VisiCorp's Visi On was a short-lived but influential graphical user interface-based operating environment program for IBM PC compatible personal computers running early versions of DOS. Although Visi On was never popular (as it had steep minimum system requirements for its day), it was a notable influence on the later development of Microsoft Windows.
In the spring of 1981, Personal Software was cash-flush from the ever-increasing sales of VisiCalc, and the corporate directors sat down and planned out their future directions. Ed Esber introduced the concept of a "family" of products that could be sold together, but from a technical perspective none of their products were similar in anything but name. For instance, to use VisiPlot with VisiCalc data, the numbers to be plotted had to be exported in a "raw" format and then re-imported.
Dan Fylstra led a technical discussion on what sorts of actions the user would need to be able to accomplish in order for their products to be truly integrated. They decided that there were three key concepts. One was universal data exchange, which would be supported by a set of common data structures used in all of their programs. Another was a common interface so users would not have to re-learn the UI as they moved from one program to another. Finally, Fylstra was concerned that the time needed to move from one program to another was too long to be useful – a user needing to quickly look something up in VisiDex would have to save and exit VisiCalc, look up the information, and then quit that and re-launch VisiCalc again. This process had to be made simpler.
In July 1981 Xerox announced the Xerox Star workstation, and by that point it was a well known "secret" that Apple Computer was working on a low-cost version that would later be known as the Lisa. Personal Software's president, Terry Opdendyk, knew of a two-man team in Texas that was working on a simplified GUI, and arranged for Scott Warren and Dennis Abbe to visit Personal Software's headquarters in Sunnyvale, California. They demonstrated a version of Smalltalk running on the TRS-80, a seriously underpowered machine, and the Personal Software people were extremely impressed.
A contract was soon signed, and work on "Quasar" started almost immediately. The name was later changed to Visi On, a play on "vision" that retained their "Visi" naming. A port to the ill-fated Apple III was completed in November, and after that, development work shifted to a DEC VAX which had cross-compilers for a number of machines. In early 1982 Personal Software changed their name to VisiCorp, and was betting much of the future success of the company on Visi On.
Visi On had many features of the modern GUI, even a few that did not become common until years later. It was fully mouse-driven, used a bit-mapped display for both text and graphics, included on-line help, and allowed the user to open a number of programs at once, each in its own window. Visi On did not, however, include a graphical file manager. Visi On also demanded a hard drive in order to implement its virtual memory system used for "fast switching", and at the time hard drives were a fairly rare piece of equipment.
Tom Powers, VisiCorp's new VP of marketing, pushed for the system to be demonstrated at the fall COMDEX show in 1982. Others in the company were worried that the product was not ready for shipping, and that showing it so early would leave potential customers and distributors upset if it wasn't ready soon after. Another concern was that VisiWord was being released at the same show, and there was some worry that it might be lost in the shuffle. Powers nevertheless got his way, claiming that IBM pre-announced products in order to drum up industry "buzz" and create a receptive market for when the product actually shipped.
The demonstrations at COMDEX were a huge success. Many viewers had to be told it was not simply a movie they were watching, and Bill Gates speculated that the PC was in fact simply a terminal for a "real" machine like a VAX. The press rushed to write about the product in glowing terms, and it became one of the most talked-about products in the industry. However this very success led to a number of very serious problems.
In separate June and July 1983 Byte articles, the company mentioned a (late) summer 1983 release. But it appeared to developers within the company that a fourth-quarter release would be the earliest that could be hoped for.
Corporate civil war
|This section does not cite any references or sources. (October 2009)|
While Visi On development continued, VisiCorp as an entity was in the process of self-destruction. Terry Opdendyk, the president hand-picked by the early venture capital investors, had an extremely autocratic management style that led to the departure of many key executives. From late 1981 to the eventual release of Visi On, most of the product management of the company left, notably Mitch Kapor in charge of VisiCalc development, Ed Esber, Roy Folk, Visi On's product marketing manager, among others. The press referred to this as "corporate civil war".
It was Mitch Kapor's departure that would prove most devastating to the company, however. Kapor, developer of VisiPlot and VisiTrend, had been pressing for the development of a greatly improved spreadsheet, but Opdendyk was uninterested. This was during a time when VisiCorp and VisiCalc's developers were at an impasse, and VisiCalc was growing increasingly outdated. When Kapor decided to leave, the other executives pressed for a clause forbidding Kapor to work on an "integrated spreadsheet", but Opdendyk couldn't be bothered, exclaiming Kapor is a spaghetti programmer, denigrating his abilities.
Kapor would go on to release Lotus 1-2-3, which became a major competitor to VisiCalc in 1983. By the end of the year, sales had been cut in half. Combined with the exodus of major portions of the senior executive staff and the ongoing battle with VisiCalc's developers, VisiCorp was soon in serious financial difficulty. All hopes for the company were placed on Visi On.
The October 31, 1983 InfoWorld, in an article titled, "Finally, Visi On is here," flatly stated: "the... publisher is putting the product on computer store shelves... Visi On is scheduled to be available during the last week in October". The November 14, 1983 issue said: "VisiCorp has just released Visi On." However, the July 2, 1984 issue says: "By the time Visi On was actually shipped on December 16, 1983,..." and PC Magazine reported in the February 7, 1984 issue that they still hadn't received the product in its commercially available form.
The basic operating system, known as Visi On Applications Manager, sold for $495, and required a mouse for $250. Three applications were also released: The Visi On Calc spreadsheet for $395, Visi On Graph for $250 and the Visi On Word word processor for $375. The cost for a complete package was thus $1765.
However, the main disadvantage of Visi On was its extremely high system requirements by 1982 standards. It needed 512k of RAM and a hard disk at a time when PCs shipped with 64k-128k and IBM did not yet offer a hard disk with the PC. Third-party drives were however available at the time, typically 5MB units that connected to the floppy controller and were treated by the operating system as an oversized floppy disk (there was no subdirectory support). This brought the total cost of running Visi On to $7500, three-quarters the cost of an Apple Lisa.
The press continued to laud the product, going so far as to claim it represented the end of operating systems. The end-users were less impressed, however, not only due to the high cost of the required hardware, but also the general slowness of the system. In a market where computers were generally used for only one or two tasks, the whole raison d'être for Visi On was seriously diluted.
In January 1984, Apple Computer released the Macintosh with much fanfare. Although the Mac was seriously lacking software, notably a spreadsheet, it was faster, cheaper, included a graphical file manager (the Finder), and simply looked much better. Although it didn't compete directly with Visi On, which was really a "PC product", it nevertheless demonstrated that a GUI could indeed be fast and easy to use, both of which Visi On failed to deliver.
Adding to the release's problems was Bill Gates, who took a page from VisiCorp's book and announced that their own product, Microsoft Windows, would be available in May 1984. This muddied the waters significantly, notably when he further claimed it would have a similar feature set, didn't require a hard disk, and cost only $250. Ironically, Windows was released with an even longer delay than Visi On, only shipping in late 1985, and was lacking the features that forced Visi On to demand a hard drive.
Only eight VisiCorp employees were still developing Visi On when VisiCorp sold the source code to Control Data in mid-1984 to raise cash as it sued Software Arts, while continuing to sell the software itself. Sales were apparently very slow; in February 1985, VisiCorp responded by lowering the price of the basic OS to $99, knowing that anyone purchasing it would also need to buy the applications. These were bundled, all three for $990. This improved the situation somewhat, but sales were still far below projections, and it was certainly not helping the company stave off the problems due to Lotus 1-2-3.
Following declining VisiCalc sales and low revenues from Visi On, in November, 1985 the company merged with Paladin Software. The new company kept the Paladin name.
Official system requirements for Visicalc were: PC DOS 1.10, 512k of RAM, a Color/Graphics Adapter, 2.2MB of disk space, and a mouse. It will work on newer PCs, but requires a compatible mouse and hard disk partition under 15MB as only the FAT12 file system is supported.
Visi On required Mouse Systems-compatible mice; Microsoft-compatible PC mice, which over time became the standard, were introduced later (in May 1983). Visi On used two mouse drivers. First, loaded in text mode, made mouse registers accessible to the embedded driver, which translated coordinates to cursor position. This internal driver, built-in as a subroutine into VISIONXT.EXE, required Mouse Systems PC-Mouse pointing device. It is not compatible with the Microsoft Mouse standard.
Writing Visi On applications required a Unix development environment. Visi On was targeted toward high-end (expensive) PC workstations. Visi On applications were written in a subset of C VisiC, and a third-party could have ported the core software (VisiHost, VisiMachine virtual machine, VISIONXT.EXE in IBM PC DOS version) to Unix, but that never occurred. In 1984, VisiCorp's assets were sold off to Control Data Corporation.
Making working copies of the original floppy disks using modern methods is difficult - they are protected using pre-created bad sectors and other methods of floppy disk identification. However, Nathan Lineback has made Disk Images available. The legal status of these disk images is unknown.
- Lemmons, Phil (June 1983). "A Guided Tour of Visi On". BYTE. p. 256. Retrieved 20 October 2013.
- Woodmansee, George (July 1983). "Visi On's Interface Design". BYTE. p. 166. Retrieved 20 October 2013.
- Finally, Visi On is here, InfoWorld, Oct 31, 1983
- "Integrated software: Windows on the future". InfoWorld. November 14, 1983.
- Can Visicorp come back?, InfoWorld, July 2, 1984
- Langdell, James (February 7, 1984). "Who'll Clean Up with PC Window Software?". PC Magazine.
- Langdell, James (January 24, 1984). "Visi On Comes Into Sight". PC Magazine.
- Webster, Robin (1984-10-02). "Troubled VisiCorp Shifts Visi On Code To CDC for Cash: Sticks to Selling". PC Magazine. p. 41. Retrieved 25 October 2013.
- Mouse and new WP program join Microsoft product lineup, InfoWorld, May 30, 1983
- Lemmons, Phil (December 1983). "Microsoft Windows". BYTE. p. 48. Retrieved 20 October 2013.