Dead End: WordPerfect

Abstract

WordPerfect owned the word processor market of the 1980s. By 1990 the company in Orem, Utah had passed $500 million in sales with more than 4,000 employees, the third-largest PC software vendor in the world, on essentially one product. It still held 85% of the DOS word processor market in 1992, by which time DOS had stopped being the market that mattered. Microsoft shipped Windows 3.0 in May 1990; WordPerfect’s first Windows version arrived eighteen months later and was slow and unstable, and the first good one came a year after that, by which point Word for Windows 2.0 had the market. Novell bought the company for $1.4 billion in 1994 and sold it to Corel twenty months later for about $115 million in stock and cash. WordPerfect is still shipping. Its remaining users are lawyers and government offices that never left.

The DOS Kingdom

Alan Ashton, a computer science professor at Brigham Young University, and Bruce Bastian, a graduate student who had been the university’s marching band director, wrote the program in 1979 for the city of Orem’s Data General minicomputer. They founded Satellite Software International in 1980 to sell it, initially as SSI*WP. The IBM PC version followed in 1982, and the company later renamed itself after the product.

Three things made it dominant.

Reveal Codes. Word processors of the era stored formatting as invisible markers in the text, and when a document misbehaved there was no way to see why. Reveal Codes split the screen and showed the marker stream directly, so a user could find the stray font change and delete it. For anyone producing formatted documents professionally, this was the difference between fixing a problem and retyping a page. Word had no equivalent then and has none now.

The function keys. Every command lived on F1 to F10 with Shift, Alt, and Ctrl, forty combinations covering the whole product, printed on a plastic template that sat over the keys. A trained typist never touched a menu or a mouse, and the muscle memory that produced was the strongest lock-in the product had.

The support line. WordPerfect Corporation ran toll-free telephone support, free, staffed by people who knew the software. In an industry where support was usually a premium-rate number and a queue, this built a reputation the company traded on for a decade.

Version 5.1 for DOS, released November 1989, was the peak: pull-down menus and mouse support added on top of the function keys, tables, print preview. Law firms standardised on it, universities taught it, and by 1992 it held about 85% of DOS word processing.

The Utah anomaly

WordPerfect Corporation ran from Orem, in a heavily Latter-day Saint community with no connection to computing culture. The founders’ values showed in the company: no alcohol at company events, and a support operation whose defining quality was patience with people who did not know what they were doing. That culture produced the loyalty that made WordPerfect dominant. It did not produce the speed the company needed when the platform changed.

The Windows Problem

Windows 3.0 shipped in May 1990. WordPerfect kept its engineering weight on DOS, where the revenue was, and treated Windows as secondary. In 1990 that was a defensible reading: nobody knew yet whether Windows would take.

Windows 3.1, April 1992, settled the question, selling in millions and carrying real applications. By then the company had three specific problems.

The codebase. WordPerfect for DOS owned the machine: direct screen writes in text mode, its own keyboard handling, interrupt hooks. Windows demanded an event loop and required that all drawing and input go through its APIs. There was no port. There was a rewrite.

The printer drivers. WordPerfect had built its own printing system with definitions for practically every printer sold, and it was a genuine asset. Moving to Windows meant handing print output to Windows GDI and to Microsoft’s driver model, throwing away that investment and that control. The company was slow to accept the trade.

Moving targets. Early Windows work was done against the 3.0 API and had to be revised for 3.1.

WordPerfect for Windows 5.1 arrived in November 1991, eighteen months after Windows 3.0. It was slower than Word on the same hardware, crashed, and mixed Windows conventions with WordPerfect DOS conventions so that shortcuts behaved differently depending on which ancestry they came from. Version 5.2, November 1992, was the first mature Windows release, and by then Word for Windows 2.0 had been on the market for a year.

Word for Windows

Microsoft had been building Word for Windows since the mid-1980s, before Windows had a market to speak of. Version 1.0 (1989) was unremarkable. Version 2.0 (1991) was fast, stable, followed Windows conventions because it had been designed for them, and shipped with a toolbar, accurate on-screen formatting, integrated spelling, OLE so an Excel range could live inside a document, and Microsoft’s sales organisation behind it.

The important asymmetry was not technical. Microsoft’s word processor had no DOS installed base to protect, no printer driver empire to defend, and no reason to slow down. WordPerfect had all three, and each one was a reason to wait.

WordPerfect held about a third of the Windows word processor market against Microsoft’s majority. In DOS it still had 85%, and DOS was shrinking under it.

Novell

Novell, which owned NetWare, agreed in March 1994 to buy WordPerfect Corporation for $1.4 billion in stock, closing in June; Ashton and Bastian took roughly $700 million each in Novell shares. Novell also bought Quattro Pro from Borland. The plan was a suite to fight Microsoft Office.

The plan needed the acquired company to keep functioning, and it did not. The people who had built the support reputation left, as did engineers who knew the codebase and the leadership that had run the product. Meanwhile Microsoft Office was bundling three applications that were each competitive and selling them through the OEM channel that put Windows on new machines.

WordPerfect 6.0 for Windows (1993) improved things and 6.1 (1994) was technically competitive. Neither moved the share.

In January 1996 Novell agreed to sell WordPerfect to Corel, the Ottawa company behind CorelDRAW: 9.5 million Corel shares worth about $100 million plus $10.75 million in cash, with licensing commitments on Novell technology that brought the headline value to roughly $180 million over five years. The deal closed that March. Novell had paid $1.4 billion twenty months earlier.

Novell v. Microsoft

Novell’s grievance did not end with the sale. In November 2004 it sued Microsoft over the period when it owned WordPerfect, arguing that Microsoft had used its Windows monopoly to disable competing applications.

The centrepiece was a memo of 3 October 1994, in which Bill Gates wrote about the namespace extension APIs that let an application put its own file-handling into the Windows 95 shell: “I have decided that we should not publish these extensions.” Novell’s developers had built against those interfaces on the understanding they would ship, and had to tear the work out.

The case reached a jury in December 2011, which hung 11 to 1 in Novell’s favour. In July 2012 Judge J. Frederick Motz granted Microsoft judgment as a matter of law, ending it. The appeals failed and the Supreme Court declined the case in 2014. Whatever Microsoft did or did not owe WordPerfect, the courts settled it eighteen years after the product had lost.

Corel and the Long Tail

Corel kept shipping WordPerfect Office against Microsoft Office at a permanent disadvantage in price and distribution, serving the markets that had specific reasons not to move.

American law firms were the strongest of those. They had macro libraries built for legal production, staff whose skill was Reveal Codes, and court formatting requirements that punished surprises. Migration to Word happened slowly through the 2000s, sometimes only when a court’s filing rules forced it, and some federal courts kept accepting .wpd files.

Government offices behaved similarly, for the same reason: the cost of the switch was concentrated and immediate, the benefit diffuse.

Corel has shipped versions steadily since, through WordPerfect Office 2021. The product still has Reveal Codes.

Reveal Codes, still

Word has never shipped an equivalent. “Show formatting” reveals attributes; it does not expose the marker stream so you can delete the one that is wrong. People who produce heavily formatted documents under externally imposed rules still name this as the reason they keep a WordPerfect licence. It is a capability gap that has survived thirty years of Word releases.

Dead End: The Cost of Owning the Old Platform

Eighteen months late with a bad product, thirty months late with a good one, into a market where the competitor had shipped a good product first: that is the whole story, and it was not caused by stupidity. The rewrite was real, the API churn was real, and in 1990 DOS was where the money was.

The mistake was in the arithmetic of the transition. WordPerfect assumed its reputation would carry across, and reputations do not survive a first impression made by a slow, crashing port. Buyers in the new market evaluated the two Windows products side by side with no memory of who had been better under DOS.

Incumbency in the old platform is the handicap: every reason to protect the existing revenue is a reason to be late. Microsoft’s Word team had nothing to protect and could optimise for one thing, being the best Windows word processor, which is precisely what a challenger can do and an incumbent cannot bring itself to do in time.