his records mainly for loan processing. But the “SALE:PERSN” field, combined with the “SALE:PRICE” one, could tell him which outside sales reps were selling the most expensive homes. In other words, the loan-processing data base was one good way to keep track of the salesmen catering most successfully to the $135,000 buyers. For who cared what Bowie called his data base—“Loan Processing” or “Sales”? The point is, he could follow the salesmen’s performance, too, through cross-references between fields. dBASE II was a treasure trove of information about trends, sales, or otherwise. “The problem,” said Bowie, “is that dBASE II would work fine with my budget summaries for my houses, except they require sixty fields and dBASE II is limited to thirty-two. So Sue suggested the Multiplan spreadsheet program. Multiplan is easy to use, even easier than dBASE II. Julie got started with Multiplan with just fifteen minutes of instruction.” Multiplan works on the same idea as the better-known VisiCalc program, which, like WordStar, has sold hundreds and hundreds of thousands of copies. The true origins of these spreadsheets go back to 1978. A Harvard MBA student disliked the tedium of using a calculator to tote up columns and rows of interrelated numbers; it was boring even with a pocket calculator. A change in just one number could throw off dozens of other entries, so imagine the brain-numbing effect of making an error and then having to recalculate an entire spreadsheet. Why not write a computer program to alter all the other variables if one changed? And so was born the electronic spreadsheet; it and word processing are the single most popular uses of microcomputers—the real justifications for their existence. A VisiCalc-type spreadsheet can add, subtract, multiply, average, do partial sums, find minimums, maximums, simplify your life in a number of ways. A description of VisiCalc in _CPA Micro Report_ ticks off an awesome number of applications: “sales forecasts, profit and loss statements, rate-of-return calculations, project scheduling, tax calculation, pricing strategies, financial planning, loan amortization, league standings and report generation.” An electronic spreadsheet can help do your checkbook; or it can assist in the preparation of a small country’s budget—which, in fact, has happened. Some even say that spreadsheets are contributing to the paperwork deluging American business. VisiCalc coauthor Dan Bricklin disagrees. “A lot of calculations,” he said of the pre-VisiCalc days, “were being done on the backs of envelopes or corners of envelopes or the corners of newspapers. VisiCalc isn’t causing people to produce more numbers and reports. Those numbers were always there, but they weren’t always being identified.”[28] Footnote 28: The Bricklin quote comes from Steve Ditlea’s excellent article in _Popular Computing_, September 1982, page 48, which helped me appreciate VisiCalc’s many uses. “Multiplan is incredible,” Bowie said of his VisiCalc-style software. “I have generated budgets to see if they’ll do all I want them to, to consider all the what ifs. “And it’s a big help in scheduling production. We have a factory producing cabinets for our homes, and our scheduling system is critical, since at most it can produce cabinets for only fifteen units a week.” The cabinet plant manager, caught between the demands of Washington Homes’ northern and southern divisions, had scheduled production according to his own whims. “It was costing us a lot of money in missed settlements,” said Bowie, “and cabinets were not delivered, or they were the wrong color and size.” He was talking in late February, telling how, the other day, he had loaded in thirty-five more cabinet orders and instantly learned “we were out to April 15 on cabinet deliveries. That’s the soonest we could get them based on the plant’s production capabilities. I’m meeting with the manager this morning to see if in the future he can up his output. I’ve found that everybody’s got to live with lead times. Everybody’s got to give us the right lead time on orders, which means we are now ordering cabinets for delivery at the end of April and May, where we generally in the past would not have ordered May delivery until April. If I miss five house deliveries in a month because of cabinets not being there on time, then that’s ,000 to $5,000 in carrying charges over the next three to five months.” Similarly, Multiplan was probably saving Bowie several thousand dollars more over the same period by “identifying houses not started because building permits have not been issued. And it will help identify the reasons why they haven’t been. “To get a building permit, all sorts of things must fall into place. And now we can look at any particular job at any particular time, and if we see it isn’t started, find out why the permit hasn’t been issued. “We can determine if we’re waiting for site plans or a plumbing permit or electrical permit or whatever it happened to be and instantly target and solve that problem. “Before, the record keeping was ‘Go get a permit. Have you got it yet?’ There were no details, no backup. We relied on our field people to get the permits, and they would get the permits in a way that was timely for them and their production schemes but not in a way that was timely for us in bottom-line deliveries. So we are now able to target all the permit process and make it happen at our rate.” Not only was Bowie using Multiplan as a spreadsheet to plug in all the what ifs; he was also using it as a data base, even if it wasn’t as nimble in manipulating nonnumeric data as dBASE II could be. Whatever chore Bowie was using Multiplan for, he loved the “Help” screens. He could turn them on to guide himself through the commands he wasn’t familiar with; besides, all of Multiplan’s basic commands were normally at the bottom of the screen, anyway. Multiplan, by the time you’re reading this, may not be the best spreadsheet on the market, at least not for you. But at one point _CPA Micro Report_ was pronouncing it the “new Empress of Spreadsheets” for accountants using a wide range of computers. Multiplan even worked with VisiCalc files so that users of the older program could easily convert. “Multiplan,” said _Micro Report_, “... can sort a line-by-line record of events by account number or name—a frequent requirement in CPA applications.” Keepers of expense accounts, presumably, could cherish such a capability. Also, Multiplan, as the newsletter pointed out, lets you name your variables; you could refer more easily to the =cells= or the exact locations in the columns and rows. Instead of saying cells “A26” or “R38P,” you could refer to “Sales” and “Total Costs.” With programs like Multiplan, software designers are more successfully catering to the needs of businesspeople who want computers to adjust to them rather than the other way around. That’s how it should be. Bowie is a construction executive, not a computer expert, and Boland’s an accountant rather than a hacker. Bowie, however, plainly seemed more willing to live with the complexities of existing software. Many in his place would have used a consultant—and wisely, I think, for Washington Homes was a multimillion-dollar operation—yet Bowie had the background and patience to computerize just with guidance from a store. He might not have been a computer expert. But he loved the new. He loved complexity if he could logically unravel it. He didn’t mind mistakes. He felt in control because he had backups on disk and on paper. The manuals didn’t scare him. Instead of soaking up every word there, Bowie, like Seymour Rubinstein, had a gift for knowing which page to flip to if he had trouble. Bowie was a born micro user. However serious about his job, he might as well have been a child relaxing after school with a few rounds of Pac-Man. He loved seeing instant cause-and-effect relationships. He took as much delight in learning where his division could be six months hence as a child might take in winning an arcade game. He considered his computer “the greatest therapy in the world,” an opportunity to “sit down and feel really good” in “a fairly high tension business.” Edward Boland, too, however, in a different, more structured way, was curious about numbers and life, and in the end, I suspect, the two men’s learning styles didn’t entirely explain their opinions of their programs. Bolands’ general-ledger software just wasn’t right for his needs. It straitjacketed him. Bowie’s programs, on the other hand, helped him do just about anything he wanted. “Anything,” incidentally, included advancing his career. When I next caught up with him, in May 1984, he was president of another construction firm and owned one-third of that company and was taking home a paycheck thirty-five percent bigger. Bowie said his computer skills “had a great deal to do with it. I had management tools that not very many other people had.” Alan Scharf: Integrated Program, Including Graphics Alan Scharf, a forty-three-year-old New York executive, also has a nice touch with software—a good-enough one, in fact, to have helped win him a job at a blue-chip firm at triple his old salary. “It’s done wonders for my earning power,” he said from his offices at Merrill Lynch Leasing, Inc., where he was a $75,000-a-year assistant vice-president. “I got this job because I walked in and told them I could do a better job on an Apple. “I didn’t own one at home at the time. But you can be sure that I bought one promptly and boned up on it for the next three weeks, and of course I’d done a lot of research on the Apple before then to make sure I could deliver on my promise. “My previous company had refused to let me get one to improve operations there and do estate taxes. I had to do them by hand on a calculator. It took hours per client. And I got mad. Most people my age are afraid of computers, but I’d worked with word processors. And what are word processors but another kind of computer?” So Scharf left his job as an estate tax planner with a staid old brokerage firm and set up shop at Merrill Lynch’s division dealing with real estate and equipment leasing. It was a VisiCalc devotee’s dream job, one calling for quick, repetitive, accurate math in deals as big as $150 million. Merrill Lynch Leasing made bids to companies hungry for better cash flow. The leasing company (and rivals) offered to buy their headquarters buildings or other real estate, freeing the money for bigger factories or research and development. It was a series of leaseback arrangements. Merrill Lynch organized syndicates for the ultimate buyers—people or companies eager for tax shelters. And that meant more than a little numbers crunching. Imagine the variables. The deals had to be sexy enough to the selling companies for Merrill Lynch to win the bids. At the same time, the tax shelters couldn’t leak. The deals must provide the buyers with the write-offs that the prospectuses from the leasing company promised. Ideally, too, they would yield maximum tax advantages on minimum investments. And for investments of different sizes and at different tax rates, just what would the various benefits be? When Scharf reported for work, he found that the real estate department of the leasing company was on the verge of spending $200,000 a year tapping into an outside firm’s computer to come up with the right numbers. The big machine would have been able to do simple debt-amortization calculations. Scharf could have told a company, for instance, how long it would take to pay off a mortgage on a building for which Merrill Lynch proposed a leaseback. But that was only a small part of what the job needed. And what about the costs? So Scharf instead used an Apple system costing less than $7,000, a one-time investment. The Apple couldn’t do all the calculations needed, but it could actually outperform the time-sharing system in some ways. Consider simultaneous equations. The software on big machines—at least by way of the terminals at Merrill Lynch Leasing—just didn’t include them. But the Apple could simulate this capability. With the VisiCalc spreadsheet it could juggle around dozens of interrelated statistics, using nightmarishly elaborate algebra with Catch-22-like mathematical spirals. In other words, you wanted to know the value of _x_, and it depended on the value of _y_ and _z_, and you couldn’t solve for _y_ until you solved _z_, and you couldn’t solve for _z_ until you knew _x_. That’s how it worked, except, quite possibly, Scharf and his staff would be wrestling with, say, _a_ through _k_ instead of just _x_ through _z_. Struggling with these Catch-22s, the Apple was a slowpoke by computer standards. It still took half an hour. That might seem like the Indy 500 to someone accustomed to hand calculations. But Scharf must have felt the same way I did about inferior word-processing software. However faster than without a computer, it still limited your possibilities. You didn’t have as much time to experiment with all your choices. And the more time Scharf had, the more closely he could consider all the variables and the more attractive could be Merrill Lynch’s leaseback bids. The Apple did its job. “We used it to compete successfully for work with a number of well-known clients,” Scharf said. “Anheuser Busch—we did their office building in St. Louis. We worked with Beneficial Corporation. We’ve done a number of K mart stores.” Scharf, never smug, still tinkered with the Apple and its software. An observant computer dealer noticed he would keep asking for larger RAM boards to allow him to do bigger, fancier spreadsheets. And so it was that the dealer nominated Scharf a tester for Lotus 1-2-3 in late summer 1982. 1-2-3 was the new =integrated software= from Lotus Development Corporation, a Massachusetts firm started by a former rock disk jockey rich with $500,000 in royalties from programs sold to the makers of VisiCalc. 1-2-3 combined a spreadsheet, graphics, and data base. You could, for instance, pump figures from the spreadsheet program directly into the data base with a few simple keystrokes. You didn’t have to go through unwieldy computer rigamarole to transfer facts from one kind of electronic file to another. More important, however, Lotus, at least for Scharf’s use, was a more powerful numbers cruncher than the VisiCalc he ran on his Apple. Lotus was for the 16-bit IBM PC. Sixteen-bit machines were speed demons for numbers crunchers, especially with powerful programs like 1-2-3. An Apple-VisiCalc duo handled worksheets with 254 rows and about 65 columns. But an IBM and 1-2-3 duo could take on 2,048 rows and 256 columns. Scharf’s first test version of 1-2-3 cracked simultaneous equations in four minutes, one-tenth the time that the Apple-VisiCalc combination took. Income and cash-flow statements came out calculated to the nearest penny. ■ ■ ■ Alan Scharf’s Tips on Choosing the Right Spreadsheet Not every spreadsheet user has needs as complex as those of Alan Scharf, a whiz with Lotus 1-2-3 and Symphony, but here are traits he says you might look for: