Software Successors

PERSONAL COMPUTERS have entered their rococo age. Much is rewrought and embellished; little is truly new. In the past few months several of the most popular computer programs have come out in revamped versions. Many of the revisions make the programs more attractive. A few raise basic questions about how computers can and cannot be useful. In this article I will be discussing changes in word-processing and database-management programs.

The most touching new product is WordStar 2000, successor to the famous word-processing program WordStar. For years the MicroPro Corporation, the creator of WordStar, found itself in a position analogous to that of the United States during the “Ugly American” era. It bestrode the world, but it felt itself inadequately loved. Everyone seemed to have a complaint about WordStar, even as it dominated the market and set the industry’s standard.

In an attempt to stop all the whining, or out of a genuine desire to please, MicroPro has offered up WordStar 2000. Was the original WordStar considered intimidating and hard to learn? WordStar 2000 comes with an entertaining and informative tutorial program, complete with a Tinkerbell-like animated character who leads you from lesson to lesson. Were WordStar’s commands arbitrary and obscure? The new commands are streamlined and mnemonic — to start editing, you type E, instead of the old WordStar’s D. Were more and more users calling for split screens, which enable you to view and work on two documents at once? The new program has them. Did WordStar lack an “undo” command, with which you could retrieve material you had accidentally erased? WordStar 2000 has that, too.

Considered strictly on the basis of its specifications sheet, then, WordStar 2000 would seem to represent an enormous step forward from WordStar, perhaps even justifying its $495 list price. (Like all the programs mentioned here, WordStar 2000 is usually available at half price from mail-order firms.) Its manuals are excellent; it can handle footnotes and prepare indexes and do math; it is designed to make the most of the “subdirectories” and “pathnames” that add such convenience to new versions of the operating system called MS-DOS. But taken as a whole, the program is a catastrophe.

WordStar 2000’s problem is that it operates with such paralytic slowness as to convince the user that either he or his computer has died. When I first tried the program, on a friend’s IBM Personal Computer, I thought there must be something wrong with his machine. Surely MicroPro would never try to sell a product so much more sluggish than its old stalwart. Later, after I had tried the program on other computers and faced the horrifying truth, I got out my stopwatch to determine just how slow it was. When I was using the old WordStar and working on a document perhaps the length of this article, it took about three seconds for the document to be loaded from a disk drive and appear on the screen. With WordStar 2000, it took about thirty seconds. To “read” one document into the middle of another took four seconds with the original WordStar and almost forty seconds with the 2000. Searching for a certain phrase or character took about eight times as long with the new program. Moving a paragraph from the end of the document to the beginning took about six times as long. To be fair, WordStar 2000’s automatic-reformatting feature is easier than WordStar when you change margins or spacing.

I know: complaints about a computer’s speed always sound petty. Still, I cannot imagine that anyone who is accustomed to the original WordStar will find that footnotes or split screens justify the living-dead performance of WordStar 2000.

In its program’s defense MicroPro has been claiming that the new program was never intended as a replacement for WordStar, which has been rechristened WordStar Professional and which will continue to be improved and sold. It is, instead, a complement, aimed at a different market. MicroPro representatives have also argued that speed can mean many different things. If you consider how long it would take someone who has never before used a computer to compose his very first letter or memo, WordStar 2000 probably is faster than WordStar. For an office full of people who use a computer once a month, a sluggish-butsimple system of this sort might make sense. But for everyone else, choosing an “easy to learn” program is like choosing a bike with welded-on training wheels. For a week you are happy; for the next ten years you grit your teeth.

Two OTHER REVISED word-processing programs offer far more attractive combinations of features and speed. The one I began using about six months ago is version 4.0 of WordPerfect, which also has a list price of $495. (The name is characteristic of an industry with no gift for understatement.) WordPerfect’s approach is the opposite of WordStar’s in several ways. While both the old and new versions of WordStar rely heavily on “control sequences”—hitting the control key along with a letter—most of WordPerfect’s commands are triggered by the “function keys,” the bank of extra keys at the top or the side of the keyboard. WordPerfect is based on “menus” —you make choices from an array of options on the screen—while WordStar makes it easy to bypass the menus and also uses more non-menu commands. The WordPerfect screen display is not cluttered with the ruler lines and unasked-for information most programs force you to look at, and it gives a very precise idea of what the printed version will look like. Underlined words are underlined, boldface is bold. The printing controls are more flexible and powerful than those of any other word-processor I have seen. You can instruct the program to print several documents in sequence, and meanwhile you can be working on another. The footnoting commands, difficult to describe but very easy to use, are a tremendous benefit to anyone facing the burden of annotation. I switched to WordPerfect from the estimable WordStar when working on a project that required numerous footnotes.

The program’s greatest asset may be the attitude of its parent company, Satellite Software International, of Orem, Utah. In the past two years SSI has repeatedly revised WordPerfect, each time rectifying problems that had been pointed out by users and in magazine reviews. When I first received the program, it had a defect that caused infrequent but annoying crashes, in which I would lose whatever material I had been working on. Even though this bug affected only users of the Victor 9000 computer, who hardly constitute the majority of the computing public, SSI put a technician to work, fixed the problem, and mailed out revised disks free.

WordPerfect has a small drawback and a larger one. The minor problem is its spelling checker, which works quite slowly and is unduly permissive. As far as it is concerned, ar, fallot, and ostalgia are all real words. The dictionary is advertised as containing more than 100,000 words, but since it is only about 240,000 characters long, it obviously cannot store all those words in their entirety. It relies instead on formulas for manipulating roots and suffixes, which lead usually to sensible results but sometimes to grief. The more important problem is one of speed. In most of its operations—loading, searching, saving—WordPerfect is about as fast as the original WordStar, which is to say, perfectly acceptable for most purposes. But because the program works so hard to make the screen display resemble the printed output, it hesitates and shivers as the cursor moves from one line to the next, sometimes giving the screen a sluggish, palsied look.

If I were sufficiently annoyed by the WordPerfect cursor, and if I had an IBM or an IBM-compatible machine, I would turn to XyWrite II-PLUS. (The program is pronounced zy-write. That its versions should, like Superbowls, be counted in Roman numerals is another sign of the industry’s lack of restraint.) XyWrite is the fastest word-processor I have ever seen. Virtually everything that happens happens instantaneously. The same paragraph that took ten seconds to move with WordStar, and a minute with WordStar 2000, moved in less than a second with XyWrite. Most of the program’s features are roughly the same as, or slightly less advanced than, WordPerfect’s. Its footnoting commands are less flexible. It has no spelling checker (although its files are compatible with most independent spelling programs, including the estimable The Word Plus), and its printer controls are slower and clumsier than either WordPerfect’s or the original WordStar’s. However, it offers split screens, and it lets you escape to the operating system and run other programs without actually leaving XyWrite, an obscuresounding feature that in practice is very handy. At a list price of $295, it is substantially cheaper than the competition.

DATA-BASE-management programs have evolved in similar ways. Why should normal people care about database managers? Most of them shouldn’t. The programs cost at least as much as good word-processors and are harder to learn. They are worth the expense and effort only if you must keep track of a sizable body of information and examine it in different ways. One week you may want to list your expenses chronologically, another week by tax category or check number or payee. You may want to arrange a bibliography by author or by title, or a mailing list by ZIP code or by last name. Data-base managers are designed for precisely such purposes, and they are judged on their flexibility and their speed.

The genealogy of data-base managers for personal computers runs back to a program called dBase II. (Didn’t they teach punctuation and capitalization in the Silicon Valley schools?) Like the original WordStar, it was powerful and it set a standard; even more than WordStar, it was regarded with fear, because of its reputation for being so difficult to comprehend. Apart from several serious technical limitations, dBase II had one supremely irritating flaw, which became ever more apparent as competitors emerged: it was terribly slow.

This weakness was ruthlessly exploited two years ago by the R:Base family of programs, which are the XyWrites of the data-base world. My technical friends tell me that in principle one competently written program should sort data about as fast as another. They say that sorting is the kind of irreducible operation that is limited mainly by the speed of the computer chip. Maybe so, but there was no denying how much faster R:Base 4000 was than the competition. The program’s creator, Microrim, ran a series of ads listing comparative sorting times, showing “d Old Way” on one side and “R Way” on the other.

R:Base, then, would have seemed the obvious choice, except for an unfortunate limitation of its own. It was faster and easier to use than dBase, but it was less flexible, because dBase was really a programming language masquerading as a program. By the time you had learned all the “do while” and “accept to” commands, you could design a program that would print subtotals and sort columns whatever way you chose. The R:Base command language, if more clearly related to English, was not quite as complete.

Each program has subsequently evolved, in an attempt to eliminate its weakness. By now there is dBase III, which is more flexible, easier to understand, and much less sluggish than the original. The speed comparisons with the original dBase vary, depending on just what job the programs are doing, but by my reckoning the new dBase is two to three times as fast as the old— though still only about half as fast as R:Base. For anyone who is familiar with the old dBase commands and is willing to put up several hundred dollars for improved speed and power (the program lists for $695), dBase III is a sensible choice.

Meanwhile, Microrim has put out numerous supplements to and revisions of R:Base, which have moved the program in two directions at once. Most of the changes have made the programming language more powerful and useful, shoring up R:Base’s one vulnerable spot. Its latest incarnation, called R:Base 5000, can match dBase III practically feature for feature, while running significantly faster and offering several valuable extras. For example, R:Base provides for something that it calls “automatic rule checking”—if you were entering baseball players’ batting averages, you could tell it to accept no number greater than 1.000 (more realistically, nothing over .400), which would help you catch typing errors in your original. The R:Base manuals are excellent, and the program itself offers online advice when you lose your way. In an appeal for converts Microrim also includes a system for automatically converting data prepared using dBase II, Lotus 1-2-3, and other programs for use with R:Base. (R:Base 5000 lists for $700.)

At the same time, Microrim has tried to make R:Base 5000 “easier” to use, with results reminiscent of WordStar 2000. With a program called CLOUT, the company has attempted to bring “artificial intelligence” to the personal-computer user. Instead of asking for, say, a sales report in the usual way (“Select name, id, deptno, sales from salesfile sorted by deptno, sales for date ge 0901-85 and date It 10-01-85”), you could tell the computer, “Show me each department’s leading salesmen for September.” Depending on how much time you wanted to invest in whimsy, you could teach the computer to respond to “Who were the winners and losers last month?” or “Who’s letting down the team?” or “Who gets a key to the executive washroom?”

This seems like fun, and the program—like WordStar 2000—is convenient the very first time you try it. But once you become familiar with R:Base or any other program, you realize that the “hard” way to ask for information is not really any harder—the “Select name” sequence above could be boiled down, through programming tricks, to something as terse as “Print S”—and it is far more efficient and precise. CLOUT drags R:Base’s speed down to the dBase range, and it adds an element of chaos to data-base work. For reasons too complicated to explain here, you are always in suspense about how CLOUT will choose to display the information you are seeking. This kind of artificial intelligence resembles a pidgin language. To the newcomer it seems simple, but it forces you to work harder to make your meaning clear.

I WOULD LIKE TO end with a word in favor of another, less well known database program. I was initially drawn to KnowledgeMan out of plain curiosity. Micro Data Base Systems, which produces the program, ran peculiar-looking advertisements, which featured a charcoal sketch of a man holding a computer disk, done in the fashion of Social Realist murals from the 1930s. Micro Data gave elements in the program names like UserMan and KeyMan, and claimed that it could do practically anything better than the competition. I wondered how so campy an outfit, based in Indiana, could survive against the smoothies from Cambridge and LA.

The explanation, as I discovered after using the program for my own accounting, is that KnowledgeMan really is more powerful and flexible than any competitive program, R:Base and dBase included. Many of the differences between KnowledgeMan and the others are more significant on the spec sheet than in daily use. For example, KnowledgeMan lets me enter 255 “fields per record”—that is, 255 categories of information about each book I am citing or transaction I am recording—versus dBase III’s mere 128, but I’ve never used more than twenty. Other features, however, are very useful. To me, the most valuable is the ability to draw information from several different “tables” simultaneously, without going through the time-consuming agony of “joining” or “merging” or “intersecting” them. KnowledgeMan also has its own spreadsheet, which is not as fast or pretty as the industry leader, Lotus 1-2-3, but which can automatically draw its entries from your data base and thereby calculate financial totals or your income tax.

KnowledgeMan runs about as fast as dBase III, although in some operations it approaches the R:Base realm. Its worst drawback is an impenetrable instruction manual that could double as an entrance exam for the Carnegie-Mellon Institute. Micro Data, which sells KnowledgeMan for $500, says that prayers for a new manual are about to be answered. By the time you read this, KnowledgeMan/2 is supposed to be on the market, with a brand-new manual, an even more powerful command language, and other features that will make the program easier and more inviting to use.

I just hope the company doesn’t get carried away with making the program “easier.”It should reflect on what happened to WordStar, and beware.