Click for next page ( 29


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 28
Trends in Personal Computer Software Mitchell Kapor* The personal computer industry has, relatively speaking, no history. I've been involved with PCs since 1978, and that makes me an old-timer. In business personal computers became legiti- mate only a few years ago, with Visicalc on the Apple II. The first hardware legitimacy for personal computers in business came even more recently, with the IBM PC. Thus, to forecast what the industry may look like, what the products and their uses will be, is in my mind like looking at a first grader and determining what success in what profession that child is going to have. Several characteristics of current product development and the computer fielc! in general make the future uncertain. First it is important to note that the personal computer software products that have been most successful in the marketplace Visicalc, dBase, Worcistar, ant! 1-2-3 are a few were conceived of, inspirer! by, and clevelopeci through the efforts of single individuals or, at most, teams of two people working in isolation. In other words, those products came about through the heroic efforts of indivicT- ual geniuses. Products developed through a more structured ap- proach, which included a marketing requirements document and a formal development team- that is, products with good methoc3 *Mitchell Kapor is president of Lotus Development Corporation, Cambridge, Mas- sachusetts. 28

OCR for page 28
TRENDS IN PERSONAL COMPUTER SOFTWARE 29 ology have not yet achieved the same degree of success in the marketplace. Two examples are Vision and Lisa; much is made of the huncireds of work-years that went into their clevelopment. This is a significant fact for us at Lotus because we worry about how as an organization we are going to clepend less on the efforts of individual geniuses. It is also a concern from a buyer's stand- point. If I were a ciata processing manager thinking about making a multimillion-dolIar commitment to personal computers in my company, it would be important to clear with software vendors that hacT a track record in the marketplace. At present this is an unsolved problem. To add fuel to the fire, the successful products I mentioned- 1-2-3, dBase, WordStar, and Visicalc-were all originally devel- opecl in assembly language. To my knowledge, the currently re- leased versions still run in that language, which, for a number of reasons, gives much better performance. When you're clearing with limiter! resources performance in terms of speec! and power is crucial, but such software is not very portable or responsive to the change in hardware environments. Successful commercial prod- ucts for personal computers in a higher-level language, such as C or Pascal, have not yet been developer! ant! in my view probably will not be accomplished with today's 16-bit technology. We will have to wait for 32-bit technology and go through our investiga- tion again. In the 16-bit worIcI you can make a better product- better in the sense of user acceptance if you write it in assembly language. Portability and other assets are the trade-offs to make a product people will be happy with. Another peculiarity of these successful products is that they were not developed out of any clear sense of market need. That is almost heresy. In the case of 1-2-3, for example, we looked at what was coming and said, "Aha, 16-bit technology is much better than 8-bit. We can do some important things." We looker! at Visicalc and the other products and said, " It would be great if Visicalc hac! a graphing command in it." That was our market research. This peculiar situation may stem from the fact that no one can be taught how to design a personal computer software product. It is possible to take courses, get degrees, go to seminars, ant! put together a huge bookshelf on how to write a compiler, how to do top-down structure systems design, ant! how to write transaction processing-oriented applications. There is training ant! a body of knowlecige; there are experts anti some well-recognized principles

OCR for page 28
30 MANAGING MICROCOA~UTERS and practices. But none of these things tells you how to write a good productivity application for a personal computer. Writing such an application is an art as well as an exercise in trial and error; software designers are like medieval artisans. There is also a subjective and psychological element in the development of software products that please users. It involves recognizing what will fee] right and what will work for the user. It's a big factor because end users are not programmers ant! are not necessarily comfortable with the technology. Thus, software design is not the structured, orderly, system- atic, rational, and controllable process that one would like to present to potential investors in a company's software technol- ogy. I assume that the training and the development of a design methodology will come in time, but it will be a long process. I also know that more structured market research not only is possible but is already being done. In general, however, success- ful products still come about because someone has the germ of an iclea ant! someone, either the same person or a different person, has the technological capability to begin working on that idea. They go off somewhere for six to nine months ant! come back with a product. That should make anyone nervous. For software com- panies such as Lotus or any of its competitors, long-term pros- pects clepencI on moving to another stage of development and im- plementation and even inspiration. We have a long way to go. Besides the strongly individual and unpredictable nature of software design and development thus far, two other characteris- tics of the industry make forecasting difficult. The first is that software companies do not control their own destinies. Major software products are hardware-driven, which today means IBM- driven. The area of micro-to-mainframe communications is a good example of the kind of problem that affects Lotus and many other software companies. The market need is obviously there. PCs have to be hooked up, and we can see many possible approaches. Local area networks are one. Once you have 5, 10, 50, or 100 PCs handing off floppy disks from one user to the other soon becomes tiresome. If these users were connected in a local area network, they Louis share files and clo electronic mail. In fact, organiza- tions are aireacly asking for these capabilities. The software company, however, understanciably wants to be cautious in implementing something that may have to be thrown out once IBM or another company announces something new.

OCR for page 28
TRENDS M' PERSONAL COMPUTER SOFTWARE 31 Major hardware shifts upset everyone, and long-term prediction is quite difficult at this point. I can't say what we will be cloing three years from now because I honestly clon't know. I can only talk about what I think will happen in the next year. A third characteristic that affects software forecasting is that nobody really unclerstands what to do with a personal computer. We have this wonderful piece of technology for which creative people invent uses. The general public buys the tool because it does something for them on a day-to-day basis. But for our target market of managers and professionals I don't think the applica- tions of personal computer technology will occupy more than 5 to 10 percent of their time. In the area of artificial intelligence, for example, I have given speeches about terrific products that could make managers and professionals more productive. But I don't know what shape such products might actually take or what their impact on ens! users or organizations might be because the crea- tive spark for matching up the technology to market need is sim- ply not there. You can't just hire a consultant to tell you, "Builct this product." I have never seen a successful microcomputer product cleveloped that way and I am quite pessimistic about such efforts. So we are back to our small teams of heroic geniuses sitting in the corner. Even if they have a pipeline to a national corporate- user group, such as we have at Lotus, they simply cannot ask, "What do you want?" The technology is evolving so rapidly, both in hardware and software, that no one can fully articulate what is needed. An organization can only say, "Well, we think we need this, this, and this." But could any organization have said several years ago that what it specifically needed was Visicalc? Abso- lutely not. The idea for Visicalc came out of creator Dan Brick- lancl's head, was realized as a product, and now has several million quite happy electronic spreadsheet users. Visicalc could not have been developed by going out and cloing market research. I expect the industry's long-term future to be very exciting, with innovative and useful applications being creates! and per- sonal computers continuing to have a major impact in organiza- tions. But I am quite uncertain about the nature of that impact and the kincis of developments that may occur beyond the next 12 months. Therefore, if it were my job to figure out what to do with personal computer technology in an organization, I would adopt a fairly cautious long-term approach. In the near term, however, we

OCR for page 28
32 MANAGING MICROCOMPUTERS can say a number of things about what is likely to happen with software. Recently some new buzzwords have emerged that offer some clues. One of them is "integrated software." What this says to me is that the ground rules are shifting and that the market expects IBM PCs or XTs to have available on them some multifunction application. Further, this multifunction application should ad- ciress itself to a generic set of needs of users in organizations. There are five such broac! application areas: spreadsheet, data- base, word processing, graphics, and communications. These needs are not new, but there is a much sharper focus and emphasis on new products coming out in these areas. Previously we had incliviclual products such as Visicalc and Visiplot; tociay we have integrated offerings from Lotus, Visicorp, Apple with its Lisa machine, Context with its MBA package, and about 15 other companies. It is apparent that, at least as a concept, integrated software is a big-win item. If you can cleliver an integrated product that floes several things well at an acceptable speed, that has a common command structure or user interface, and that has some ability to share data, you are removing a lot of the fragmentation that existed previously with individual products. If by learning one set of commands you can access database data in a spreadsheet, make graphs from a spreadsheet, and incorporate pieces of the database into your word processor, you have a much more versa- tile product. There seems to be general agreement that the five applica- tions spreadsheet, database, word processing, graphics, and communication while not exhaustive, answer basic productiv- ity needs. Beyond that, there is a diversity of approaches being taken toward integrated software. No one has yet delivered the perfect product, which, ~ think, would have to perform the five individual applications quite well. In other words, if you give peo- ple a spreadsheet application so they can do their forecasting or their planning it has to be at least as good as stancl-alone spread- sheets. Similarly, a word processing application that will be user! for everything from dashing off a quick memo to doing a complete report or document should not be a substantial compromise below a stand-alone word processor. The inevitability of some compromise is one of the many criti- cisms of integrated software packages, but I believe the geniuses will solve that particular problem, as they have others. They will

OCR for page 28
TRENDS llV PERSONAL COMPUTER SOFTWARE 33 do so because you cannot expect people to accept a product that forces them to give up features, convenience, power, or anything else they are used to. They won't buy it, they won't use it, ant! it will sit on the shelf. Simply meeting application needs, however, is insufficient. There is further agreement, I think, that integrated products as application have to be open-ended! so that they can connect in clif- ferent ways to other applications, to vertical market products, to mainframes, to minicomputers, to voice. But there is no general agreement about the best type of open system. We will continue to see diversity until the market votes with its dollars. I Louis like to focus on the prospects for two broac! approaches to integrated software. The first is the integration of the applica- tions themselves. Recently, what I call the window-mouse type of system, both in the Apple Lisa and the Vision product, received a great deal of attention. Quite a few other products adopted that metaphor. With this system it is possible to create different win- clows or physical regions on a screen, each of which can have a different application in it. As a user I can set up my system so that I have my word processing in one window ant! my database in another. ~ can blow up a single winclow and work with a full-screen spreadsheet, then shrink it back clown. And -I control the entire application using a pointing device such as a mouse. The general enthusiasm for the Winslow approach, particularly with the mouse, appears to be leveling off. There has even been some mouse backlash claiming that perhaps keyboards are not so bad after all, especially for creating a document. For me, the main problem of the winclow approach is that there are different applications in each window. These are like little is- lands connected by bridges over which data move. Let me give a concrete example using Lisacalc, which is a spreadsheet: To do forecasts and graphs, you have to cut the data out of the winclow ant! put it onto a clipboarcI. It then gets pasted into the graphing program and makes a graph. To change a number on the spread- sheet or to get different summary numbers, you must move back to the spreadsheet Winslow, enter the new number, cut the data out and paste it back in again. This turns out to be a laborious ant! cumbersome process. What people really want is to make some new numbers, hit a key, and see a new graph come up on the screen. No one has yet cliscoverec! how to do this in a Winslow system. Each winclow (ap- plication) has its own separate fiats structure sharing a common

OCR for page 28
34 MANAGING MICROCOMPUTERS interface so that things can be moved around. But from the user's point of view, why is this necessary? Why can't the user have a unitary clata structure that can be called up as a document, or a spreadsheet, or a database? Although this might be a much more satisfactory approach for the users, it has yet to be realized. So it is back into the little locked room for the geniuses to design a viable product based on user need. The second approach to integrated software and, I think, the big coup, has to do with the integration of clata rather than the integration of the user interface. Things shouIcI be simple, straightforward, consistent, and intuitive for users. A common user interface certainly goes a long way toward that goal, but un- tiT data is completely transparent to the user, many people will shy away from using the proclucts. This integration of data is involved not only in the personal computer but is also important in the micro-to-mainframe link. We are in the midst of a great clear of very painful progress in that area. Recent developments are certainly making it much easier and more convenient to move data from a mainframe to a personal computer, but we are still at a rather primitive stage in terms of how we actually put the data inside of a PC. The process will not be transparent until the next generation in clevelopment, when- ever that occurs. There is also a Tong way to go in the area of communications. I predict that it will be people in organizations willing to invest and devote more of their own resources and cleverness who will de- velop usable systems. And I expect it will take a least another couple of years to see this happen. I have no doubt that we are going to see good integrates! appli- cations ant! integrated data structure. But I think it is equally important that future software products be as open-ended! as pos- sible and have different qualities or types of open-endedness. In the near future, ~ think it is safe to say that organizations may have their own staffs of application developers. They will want to create custom applications for their own corporate needs or they may wish to create applications in some vertical market and sell them. Organizations may want either to develop an application from scratch or some higher-level language that talks to inte- gratec7 applications so users can get the benefit of the spreacT- sheet, the graphing, and the database regardless of the specific application.

OCR for page 28
TRENDS TV PERSONAL COMPUTER SOFTWARE 35 Beyond development from scratch, which is expensive, time consuming, and hard to control, a productivity application can be made open-encled so that an expert user (not a programmer) within the organization can create a mode} template application. This application can then be used as a standard within the organi- zation. To do this, however, expert users must have the necessary tools. The first, stumbling effort Lotus macle in this direction was to inclu(le a macrolanguage with 1-2-3 that contained the rudiments of a formal programming language, control structures, variables as- signments, and user interface primitives. It is clumsy and anything but elegant, but it is being used in corporations far more than we ever could have predicted. In fact, some of our corporate accounts have clevoted 5 and 10 work-years of development in this area and their results are being distributed to hundreds, perhaps thousands, of users within their own organizations worldwide. A third type of open-endedness allows the indiviclual applica- tion to talk to and work with other off-the-shelf retail applica- tions. ~ predict that in the future usage patterns may well be fo- cused on one or two core standardized products within the organization. At the same time there will be legitimate needs for perhaps as many as two dozen other more specific applications that can also be purchased off the shelf. Therefore, it is critical that a core productivity application be able to be integrated with those other applications. There are a variety of approaches that will be coming out to achieve this. In sum, open-encledness is not a simple thing. There are com- munities of users with different needs, vertical market needs, in- ternal needs, the desire to create salable adcI-on applications to a core product. I think it will take a couple of years to sort out what the most desired aspects of open-endedness are and to determine what will become the accepted standard. Blessing or curse, we live in interesting times. Despite all the uncertainties I am quite optimistic about the value of delivering personal computer software ant! in particular integrated soft- ware. I see an absolute shift in these deliveries toward the corpo- rate marketplace. As this shift is taking place, I think software companies are maturing as entrepreneurs and are learning to un- derstand the real needs of organizations. With a modest amount of patience on both sides, organizations and computer companies are going to achieve their mutual goal of delivering goof! informa- tion processing to the broadest community of users.