Windows.

This month's column was written by C. Eugene Prescott, CPA, Greenville, N.C. Mr. Prescott is the former chairman of the AICPA Tax Division's Task Force on Automation for the Tax Practice of the '90s.

Windows

When Microsoft announced its initial Windows product at COMDEX on Nov. 10, 1983, most CPAs were not paying attention; and when Windows was finally released two years later, on Nov. 20, 1985, not many CPAs bought it. Even as Windows was winning awards for Best Technical Achievement, Best User Interface and Best Software Product in 1986, most CPAs stayed with MS-DOS. Beginning with the Windows 3.0 release on May 22, 1991 (one year after it was announced as available for immediate release), some of the more technologically experienced CPA firms began serious experimentation with the concept, although the vast majority of CPAs stayed away. With the arrival of Windows 3.1 on Apr. 6, 1992, a slow but steady migration of CPAs to Windows began.

In retrospect, most CPAs were correct in avoiding Windows prior to the 3.1 release. The limitations and bugs of the prior releases were not suitable for the large variety of applications (mostly DOS) that CPAs were routinely using. In addition, the hardware demands that Windows required were considerably higher than most CPAs had available, and upgrading was relatively expensive. As PC computing power continued to double every 18 months and the cost decreased, the hardware limitations subsided even in firms continuing to run DOS over networks such as Novell or LANtastic. So by the time Windows for Workgroups 3.0 (numbered 3.0, although it was the first workgroup release) arrived on Oct. 1, 1992 (and was quickly updated to 3.1 on Oct. 27, 1992), there were considerably more CPAs who already had sufficient hardware to accommodate the platform. Windows for Workgroups (WFW) was a good fit for many small and large firms (even though they had to discover this for themselves). While in general the benefits of Windows had been consistently overstated, WFW had significant benefits that had been somewhat understated, such as an included scheduler and Microsoft Mail, in addition to networking capabilities. A few CPA firms figured out how to implement WFW (with peer-to-peer connections to workstations) on top of Novell servers. Productivity gains were achieved on a level that justified the accompanying headaches of managing the memory constraints inherent in MS-DOS (which had to be loaded prior to these versions of Windows). With the addition of 32-bit disk and file access included with WFW 3.11 (released on Nov. 1, 1993), it was clear that WFW 3.11 was a better Windows product than regular Windows 3.1. So even though Windows NT was announced on May 24, 1993, and shipped, many CPAs opted for WFW 3.11, particularly as clients to servers.

With Windows 95, the most hyped product ever, released on Aug. 24, 1995, all CPA firms are confronted with a dilemma. Those that are still DOS-based (which may be most) are wondering whether they should join the fray; those already at Windows 3.11 or WFW 3.11 are wondering whether they should upgrade. And what is Windows NT, anyway?

Regular Windows 3.1 or 3.11 (shipped Apr. 4, 1994) probably is on the largest number of machines, although WFW 3.11 began outselling regular Windows in April 1994. The user interface and most of the operational attributes of Windows 3.1 and WFW 3.11 are identical. The primary difference between these products is that Windows 3.1 depends on some other networking software for connectivity. Many sites are connected to Novell servers and some smaller sites are connected to LANtastic servers. With WFW, the networking connectivity is built in and relatively easy to implement (easier than most alternative networking concepts). While this peer-type 16-bit connectivity yields many productivity gains, heavy-duty file and application services typical of many CPA-specific software packages (such as individual tax packages) are better suited for 32-bit servers optimized for high-volume input and output requests. Consequently, the majority of networked systems running Windows or WFW are also connected to Novell servers. in addition to the included networking capability, the 32-bit disk and file access option in WFW allows for a significant increase in output for disk-intensive activity even on standalone systems. The gain from this capability is more significant for computers with integrated drive electronics (IDE) type controllers to the hard disk than those with small computer systems interface (SCSI) type controllers. While many believe SCSI controllers are preferable, most desktop computers today have IDE controllers. Although most users still prefer WFW to Windows, both of these products have probably reached the limits of their capabilities. Neither product is likely to be further enhanced or upgraded after the release of Windows 95.

While Microsoft's intended competitive market for Windows NT has been variously identified as UNIX, Novell, OS/2 and none of the foregoing, its debut was rather unglamorous. While frequent crashes and other limitations kept the initial release from gaining any significant market share, the concept was nevertheless promising. There were two versions of this 32-bit product, a server version and a workstation version; the main difference was that NT Advanced Server (NTAS) was optimized for traditional server functions and NT Work Station (NTWS) for traditional desktop functions. From an administrative standpoint, being able to use the technology knowledge required to support desktop computers to also support the server was intriguing. Up to 10 NTWSs could be connected in a peer network (similar to WFW), while a large number of Windows, WFW or NTWS computers could be connected to one or more NTAS computers. The original release of NT did not connect well with a Novell server; with the release of version 3.5, NTS (the Advanced designation was dropped) and NTWS became strong and stable platforms. The current version of both is 3.51; they connect well with Novell services.

The main consideration in planning for an NTWS desktop computer is adequate hardware. Although lesser machines will technically run the software, CPAS would be wise to have at least a 486DX2 66Mhz computer with 16MBs random access memory (RAM) before installing it. The hardware requirements for NTS are also demanding; yet, they are not significantly different from a comparable Novell server. in addition, numerous capabilities are "built in," while on competing systems they have to be acquired separately. For instance, server disk redundancy using redundant array of inexpensive disk (RAID) technology is included. This allows an NT server to have "hot-swappable" disk drives. That is, all data is updated simultaneously on two or more hard drives. If a hard drive fails, the system continues to function. The failed drive can be removed and a replacement inserted without ever bringing the system down. While the exact number of NT installations is difficult to ascertain, versions 3.5 and 3.51 have made significant general market penetration. This penetration seems to have been achieved with UNIX and Novell shops adding NT servers (as opposed to replacing them), as well as many first-time networks. Few CPA firms have used NTWS. Some CPA-specific Windows software packages do not work correctly, generally when the software developer has bypassed the Windows printing concept and made direct hardware calls to printers. Windows NT requires use of its print manager; software using bypass techniques will not print. However, nearly every software developer has committed to support Windows 95 (including most CPA-specific developers). In order to claim compatibility with Windows 95, they also have to be compatible with Windows NT (a Microsoft requirement). Thus, the potential for NTWS on CPA desktops now exists. Windows NT is a system "written totally from scratch"; it does not load DOS and does not have elements of DOS code embedded in its code. Most DOS applications will run on NT with better protection from problems than DOS applications on Windows or WFW. Some 16-bit Windows programs are less responsive on an NTWS than on WFW or Windows. Both Microsoft and Novell have released connectivity links between NT and Novell that work.

Currently, there are four relatively mature versions of Windows to choose from: Windows 3.11, WFW 3.11, NTS 3.51 and NTWS 3.51. In addition, Windows 95 has received more commitment from software developers than any operating system in history. Yet many CPAs are still operating solely under DOS. While learning Windows requires some commitment (in spite of the marketing promise that it can be learned intuitively), there is a conceptual uniformity that is pervasive between the operating system and Windows applications that is transferable between applications. That is, once a working knowledge of Windows is acquired, facility on a new Windows application can be gained faster than under DOS. A Windows application presents more coherent data on the screen concurrently than can be done under DOS, as it is a graphical interface (DOS is character-based unless the applications are handling graphics). Switching between tasks allows CPAs to be more productive. Although task-switching can also be accomplished under DOS (with specific add-on software), Windows does it better. Concepts such as electronic mail and scheduling (appointments, to-do lists, telephone messages, etc. can be better exploited using graphical interfaces; for instance, onscreen pop-up notification of the receipt of electronic messages works well under Windows. In addition, computer-assisted presentations are more effective under Windows. Many CPAs have yet to realize the potential of computer-assisted presentations to clients. With the current capabilities and ease of use of presentation software, this is now feasible, practical and effective, even for a single client on an important issue.

Collaborative projects can also be better accommodated under Windows. Moving documents around between specific collaborators and using the cut and paste capability of Windows increase the productivity of a group project. More elaborate workgroup concepts (such as Lotus Notes) also are dependent on Windows.

Many CPAs have used computers only as tools for specific tasks. A better concept is to think of the computer as the basic ingredient for automating all office functions. Client files, historically confined to paper, can be managed better if digitized. While just becoming practical for most CPA firms, imaging (digitizing documents) will provide a huge productivity gain. All business workers and most individuals will soon be routinely connected to the Internet. Browsers such as Mosaic and Netscape Navigator, which make using the Internet practical for mail, news, research and many other tasks, are dependent on graphical user interfaces (Windows, Macintosh, X Windows, etc.). Within five years, traditional analog technologies (such as telephones, modems and television) will become completely digital. Becoming completely digital means a Local Area Network (LAN) will host all communications activity concurrently with a traditional LAN activity. There are many ingredients to using technology at this level; Windows is but one of the key ingredients for the fully automated office.

While the retail version of Windows 95 was not released until Aug. 24, 1995, thousands of Beta copies have long been in use at evaluation sites. The appearance and organization of the interface have been radically changed. Some test users for whom the prior interface had become second nature have resisted the change. Users learning Windows for the first time should become competent under the new look and feel in approximately the same time as the old. While this product has been tested and evaluated more thoroughly by more people than any unreleased product in history, it will nevertheless have limitations and flaws. While Windows 95 does not require that DOS be loaded prior to Windows, some DOS code has been ported into the product. Whether this embedded DOS code compromises the promised "freedom from DOS" has not been fully ascertained. The likely reason for the throwback DOS code is the Microsoft promise that Windows 95 will run as well on existing computers as Windows 3.11 or WFW 3.11 in terms of robustyless. This decision is based more on marketing than on technology. Microsoft marketing does not want to be perceived as forcing users to make significant hardware upgrades in order to run Windows 95. The underlying hardware decision for CPAs to run Windows (any version) is different from many other users. CPAs typically run 10 to 25 different applications. Various applications stress computers in different ways. With the normal base applications most CPAs currently use, computers are stressed in all ways (processor speed and capacity, RAM speed and capacity, disk speed and capacity, network speed and capacity, backup speed and capacity). Therefore, CPAs need all of the capacity they can afford. What they can afford is driven primarily by a combination of the billing rate of the user and the extent to which that user can become more productive with computers. No CPA should be trying to run Windows on an Intel 386 or lesser processor. But neither should he still be trying to use an Intel 386 for a significant DOS workstation. All additional computers purchased should be intel Pentiums at a minimum, and the processor speed (90Mhz, 100Mhz, 120Mhz, 133Mhz, etc. should be as fast as can be afforded. Even though Windows will run in 12MB RAM, CPAs need more. With Windows 95, NTWS and NTS, more memory means more overall speed. The cost per megabyte of hard disk space will continue to drop to about the 15 cents per megabyte range; the need for the space and access speed will continue to grow exponentially. The determining factor should be whether the technology has become "standard" (a very subjective determination). Once it reaches this point, the only limits on capacity or speed will again be what is affordable. So what Windows 95 is really going to do is provide the time to plan and acquire significantly more robust hardware. If a CPA currently has sufficient hardware, NTWS 3.51 is more stable than the initial release of Windows '95 can be. It can also be configured to run with an interface (currently too buggy to use) that mimics the new Windows 95 interface. So although it is technologically different (and superior), it will look and feel the same as Windows 95. By the time the successor to NT is released (currently referred to as Cairo), it is possible (and perhaps likely) that Windows 95 and NT will evolve into a common product.

Of course, Microsoft Windows is not the only graphical interface system available. The Macintosh, UNIX (using X Windows on workstations) and OS/2 have had what may be more advanced systems longer. However, none of those systems have the market presence of Microsoft Windows. Consequently, significantly fewer application software developers write code for those systems; and when good software is available, it is generally more costly (due to less potential market). In order to have the greatest security of adequate and affordable application software choices, Windows is the safest choice. For the same reason, current DOS users should migrate to Windows as soon as possible. Since few software developers are currently spending any resources on improving DOS applications, all significant future software improvements will be available only in Windows versions.

The Microsoft victory on the desktop has not yet been duplicated to the same degree on the server. However, for CPA firms that often start with small networks and expand rapidly, NT can be more cost effective than others. The ability to add additional users on an as-needed basis is currently more flexible, achievable and affordable. NT does not require more licenses than is needed to add a single network user. Even if server competitors (primarily Novell Netware) become more affordable, NT has the administrative advantage of being Windows; more users can evolve to more server administrative tasks with less training. Paradoxically, some current network administrators prefer having some mystery surround network administration (a notion presumably driven by job security). Nevertheless, due to the current market presence of Novell on the server and the maturity and robustness of UNIX on larger servers, Microsoft is less likely to dominate the server world to the same degree that it has captured the desktop.

Thus, it will remain important that data and programs on dissimilar systems are capable of working in harmony. The Internet can be used as a model of what is possible and desirable. Most Internet Web sites are currently UNIX systems (although NT-based Web sites are growing rapidly). A user site (client) armed with a Web browser (such as Netscape) can access information on these different systems as readily as if it were on their local LAN or hard drive (depending on line speed and bandwidth, of course). Browsers on a Windows machine, a Macintosh machine and an X Windows machine can access and display meaningful data from the same source. The clear implication is that all digitized information deemed shareable can be put up on Web sites around the world and effectively accessed on demand by people everywhere. Ironically, we can currently interoperate globally more effectively than on most LANs. Interoperability makes useful digitized information more available and consequently more valuable. intelligent agents can be configured to continually monitor areas of interest (globally) and provide options to read important items they select. As historical paper files (client files) become digitized, this same concept can help organize and manage these files so that any piece of relevant data can be quickly located, displayed, edited and printed from any workstation (with permission).

Since a significant portion of shared digitized information will be graphical (photographs, videos, animations, etc.) and will consume huge amounts of storage space (a three-minute compressed video can be a 25MB file), the bandwidth of the transport mediums is the key to full implementation. Today's processors, disks and wires can accommodate these huge piles of bits. It is the bits flying through the air without invading each other's space or overwhelming governmental regulators that present the greatest challenge. What has historically been treated as regulated monopolies for the greater good and protection of the most people (television, radio, telephony, cable, etc.) will have to be reinvented. Neither the current business model nor the political model of these industries is going to be applicable in the short-term and long-term future. While the future is the network, huge bandwidth throughout the global network(s) is essential.

Many people have become very wary of the success and continuing aggressiveness of Microsoft. A reason offered by some for avoiding Windows is to avoid becoming dependent on a vendor who may become arrogant by gaining a dominant position in the market. Microsoft has in fact outmarketed the competition for some time. However, in order to sustain its growing market share, it has also produced more good products at more affordable prices than the competition. While that can change and justify more government intervention, competitors would be better served to focus on producing better and better-priced products.

Regardless of which technology vendors dominate the market, the rate of change will continue to escalate radically for the foreseeable future. As George Gilder alludes in his forthcoming book, Telecosm, the future belongs to the network more than to the processor. This factor coupled with seemingly infinite bandwidth available everywhere and affordable by everybody that is emerging on the computing horizon will drastically change all traditional ways of thinking about the information business. Windows constitutes only a small part of this change, yet it is an important part. All CPAS should be Windows-literate (even those planning on retirement). As networks become pervasive in our homes (replacing standalone computers, television sets, VCRs, telephones, etc.), Windows (or its successor) will be required. That can happen by the year 2000. Actually, much of it can happen sooner. Generally, many technologists Point to the time when most homes will be connected via fiber optics as the point at which radical merger of computing and entertainment devices will occur. But as Nicholas Negroponte, Professor of Media Technology at MIT and founding Director of the Media Lab at MIT, states in his recent book, Being Digital, using a technique called asymmetrical digital subscriber loop (ADSL), plain old copper phone lines can pump up to 6 million bits per second four times the rate required to sustain VHS-quality video). while fiber optics is nevertheless the long-term solution, we can now move more information around than we know how to manage. The current most limiting factor is the speed of moderns (even 38,400 baud modems). Although asynchronous transfer mode (ATM) will ultimately replace modern-type connections, integrated services digital network (ISDN) technology will be more easily available and affordable in more places sooner. in this evolution (and it is evolving rapidly), Windows will likely be the glue that allows nontechnical persons to become as proficient as highly technical persons and will shield the user from the complexity of the underlying architecture. While any graphical user interface (such as Macintosh, OS/2 and X Windows) could conceivably do this, it is more likely that Windows will dominate the next three years. And in a world of bits, three years can be an eternity.

Editor's note: Dr. Hicks, Mr. Love and Mr. Prescott are members of the AICPA Tax Division Tax Computer Applications Committee.

Editors: SAM A. HICKS, PH.D., CPA Associate Professor The R.B. Pamplin College of Business Virginia Polytechnic Institute and State University Blacksburg, Va.

JERRY L. LOVE, CPA Shareholder Davis, Kinard & Co., P.C. Abilene, Tex.