The CHI literature includes few studies of applications such as spreadsheets and databases that are used more in business than by consumers and researchers. However, the spreadsheet inspired many early proponents of user-centered design. Melissa Rodriguez Zynda recounts its emergence and evolution.Jonathan Grudin
If we said that the history of spreadsheets was in many ways inextricably linked with the history of personal computing, we would not be the first [1,2]. Products such as VisiCalc, Lotus 1-2-3, and Multiplan paved the way for computing without coding, bringing the power of computing to non-programmers at a time when few applications did so. Spreadsheets do not teach their users how to code or how to understand programming; rather, they allow the user to be “actively engaged with the spreadsheet table as a problem-solving device throughout the process of model building,” without requiring low-level programming knowledge .
While the tabular nature of spreadsheets resembles that of earlier ledgers and computational tables, computing historian Martin Campbell-Kelly argues that such designs cannot be seen as the progenitor of electronic spreadsheets. Rather, what set the spreadsheet apart from other tables is that they allowed for not only data organization and storage but also near-instant calculation and manipulation. Inspired by “what you see is what you get” (WYSIWYG) word-processing applications, spreadsheets allowed “what if” projections, especially financial ones, to immediately be cast and analyzed without a major commitment of resources and, perhaps most important, without writing code. The success of the spreadsheet encouraged the demand for new applications, productivity tools, and ecosystems, and fostered what became personal computing.
At the beginning of our investigation, we sought to trace how spreadsheets changed over the years, how the product evolved. As we examined screenshots and pored over literature, we realized that electronic spreadsheet applications have remained relatively consistent over the years. Aesthetic changes occurred as graphic capabilities improved and operating systems came and went, yet most functional developments occurred at the periphery as added features for specialized tasks. Spreadsheet adoption has actually benefited from the creation of a standardized product. How has a product that at its core has remained unchanged for decades continued to meet the needs of its users? And if the history of personal computing is so linked with the history of spreadsheets, what can it teach us about computing for non-programmers?
Here, we examine the histories of the most widely adopted spreadsheets: VisiCalc, Lotus 1-2-3, Multiplan/Excel, and the recent Google Spreadsheets.
Although Dan Bricklin and Bob Frankston were the first to commercially develop the electronic spreadsheet, the original concept was initially conceived of by Richard Mattessich, of UC Berkeley at the time, in his paper “Budgeting Models and System Simulation,” published in 1961. Two of Mattessich’s graduate students built a program in Fortran; however, it was not fully realized until nearly two decades later.
In 1978, while attending Harvard’s Sloan School of Business, Dan Bricklin coded the first version of what would become VisiCalc. One year later, Bricklin and fellow MIT alum Bob Frankston would form Software Arts and partner with their publishing arm, Personal Software, to develop, market, and release their “killer application,” as it was described by computing enthusiasts of the time.
Bricklin and Frankston’s spreadsheeta riff on the grids used in accounting worksheetswas inspired by the WYSIWYG interfaces of early word-processing programs . The grid serves as a blank canvas, empty cells into which people craft and model infinite scenarios . Unlike analog calculating tables, VisiCalc allowed for the near-instant calculation and manipulation of data, allowing for “what if” projections that “thoroughly changed how mergers and acquisitions, and budgets, were analyzed” , in part because it harnessed the power of computing without the need to write code.
One of Bricklin and Frankston’s major decisions in the development of VisiCalc was to launch their product first on the Apple II. This was decided for several reasons: the computer’s well-documented architecture, its ease of use as a gaming machine, and Apple’s marketing and distribution plan. Unlike, for example, Radio Shack’s Tandy, Apple’s computers were sold at a wide range of computer stores, so pairing VisiCalc with Apple’s distribution plan helped ensure the product’s wide availability. Spreadsheets offered the power of computing to non-programmers at a time when few other applications did, so they appealed to people who had no previous interest in computers. This prompted many customers to purchase entire systemscomputers, printers, and spreadsheet softwarejust to use spreadsheets.
Because VisiCalc was one of the first commercial products for non-programmers, designing it to be straightforward and easy to understand was of the utmost importance. During development, Frankston wrote the code while Bricklin created a reference card in parallel to the build. Any function that Bricklin could not explain clearly and concisely on the reference card was revised or eliminated. The card was a 10-“page,” accordion-style pamphlet, copies of which can be found all over the Internet, including on Dan Bricklin’s website documenting the history of VisiCalc (http://www.bricklin.com/visicalc.htm).
In 1979, when VisiCalc was first published, patents for software were rarely granted, and Software Arts’ lawyers advised Bricklin and Frankston against pursuing patent protection. Bricklin has described software patents as both a blessing and a curse: On the one hand, they protect your design, but on the other, they discourage the growth of the industry.
Rather than evolving their product, Bricklin and Frankston threw themselves into building more versions of VisiCalc to be compatible across different machines and operating systems. Additionally, Personal Software and Software Arts remained separate corporations. In the face of their unprecedented success, making decisions that satisfied both parties became difficult, so much so that the two corporations became locked in legal battles of one kind or another from 1980 to 1984. This left the door open for competitors to appear with their own versions of the product.
Easy as 1-2-3
Given VisiCalc’s wild success, “clones,” or knockoff applications, quickly appeared on the market. It was not until the 1983 release of Lotus 1-2-3, however, that VisiCalc faced a serious contender. Envisioned by Mitch Kapor, a former developer of VisiCalc “add-ons” VisiPlot and VisiTrend, and written by veteran developer Jonathan Sachs, Lotus 1-2-3 advanced spreadsheet functionality in ways that VisiCalc never achieved.
Central to Lotus’s success was its design for the relatively new arrival, the IBM PC. For one, IBM was a recognized and respected name in the business world, so corporations were willing to invest in IBM computers. Kapor understood the potential of this, so he and Sachs specially tuned Lotus 1-2-3 to run on the IBM PC. Taking advantage of the IBM PC’s architecture, Sachs designed Lotus to write directly into the screen buffer, allowing for faster graphics display.
Lotus not only bet on the right hardware; it also made well-received changes to spreadsheet functionality. The most memorable, perhaps, is its development of macros. In Sachs’s own reflections on the development of 1-2-3, he describes macros as “something that nonprogrammers typically didn’t find intimidating, compared to having to learn a programming language” . Though a milestone in the history of spreadsheets, macros also hindered further development, specifically, changes to the menu. Because the macro language used keyboard shortcuts, developing them required freezing the menu, preventing future changes to it.
Part of the success of Lotus 1-2-3, and indeed one of the only conduits for the innovation of spreadsheets, is that Lotus released one of the first development kits in 1985. Lotus quickly realized that customer-created add-ons addressed actual consumer needs and tailored spreadsheets for the people who used them. By releasing the development kit, Lotus encouraged the growth of niche markets and product development for software that, at its core, remained essentially unchanged.
Lotus too became entrenched in legal battles, though unlike VisiCalc, its disputes were external rather than internal. In 1987, spurred by the threat of yet more clones (keep in mind that Lotus was a clone of VisiCalc), Lotus sued Paperback Software over the “look and feel” of its product, VP Planner. Central to Lotus’s case was that VP Planner used a menu structure identical to its own, as did several other clones of the time. Paperback Software argued that “1-2-3 had such a lock on the market that VP Planner had to be compatible to induce users to switch products” . Computing historian Martin Campbell-Kelly explains:
“This was a controversial decision because it implied that software products would have to have perversely different user interfaces to avoid look-and-feel suits, whereas usability was enhanced by having a consistent user interface across applications. It was analogized that if this principle were applied to automobiles, for example, then safety would be compromised and the rental car industry would be unworkable” .
Lotus won the case against Paperback Software and then shifted its attention to a suit against Borland Software, the makers of the spreadsheet Quattro. Quattro allowed its user to choose between two interfaces: one whose menu structure was entirely distinct from 1-2-3’s and one whose menu structure was identical to that of 1-2-3. Lotus won the first trial against Borland; however, Borland appealed and the appeals court decided in Borland’s favor. The Lotus versus Borland case went all the way to the Supreme Court, which upheld Borland’s appeal in 1996. This ruling was relatively inconsequential for both parties, however, as the legal battles had blinded them to pivotal shifts in the industry.
Microsoft entered the spreadsheet market with Multiplan on CP/M, MS-DOS, and Apple computers. It was, however, for the first several years after its release in 1982, still dominated by 1-2-3. While Lotus was busy developing its integrated suite, Jazz, Microsoft released its newest spreadsheet, Excel, for the Macintosh in 1985 to take advantage of its GUI interface, perhaps the single greatest advance in the history of personal computing. Still, the IBM PC was far more widely adopted, in part due to the acceptance of IBM machines by the business world, and Microsoft was intent on carving out a space in that market. “Excel 2.0 was aimed at high-end machines using a 286 or better processor. Microsoft ... was consciously positioning its more user-friendly software for a time when a standard IBM-compatible PC would support Mac-like interfaces” . That time did indeed come, thanks to improving processor capabilities and the Windows operating system itself. Windows 2.0 was released in 1987 with Excel as complementary software, giving it valuable exposure on a platform for which it was designed.
In another historical investigation of spreadsheets, Campbell-Kelly describes what has been termed Microsoft Economics. It posits that having two products with equal market share is not a stable situation; the scales will inevitably tip in one direction, and that product will have the financial upper hand. “Because the marginal cost of software production is trivial compared with the original development cost, incremental sales generate very high profits ... [that] can then be reinvested into product improvement, advertising, and lower prices” . As the Windows operating system gained market dominance in the 1990s, Excel eclipsed Lotus as the most widely adopted spreadsheet.
Today, Microsoft’s Excel maintains near-saturation of the spreadsheet market. While products such as Open Office and Google Spreadsheets have been introduced, neither can claim the prevalence of Excel, largely due to the success of Windows and the ecosystem of Microsoft applications it supports. Additionally, Campbell-Kelly begins to examine who the users of spreadsheets were and are. He explains that the most prevalent titles for literature on spreadsheets are first manuals for specific applications, followed by titles for business and accounting uses . As the history of spreadsheets teaches us, much of spreadsheet development has centered on creating a standardized product, one that functions, looks, and feels like all other spreadsheets. The individuals and industries that use spreadsheets, particularly business and accounting industries, have much invested in applications that remain consistent, both for software compatibility and ease of use. Jonathan Sachs explains:
“I have been asked why there have been no improvements on the spreadsheets concept. I think that it’s because with Lotus 1-2-3 the technology matured, like the word processor, and it basically does such a large percentage of what people want, that it’s enough as is. When you keep adding to it, you make it worse” .
While spreadsheets have certainly suffered from “feature creep,” many of their developments have taken the form of add-ins, add-ons, and niche templates that allow their users to tailor the applications to suit their needs. This fosters peripheral development while still allowing the core functionality of spreadsheets to remain relatively unchanged. One of the newest incarnations of spreadsheets, Google Spreadsheet, represents the reverse of the feature-creep phenomenon: It was designed to provide only a fraction of the functionality of Excel and other spreadsheets as a convenient supplement rather than as a replacement.
As discussed here, spreadsheets were inspired by WYSIWYG word-processing applications, and they emerged as a “killer app” that paved the way for personal computing. However, the extent to which spreadsheets can be considered a “personal computing” application can be debated. As Campbell-Kelly and others have pointed out, spreadsheets were never intended for casual users; they were intended for task-specific, and, most commonly, professional use [4,6] by non-programmers, who could work out problems themselves rather than relying on expert programmers . This certainly granted non-programmers access to computing, and many purchased machines simply to use spreadsheets.
Once the investment was made and interest piqued, people demanded not programming knowledge but more applications.
The beauty of spreadsheets was that they eliminated the need for low-level programming and allowed their users to focus on domain-specific tasks. The blankness of them lets people be creative in their use . The same dynamic is echoed today in applications such as Wix, a WYSIWYG drag-and-drop website builder, and Sketchup, an image-based 3-D modeling software, that use visual interfaces to allow creative programming without writing code. Perhaps the lesson to be learned is that such applications can remain consistent because they are blank, designed to be adapted anew with each use.
I am especially thankful to Erik Stolterman for his advice and support. Thanks to Wishaya Jeen Piyasirisin for creating the timeline graphic, and to Jonathan Grudin for his helpful suggestions.
This article represents the work of the author alone and does not necessarily represent IBM’s positions, strategies, or opinions.
Melissa Rodriguez Zynda is a recent graduate of Indiana University’s HCI/Design program. She is now a UX researcher and designer for IBM in Austin, Texas. Her primary research interest is the intersection of digital technologies and urban built environments.
©2013 ACM 1072-5220/13/09 $15.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2013 ACM, Inc.