Computer Chronicles Revisited 6 — Wordvision, Word Plus, and the Writer's Workbench

Since the 1990s, word processing has largely been synonymous with Microsoft Word. Of course, Word didn’t start out on top. It was first released in October 1983. At that time, the dominant word processing program was WordStar, which had already been on the market for several years. The first Chronicles episode to discuss word processors, taped in December 1983, largely framed the discussion in terms of discussing the competitors to WordStar yet made no mention of Microsoft’s new offering.

Would Computer Word Processing Make Language Too Bland and Dry?

Stewart Cheifet and Gary Kildall opened this episode by looking at a large pile of word processor instruction manuals. Cheifet noted there were about 150 word processing programs currently on the market. Cheifet also displayed an example of a handheld text editing device, the TRS-80 Model 100 from Radio Shack. Cheifet asked Kildall how this differed from a word processor. Kildall explained that a text editor was something used to prepare programs. When a user wrote a program for the first time they needed the ability to go back and make changes. But a simple text editor was not good for working with finished documents. That is, they could not be used to cut-and-paste text or prepare a document for printing. That was what a word processor was for. Kildall said word processors came in a “variety of shapes and styles,” which made it difficult for many users to choose a product.

Cheifet then introduced our B-roll feature for the week, which focused on how word processors had changed the world of business writing, specifically newspapers. Cheifet noted that word processors had eliminated many of the “intermediary steps” between a reporter’s notebook and the camera-ready copy. In many instances, a story would not touch paper until the final stages of printing. Thanks to word processing and modems, an article could be transferred over phone lines from a writer’s terminal to the newspaper’s editorial department 30 miles away. Indeed, Cheifet said entire books were now written this way, with files stored on disk until a final draft was “shipped” via modem to the publisher.

But word processors offered more than electronic transmission, Cheifet said. They also allowed for “different degrees of interaction with the creative process,” from correcting spelling errors to offering stylistic advice. As a result, there was disagreement among writers as to the overall effect that word processors would have on the quality of written language. Cheifet said some writers were concerned that computer assistance would promote “dry, bland writing” that diluted individual style. However, other writers believed that word processors would improve their ability by providing as much or as little assistance as needed.

Did Price Matter When It Came to Word Processors?

Paul Schindler and Jim Edlin joined Cheifet and Kildall for the next segment. Schindler was a journalist then working for Information System News. Edlin was the designer of Wordvision, a recently released word processing program designed to work with the IBM Personal Computer and certain compatibles. Kildall opened by asking Schindler to describe what he looked for in a word processing program. Schindler said the most important thing for him was that the program was “easy to use” and “easy to learn,” which he admitted were “popular buzzwords” used to describe many word processors. Essentially, Schindler said he wanted to start with a blank screen and write a document with minimal fuss. And since most writers, including him, were not perfect, the program also had to make it easy to correct errors and move text around, as well as format the final document in a “precise and pleasing way.”

Kildall suggested the user may think of a word processor like a car where you “go to the the dealer and try it out.” Schindler quipped you needed to “try several” before committing to one.

Cheifet pointed out that Edlin’s Wordvision was priced in the $50-$70 range, which was considerably lower than WordStar, which could retail for as much as $500. Cheifet wanted to know how much did word processors differed to cover that price range. Schindler said they tended to differ in function. He noted that the price usually reflected performance. For example, there were some things WordStar could do that Wordvision could not (and vice versa). It basically came down to what you needed as the user. As a reporter, Schindler said he just needed to be able to “slap words on the screen.” He actually used a word processing program developed by a friend. This program had no ability to format documents on the backend. But Schindler did not require such features, as he never printed his stories on paper. He sent them to a computer in New York. That said, if you required formatting and printing capabilities, you needed to find a word processing package with “sophistication” in those areas.

Kildall observed that as more companies entered the mass market of word processing software, prices would come down across the board. Edlin chimed in and said there was also a time factor. For instance, if you bought a $5 calculator from Safeway today, it would likely out-perform a $500 calculator that you purchased a decade ago. Similarly, with IBM putting out 600,000 personal computers–and counting–in a single year, the numbers at which a small software company like his could “amortize its development costs” were coming down. That meant Wordvision could offer more and still charge less.

Kildall said it was therefore important to look at a software package’s functionality and not just its price. Schindler added the key was to not buy more than you needed. The typical writer did not require “some giant 32-bit super microcomputer” just to do simple word processing. Schindler said he’d seen articles comparing all available word processors “along an axis of 42 features.” But if you only needed six of those features, buy the least expensive package that contained those six.

Cheifet then asked Edlin to demonstrate Wordvision on an IBM PC. Edlin clarified that he considered his software a “writing tool” rather than a word processor. By this, he meant Wordvision was something a person would use to work with “his or her own words,” as opposed to someone like a secretary who was primarily taking down someone else’s words. Edlin also said Wordvision was designed for “people literacy,” i.e., it was about making more literate computers that understood people, not the other way around.

Edlin said this philosophy informed the user interface. He pointed to the Wordvision title screen, which displayed menu options using red, yellow, and green symbols to mimic a traffic light–a concept that most users would instinctively understand. Edlin also said Wordvision came with labels (stickers, really) that could be applied to the standard IBM keyboard to guide users on the proper commands. (He said more recent versions came with slip-over keycaps.) Files were also displayed visually so that the user did not need to type in a file name. Edlin then demonstrated the basic process of working with a document.

Do You Want to Buy Spellcheck as a Service?

For the final segment, Warren Kuhl of Western Electric and Wayne Holder joined Cheiefet and Kildall. Holder was the principal at Oasis Systems, which developed a spellchecking package for word processors called The Word Plus. Kildall noted that up to this point, most word processing packages turned the personal computer into a “fancy typewriter,” but Holder’s software went beyond that. Holder said yes, The Word Plus was meant to take over where the writing process left off. When someone was engaged in the creative process of writing, they did not want to stop and check their spelling or look for redundant phrasing. His software took already created files and ran them through a series of spelling and stylistic checks before giving recommendations on how to improve the text.

Holder then demonstrated The Word Plus on a Kaypro 10 computer. (This machine was quite noisy and could be heard whirring throughout the demonstration.) He showed that it took about 15 seconds for the software to scan a text file and flag any spelling problems. The user could then look at each spelling error and consider alternatives from the built-in dictionary. The next step was to perform a “style analysis.” Holder noted that when you’re writing with a style book, there was often a tendency to fall into certain traps with respect to phrasing. The Word Plus could identify such traps and once again present alternatives. For instance, instead of using “SMALL IN SIZE,” the software suggested just saying “SMALL.”

Kildall asked for clarification as to how the program checked for possible alternatives–that is, did it actually rely on a dictionary? Holder said yes, there was an actual dictionary. Along the same lines, the style analysis used a list of phrases compiled from phrasebooks and other sources. In fact, a large part of this work was based on what AT&T had done in developing its Writer’s Workbench, the product Kuhl was there to demonstrate. Cheifet followed up by asking if The Word Plus could be purchased on floppy disk to run on a microcomputer. Holder said yes, it was designed to be a “practical package” that users could purchase and use on most available word processors and microcomputers.

Cheifet then turned to Kuhl and asked him to demonstrate Writer’s Workbench. This demo was actually performed on a terminal tied to an offsite machine running UNIX System V. Essentially, Kuhl said Workbench was a set of tools designed to work on UNIX to help writers. Like The Word Plus, the Workbench included both proofreading and prose analysis. He pointed out that beyond looking for things like spelling or punctuation errors, Workbench could also identify split infinitives. It was also designed to identify sexist language in a document and suggest gender-neutral alternatives–e.g., use “PERSONNEL” instead of “MANPOWER.” The other notable feature of Workbench was its ability to analyze the grade-level comprehension of the entire document by looking at the mix of simple, compound, and complex sentences. For example, the sample document was assessed at a 13th-grade level of education. If this document were meant to be an “instructional text,” Kuhl noted, it should be rewritten to require no more than 10th-grade comprehension level.

Cheifet asked Kuhl where the Workbench was currently being used. Kuhl said it was still being tested at Bell Labs and Western Electric. But it had been deployed at Colorado State University, where it had been used by about 2,500 students as part of their composition courses. Kildall asked if people would eventually become dependent on Workbench, to the point where if it passed its analysis, the writing would be considered good. Kuhl said that was the idea. He also emphasized this type of analysis would likely be used more in fields that required writers to follow a particular style, like journalism, as opposed to creative writing.

Workbench’s Legacy Lives on in Today’s Linux and BSD Tooling

David Silverman offered some additional background on the history of The Writer’s Workbench and its use at Colorado State as part of Unix: An Oral History. Silverman said the project that became Workbench actually grew out of an effort by two men, Bob Morris and Lee McMahon, to determine the authorship of individual contributions to the Federalist Papers based on text processing and statistical analysis. When Workbench was later “beta tested” at Colorado State, it “succeeded for three main reasons – its reliability, structure, and the programmers’ understanding of the writing process.” In particular, Workbench proved more capable than a rival offering from IBM called Epistle, which was “slow and incapable of coping with incorrect student grammar.” Workbench was not a grammar checker either, but it could “illuminate the style of sentences employed,” according to Silverman.

Workbench actually continues in use to this day, at least in spirit. During the 1990s, many of the individual tools that made up Workbench were rewritten for the open source GNU project. You can still download and install these tools on modern-day Linux and BSD distributions.

Wayne Holder’s The Word Plus was similar to Workbench in that it offered a series of smaller routines designed to be used with another word processing program, such as the Kaypro 10’s Perfect Writer. A 1982 manual for The Word Plus, also written by Holder, detailed a number of these utilities, including one to insert “soft hyphens” into words and another that calculated a document’s overall word count.

And while I could not find anything about Western Electric’s Warren Kuhl, Wayne Holder has had a very active post-spellchecker career. Even while his Oasis Systems continued to market Word Plus, Holder founded another software company, FTL Games, in 1982, as part of his Software Heaven, Inc. Holder hired an old high school classmate, Bruce F. Webster, as his lead designer and programmer. The two collaborated on FTL’s first game, Sundog: Frozen Legacy, originally released for the Apple IIe in 1984. After Webster left the company, Holder hired another programmer, Doug Bell, who developed FTL’s most famous game, Dungeon Master, and its sequel, Dungeon Master II: Skullkeep. According to The CRPG Addict, the first Dungeon Master was a “seminal game,” as it was the first “real-time 3D game” released for personal computers.

Holder ran FTL until it shut down in the mid-1990s.

Wordvision Publisher’s Financial Woes Lead to Quick Demise

Perhaps the most interesting story to come out of this Chronicles episode was the short rise–and prompt fall–of Wordvision. Jim Edlin and his business partner, Bruce McGloghlin, started the eponymous Bruce and James Program Publishers in Dublin, Ohio, to self-publish their word processing program. As Stuart Cheifet noted, Wordvision was priced significantly below industry-leader WordStar, which retailed for as little as $299 at the time according to one computer magazine I reviewed from that time period. In contrast, the list price of Wordvision was $79.95.

But Bruce and James actually sold Wordvision for an even lower price originally. Before the official launch, users could sign up to buy a “Pioneer Edition” of Wordvision for just $49.95. This was, for all intents and purposes, an untested beta version. Stephen Manes, in his review of Wordvision for the April 3, 1984, issue of PC Magazine, quipped that Bruce and James “managed to get users to pay for the privilege of testing a product,” and the included product documentation contained an explicit warning that “chances are the program will go ‘flooey’ and require turning the computer off and back on.”

Manes was also not impressed with the final release. He said that although Wordvision “has more than a couple of nifty things going for it,” the software was ultimately “so quirky and idiosyncratic that it may be your one-way ticket to involuntary servitude.” Manes was especially critical of the “mess of stickers” that users were expected to place on their keyboard, as mentioned by Edlin during the Chronicles demo. Altogether, there were 28 keys that required stickers, and it was impossible to operate the software without them, as the manual constantly referred to “keys that you will see nowhere except on your Wordvision keyboard.”

Additionally, Manes noted Wordvision did not play nicely with MS-DOS. The user could not backup their files using the standard COPY command in DOS. Instead, it was necessary to load the file into Wordvision and then save a separate copy. As a result of its “nonconformity” with DOS specifications, Manes said Wordvision would not work with the just-released IBM PCjr and would likely be incompatible with the forthcoming Microsoft Windows.

Bad reviews notwithstanding, Wordvision was likely doomed by the time Edlin’s Chronicles appearance aired in March 1984. Just three months later, David Needle reported in InfoWorld that Bruce and James “has run into serious cash-flow problems.” The two co-founders were the only employees left, and Edlin told Needle that he was contemplating seeking financial support from their own customers to stay afloat. Edlin added that while Wordvision was selling well, the lack of money hampered their efforts to continue marketing the program. As a final indignity, Needle said Bruce and James could only afford to man their customer support line for one hour each day.

Bruce and James did manage to reorganize as a California company in May 1985, but as best I can tell, it never produced another product after Wordvision. A registration card included with the original release mentioned a number of potential WordVision add-on products, including a spellchecker and the ability to import files from other word processing programs, but none of these features were actually released.

Jim Edlin himself rebounded from the demise of Wordvision and had a prolific career in the tech industry. After Bruce and James closed up shop, Edlin served co-founder and co-CEO of The HyperMedia Group, Inc., which produced customer interactive media during the heyday of CD-ROMs. After selling HyperMedia in 1998, Edlin then spent a decade at Tides, a network of nonprofit organizations based in San Francisco. During the 2010s, Edlin worked with a number of tech startups and spent seven years as a senior software engineer with microlender Kiva. According to Edlin’s personal website, he semi-retired in 2018 and now runs a small online jewelry store.

Notes from the Random Access File