Tools are funny things. Ostensibly, a tool is a device or implement used in the facilitation of work. Yet, sometimes a tool becomes more than that. Sometimes, a tool becomes an extension of the self. The perfect hammer becomes an extension of the hand, while the perfect software application becomes an extension of the mind. Tools are extremely personal in this way. While one tool may be perfect for one person, it may be a very poor tool for another. Maybe the handle grip is too small or too large. Maybe the default keybindings or screen layouts are too difficult for a certain user to master. Today’s article concerns a very specific tool that has become legendary.
In 1973, William Millard founded Informations Management Sciences Associates Incorporated, IMSAI, in San Leandro, California. In the early days of the company, IMSAI focused on consulting and software engineering. The most notable customer during this time was the United States federal government.
In 1974, IMSAI was asked to construct a machine that could handle workloads for General Motors car dealerships. This resulted in the IMSAI 8080.
The IMSAI 8080 used an Intel 8080 @ 2 MHz. The machine was housed in an aluminum casing, featured twenty five S100 bus expansion slots, and a twenty amp PSU. Like other computers of the time, the IMSAI shipped with the front panel board and the CPU board. Common expansion cards were memory (from 4K up to 64K), I/O boards (with serial and/or parallel), video display modules, tape controllers, and floppy controllers.
The most popular machine in the market at this point was the Altair 8800, which the IMSAI loosely copied. The advantage that Altair had was MS BASIC. This made the Altair much easier to use. Due to this disadvantage, IMSAI become the first company to license to CP/M and promote the use floppy disks. The ease of and speed of program loading this provided helped to drive sales to around the 20k mark through the course of the machine’s production.
This early version of CP/M didn’t have much refinement. The standard editor on the system was ED, and this was a line editor written in PL/M. One of the software engineers at IMSAI, John Robbins “Rob” Barnaby, then began writing NED (New EDitor), which retained the line editor functions of ED, but also added a visual interface. NED was written in 8080 assembly language.
The man who got that contract with DRI for CP/M was Seymour Ivan Rubinstein. Rubinstein left IMSAI with $8500 in his pocket, and he founded MicroPro International Corporation in 1978. At this point, there wasn’t a killer word processing application for microcomputers. This was a gap that Rubinstein intended to fill. He persuaded Barnaby to leave IMSAI and come to work at his new venture. The first result of this was WordMaster and SuperSort. There is some conflicting information here. I have read that WordMaster and SuperSort were released in 1976, but this predated the founding of MicroPro from most sources I can find. It is possible that MicroPro was a side-gig for Rubinstein and Barnaby before 1978. However, it is notable that in 1977 IMSAI became ComputerLand and exited the computer manufacturing business. As such, it makes sense that both men would have left IMSAI by then.
While these initial efforts did have some success, they also produced feedback. This feedback (coupled with some research Rubinstein had done on word processing technology at IBM, Xerox, and Wang Laboratories) resulted in the start of work on an improved word processing application. This application wasn’t modal like the previous WordMaster and NED. It featured more key bindings which provided refined control. It also included printing capabilities. Printing was a big deal as at this time applications had to include their own printer drivers. This new product was the first commercial, textual, WYSIWG word processor, and it was named WordStar. This first release in 1978 was followed very rapidly by version 2 the same year.
Over the course of 1978 to 1979, MicroPro was moving to a team oriented structure. Barnaby was very much a solo developer and this frustrated him. He left the company in 1979.
In late summer of 1981, MicroPro was growing quickly and they hired a team of 8 programmers in Ireland. Lyle Cowen, who’d written SpellStar, was sent to setup the Ireland office and manage that team. Before settling in for work at their Dun Laoghaire office, the team was carried to San Rafael, California for training and product familiarization. This was important because by this time MicroPro had several products: MailMerge that worked with WordStar for mailing lists and letters, SpellStar that with or without WordStar, CalcStar spreadsheet, DataStar database, SuperSort file sorting utility. All of these applications were written in 8080 assembler and ran on CP/M-80. The company was around 200 or so people, about 50 developers and the rest being sales, marketing, and management. Barnaby had done cross assembler work to port WordStar to various platforms including the Apple ][ prior to his departure, but most of those ports were not to be. The 8086 took precedence, and a port to MS-DOS and CP/M-86 were the focus of the time. The new Irish team worked on the port to CP/M-86 as part of their training in San Rafael. Apple users were forced to continue to use the Z/80 SoftCard and an 80 column card. Osborne users, however, were blessed with a native port.
1982 was a big year. WordStar 3 was released for MS-DOS. This was ported from CP/M-86 on the Intel development system running ISIS II. The build was on 8” floppies, and then transferred to an IBM PC over serial. CalcStar and DataStar were also released in 1982, and MicroPro Ireland were working on internationalization of all of MicroPro’s products.
AT&T approached MicroPro in the early 1980s in an attempt to persuade them to port WordStar to UNIX for a tidy $1 million. Unfortunately, WordStar was written in assembly language for the 8080 and 8086. Porting from the 8080 to the 8086 was aided by automatic source code translators made available from Intel. Additionally, the 8080 code was written for CP/M, and MS-DOS was a CP/M clone. The 6502 port for Apple machines was the exception to this rule, but the 6502 is a rather simple CPU and was handled largely via a cross assembler. A port to a totally different OS on much different hardware would be far more challenging.
By October of 1982, the original CP/M compatible 8080/Z80 assembly code of WordStar had been lost. A programmer at the company, Peter Mierau, was able to recreate that code, and was shortly thereafter laid off. He and two other laid-off MicroPro programmers, Stan Reynolds and Richard Post, created a new company whose intention it was to clone WordStar. Their company was called NewStar. NewStar’s first release, NewWord, was made available in September of 1983. This was nearly feature complete, and it had most of the MailMerge features as well. Their second release the following year not only replicated the functionality of WordStar, but included features WordStar lacked: unerase, built-in spellchecking, and laser printer support.
By 1983, WordStar had become the most widely used word processing program on the market, and the Irish team was working on OEM versions of WordStar for the PC-compatible manufacturers. Lyle Cowen returned to USA and Ed Kirkman was hired as his replacement in Ireland.
By 1984, MicroPro held 23% of the word processor market, and 10% of the overall software market for microcomputers. This made MicroPro the largest software company on Earth at the time. Just two months before the company’s IPO in 1984, Rubinstein suffered a heart attack. Fred Haney visited him in his hospital room, and there convinced him to make all of Rubinstein’s shares of the company non-voting to avoid the IPO being killed. Rubinstein signed and as a consequence Haney became CEO.
Unfortunately, the popularity of the WordStar program didn’t exactly translate into love for the company. From PC Magazine, February 5 1985:
Almost since its birth 4 years ago, MicroPro has had a seemingly unshakable reputation for three things: arrogant indifference to user feedback (MicroPro’s classic reply to questions about WordStar was “call your dealer”); possession of one of the more difficult-to-user word processors on the market; possession the most powerful word processor available.
By this point in our story, major competitors had emerged in the word processor market. Microsoft’s Word was clawing out some market share. WordPerfect, initially released in 1979, was on version 4 and rapidly taking market share away.
One MicroPro employee, Karen Brown, had a friend Edward de Jong working at Reed Risk Insurance. The company had an IBM 370 that wasn’t heavily used over night. De Jong, having heard about WordStar’s success, chose to educate himself by creating a WordStar clone in C on the 370. After six months, he’d cloned the basics. This achievement earned de Jong an introduction to the MicroPro staff, and he was hired as the lead programmer. Over the course of the next year, he worked with twelve programmers to create WordStar 2000 (the Japanese release was called TwinStar). This new version had heavy piracy protection, a non-backward-compatible new file format, different commands, an over-hauled UI, mouse support, and better printer support. The Japanese version of this release included side-by-side Japanese and Latin character sets. For the overall design of WordStar, MicroPro turned to Dan Druid who was trained by the US Navy in Human Factors research. His efforts made WordStar 2000 a better product in many respects. This new version was released in December of 1984. While this was still a DOS program, AT&T finally got their UNIX port. WordStar 2000 was successful, and de Jong hit his royalty cap in just two months.
In 1986, WordPerfect 4.2 was released. This version of WordPerfect overtook WordStar in the market. This same year, MicroPro stopped their WordStar 2000 strategy and purchased NewWord for $3 million and gave the programmers a one year contract each to stay on with the company. The following year, MicroPro went all in on NewWord. Other efforts were completely halted. WordStar 4 was released in 1987. This was NewWord 3 with features from WordStar 3.3 added to it. WordStar 4 was released first for MS-DOS with a release for CP/M coming just a bit later. This was the last release of WordStar for CP/M. This version was successful in the market despite WordStar having lost its position as number one.
WordStar versions 5, 6, and 7 integrated some of the features that were previously unique to WordStar 2000. Notably, WordStar 7 added mouse support, extended memory support, a macro language, and support for Windows 3’s clipboard (even though this was still a DOS program). While these versions did sell, they did little to restore WordStar to its former glory.
In 1989, MicroPro rebranded itself as WordStar International, MicroPro Ireland closed its doors in 1990, and in 1992 the final version of WordStar for MS-DOS was released, version 7.0d.
In 1991, WordStar International released WordStar Legacy for Windows. This was a legal word processor, and it didn’t do well. Later that year, WordStar 1 for Windows was released, and it was quickly followed by 1.1 which was a bug fix release. In June of 1992, WordStar version 1.5 was released. This really should have been 1.0; it added better support for tables and TrueType fonts.
On the 4th of February in 1994, WordStar International merged with Spinnaker and SoftKey Software to form SoftKey International. After this merger, one more release was made. The development wasn’t done internally but through Coyote Software, SoftKey was the publisher. This version, WordStar for Windows 2.0, popped up under various names with PDF documentation as opposed to physical. This was the last major version of WordStar. From Edward de Jong:
The real reason for the demise of the firm is that the management at Micropro refused the proposal to continue evolution of the product into bitmap graphics, as the future was already knocking on the door with the Xerox Star and then the Apple Lisa.
But the real body blow was the loss of Dan Druid, who died of AIDS before it was even recognized as a disease. They foolishly assigned a marketing person as head of R&D and there was no R&D after that.
What is WordStar like? Let’s take a look!
This level of documentation wasn’t too uncommon in the early days of the software industry. It’s something I really miss. Much of my early education in software usage and in software development came from such copious documentation as this from various products.
I've been using WordStar for 40 years (as of next month), and continue to write all my books with it (I'm a Hugo Award-winning science-fiction writer); I explain why I love it so much here:
https://sfwriter.com/wordstar.htm
I use WordStar 7.0D, the final MS-DOS version, under the DOSBox-X emulator (not plain DOSBox, but rather DOSBox-X, which has exellent specific WordStar support: https://dosbox-x.com/
I would like to correct various factual errors in this recounting of the history. First of all, the genius Rob Barnaby had written WordStar entirely in Z80 assembler which was then cross assembled into 6502 Assembler for the early Apple computers.
He was a magician in terms of fitting a very complicated program into tiny RAM. In this case he was trying to fit over 100k lines of code into 64Kbytes of RAM. He seriously abused the overlay manager which would let you swap in various sections of code as needed.
Micropro was moving to a team approach and the solo programmer of Barnaby created conflicts, which culminated in him throwing his terminal out the window of the building.
At this time period, the applications had to carry their own printer drivers, and WordStar had to know the command codes for hundreds of printers. It was a full time position just to test all the printers being made.
Micropro wanted to upgrade their already good and hot selling product into something that would be more user friendly The original Wordstar, if you put in an extra BOLD tag, would flip polarity on all subsequent text, and do some crazy stuff that beginning users didn't find enjoyable. They hired a designer, Dan Druid, who was trained in the US Navy in Human Factors research, and very scientifically designed WordStar 2000.
I had a friend Karen Brown who worked at Micropro and i learned about their massive success, so as an education to myself i cloned WordStar in C on an IBM 370 located in SF at Reed Risk Insurance, an R&D firm that only had a few people on staff, but happened to have a monster 370 machine that was entirely unused at night.
After six months i had cloned the basics of WordStar, and got an introduction to Micropro staff, and because of the potential 1 million dollar offer from AT&T for a Unix version of WordStar, they hired me as lead programmer, and for the next year i worked with 12 programmers at Micropro to develop Wordstar 2000.
WordStar 2000 was a very well received product, and i hit my royalty cap in 2 months....