THE FUTURE OF PERSONAL COMPUTING
A PAPER GIVEN IN MARCH, 1979
THE IEEE ROCKET CITY SEMINAR
ROBERT N. SEITZ, Ph. D.
THE FUTURE OF PERSONAL COMPUTING
For the purposes of this discussion, when I talk about a "personal computer", I am referring to a microcomputer with a full typewriter keyboard and an alphanumeric display like the Commodore PET 2001 or the Radio Shack TRS-80, as opposed to, say, a household robot or a chess-player or a microprocessor controlled appliance. (The technology discussions will apply to these "process control" applications, though, as well as to the TRS-80 class of personal computers.
It has been four years since MITS introduced the first Altair 8800 "MITS Kit", inaugerating the age of personal computing, and two years have elapsed since Commodore revealed that it would market a personal computer for "under $500". In reviewing personal computer progress over the last four years, three facts stand out. First, the Altair 8800 was no bargain, even by the standards of 1975. By the time it was expanded to do useful work, it cost about as much as the low end minicomputers of 1975, with considerably less software and company backing, and about three-or-four times as much as a comparably equipped PET or TRS-80 would later cost in 1978. The point is that personal computer prices have dropped by a factor of three-to-four since 1975. Second, Commodore and Radio Shack haven't found it necessary (or maybe even feasible) to drop their prices or discount their products over the past year-and-a-half, nor have most of the other vendors who have introduced personal computer products. Third,
although personal computers are available at entry-level prices as low as $350, the minimum entry-level price for a personal computer that, in my opinion, is more than a toy is $800 to $1,000, and the price of a minimum very-small-business computer is several thousand dollars. This leaves plenty of room for price-performance improvements over the next four yearsthat is, by 1983.
So much for the past. Where are we now? Today, the leading suppliers of personal computers are Radio Shack, Commodore and the Apple Computer Company, but there are about twenty other manufacturers trying to establish or maintain a presence in the marketplace. Recently, four of the largest video-game makers have announced personal computers - Bally, APF, Exidy and Atari. RCA and Magnavox are also marketing personal computers, though not in an aggressive way. (Their lack of aggressiveness has been a surprise to me because of the large potential market for video displays and tape recorders.) Two other large companies that are rumored to be preparing personal computer entries are Texas Instruments and Hewlett-Packard. Neither of these comapnies makes a policy of discussing future products before their official announcement dates but some news has leaked out to large investment houses. The two Texas Instruments entries are expected to sell for about $500 and about $1,000. They will reputedly both use the TI 9900 microcomputer. The $500 model will contain 4 kilobytes of random access memory (RAM) and 40 kilobytes of read-only memory (ROM) which will contain an operating system, BASIC, and, perhaps, PASCAL. It will feature plug-in Read-Only Memory program modules like TI's SR-59 programmer calculator. It will attach to a conventional color TV set for display. Delays in obtaining FCC approval for the TV-set interface may be the main reason the TI computer isn't already on the market. It is now expected to be introduced this June.
The Hewlett-Packard entry is also expected to cost about $500, and might well be a low-priced edition of the HP 9830 BASIC calculator. The 9830 was introduced in 1972 at a price of $6,000 with 4K bytes of RAM, a 32-character LED alphanumeric display and a cartridge tape drive. It is now obsolete but it might reappear early in 1980 in a $500 package if HP announces a personal computer then as predicted. If these two giants enter the personal computer market aggressively and begin to compete, some of the smaller companies will surely be forced out of the race - a fact to consider in choosing a personal computer.
Although Radio Shack and Commodore have not reduced their prices, two new personal computer products appear to offer more capability for the money in the $1,000 price range. - the Exidy "Sorceror" and intelligent Systems Corporation's "Compucolor II". The "Sorceror" sells for about $1,000 with video monitor and 8 K of ROM and is made by the nation's third-largest video game manufacturer. It includes serial and parallel interfaces, S-100 bus compatibility, a 240 X 512 resolution graphics display, a numeric entry pad, and provision to plug in "ROM - PACS" containing languages other than BASIC. The Compucolor II sells for$800 with 4K of RAM, 16K of ROM, a 51-kilobyte minifloppy disk, and a wide-band, non-interlaced color graphics display. Or it is available for $1,500 with 8K of RAM, a serial interface, and a 128 X 128 8-color graphics display with separate refresh memory. In order to appreciate the Compucolor II, we need to briefly review some facts about color computer graphics. Television sets, whether color or black-and-white, are not well-suited to computer graphics displays because their bandwidths are too low for crisp, flickerless displays. Unfortunately, high-quality color monitors cost $2,500 to $3,000. However, for the past few years, Intelligent Systems Corporation has offered low-cost color graphics terminals using wide band color displays which they produce inhouse. Furthermore, a 60-Hz repeat-field refresh technique is used for sharp, flicker-free color rather than the standard TV interlace approach. The display is limited to the eight colors red, green, blue, magenta, cyan, yellow, white and black, because these are the colors that can be obtained by turning on different combinations of the picture tube's three color guns. (More expensive color graphics displays costing $10,000 to $20,000 allow a virtually unlimited set of color combinations but this requires intensity-modulation of the three color guns, using high-speed D/A converters.) Providing low-cost, high-quality color graphics for home computers is part of the technological challenge faced by electronic engineers during the next few years.
When the Commodore PET was first announced in 1977, I thought that the personal computer would take hold like pocket calculators and would move rapidly into very small businesses and college students' homes. But that hasn't happened yet. Perhaps everyone could see what to do with a pocket calculator and could afford to pay $20 to own one, whereas they don't know what to do with personal computers and don't want to part casually with $1,000. I think the first key to overcoming this problem is more, better and cheaper software. For example, a good advanced math cassette written in BASIC for the PET, TRS-80, etc., would be helpful, as would specialized libraries for college students in physics, electrical engineering, statistics, etc. These need to be accompanied by instruction manuals that are "cookbooks" and include information about the subject. Another useful program tape would permit your home computer to emulate several different specialized calculators. Another class of tapes would provide programmed learning or review aids in various high school and college subjects such as algebra, calculus, chemistry, etc. Still another class of software package would be a "living drama" or game or simulation in which the user plays an interactive role. The computer would select pre-programmed responses for other fictional characters, depending upon what choices the user made. Another potential software market rests upon the repackaging of various combinations of existing personal computer programs on tape. Useful software packages are rapidly becoming available but they are still expensive considering that their quality and capabilities are unknown. I believe computer software will become a new publishing medium like books, records and audio tapes. The chicken-and-egg problem is that a very large sales volume is necessary to pay for high-quality software development. At the same time, large sales volume isn't a reality yet, and it depends partly on the availability of high-quality software. However, consumer software could become a major marketplace in itself with greater potential dollar volume than personal computer hardware.
Another personal computing development that doesn't seem to have happened yet is a massive takeover of very small business computing. Small business computers from companies such as Burroughs and IBM cost $20,000 and up, and even when personal computers are advertised for small businesses, they are priced at $8,000 to $12,000 - far more than the $3,900 of a TRS-80 with printer and dual floppy disks. Since small business computers are still selling for $8,000 to $12,000, there must be a reason. I can think of four.
First, low-cost business software packages have been few until the last few months. Second, the small-businessman needs help in selecting and installing hardware and software, and such expertise is somewhat rare with respect to hardware, and non-existent with respect to the new software packages that have just been announced. Also, the amount of=ne can afford to pay for such consulting services diminishes as computer prices drop. Third, many businessmen are going to need more storage than floppy disks can provide, and may find themselves willing to pay twice as much for 50 times as much storage. This is particularly true when the more expensive computer is a solid-looking, company-backed product from IBM in which they have confidence and which will have good resale value. Fourth, some of the bigger companies are willing to lease their products and to work out short-term trial rental arrangements with first-time users.
To summarize, the small-businessman wants to buy a solution to a problem, satisfaction guaranteed, and the larger computer companies will sell him just thatat a price. If you and I want to start a computer company to sell personal-computer-based turnkey systems to small-businessmen (including installation, training, maintenance and "hand-holding"), we can probably do it, but we'll probably lose money holding the customer's hand on the first few installations. One of the problems we'll fce is that of evaluating the business software packages that are now appearing on the market and selecting the one that most closely fits our client's needs. In selecting software, we'll also have to pay attention to software bugs, documentation, update likelihood and vendor stability. Another potential pitfall is the need for very expensive customized tailoring of software packages to fit each user's individual requirements. This becomes less and less practical as computer prices drop.
Another problem with personal computing is that computer prices are still high compared to other appliances or to the obvious value to the user. Technology will presumably take care of this problem over the next few years. However, it is difficult to forecast a reliable cost schedule for personal computers. The prices of electronic components are fairly predictable but about two-thirds of the cost of a fully-equipped system with dual floppy disks and printer goes into the mechanical components: the disks and printer. The disks and printer depend more upon manufacturing ingenuity and production volume than do the electronic components (although the disks and printer can also benefit from cheaper electronics). But an industry that can sell a tape cassette recorder for $20, a printing electronic calculator for $50 or an electric typewriter for $200 can surely eventually bring down the prices of disks and printers below today's levels. Of course, TV sets and tape recorders are already mass produced and are about as cheap as they will ever be so these components of personal computers will not drop in price. At the same time, integrated circuits will eventually get so low in price that personal computers cannot help becoming very cheapusing conventional TYs and cassette recorders, if necessary.
To summarize, personal computer prices will sooner or later drop considerably lower than they are now, while the pacing items will be the input/output peripherals.
Among the hardware trends that I see are the following:
1. Ever-cheaper RAM
According to my estimates, RAM prices halve about every 1.5 years, so that three years from now, in 1982, 64 kilobytes of RAM should cost about $150 and 6 years from now, in 1985, 256 kilobytes of RAM should cost $200 (with inflation). By the late 80's, one megabyte of RAM should cost about $250! If floppy disks fail to drop in price, they can eventually be supplemented with RAM, particularly since RAM can store data more efficiently than diskettes, because of the block data structure, overlay copy areas and housekeeping requirements of diskettes. (Seitz' Rule of RAM says that one byte of RAM is worth about four bytes of diskette.)
2. Higher-resoultion graphics
It takes 24 to 32 kilobytes of RAM to refresh a 256 X 256-point, 8-color display, and 96 to 128 kilobytes of RAM to refresh a 512 X 512, 8-color display. Such a display might be available for a home user by 1982 and 1985, respectively.
3. 16-bit microprocessors
The new 16-bit microprocessors could show up in personal computers in 1980 or 1981. While they will be fster than the current generation of 8-bit microprocessors, their principal selling point will be that they can address megabytes of RAM, permitting the vendor to sell more memory. A little later, in 1982 or 1983, floating point firmware may be included on these chips, followed by tru 32-bit microcomputers in the mid-80's.
4. 8-bit computer-on-a-chip
As 16-bit microprocessors appear, the 8-bit microprocessors may be re-manufactured with RAM and ROM on the chip, so that lower-priced, battery-powered personal computers become feasible. I foresee the personal computer market splitting into a high-end segment, wher prices remain at $500 to $1,000 and capabilities increase, and a low-end segment where personal computers overlap progammable calculators. For example, by 1979-80, we could reasonably expect to see Texas Instruments introduce a 4,000-byte successor to the 1,000-byte SR-59. By 1983-84, the true pocket computer could make its debut. Folded, it would measure about 3 1/2" X 5" X 3/4". Unfolded, it would present a full3 1/2" by 10" typewriter keyboard with one line of text displayed across the top. It would use BASIC as a programming language, would contain 8K to 16K bytes of internal RAM, and would use a micro-cassette for input and output. The point is that sometime in the early 80's, personal computers should blend with programmable calculators, eliminating expensive cabinetry and power supplies. An attractive fact is that the upward-growth path of desk and pocket progarmmable calculaotrs, under competitive pressures in that mass-production marketplace, also leads directly to personal computers. This provides another potential source for personal computers from an industry that is already consumer-oriented.
The introduction of a BASIC calculator by Hewlett-packard at the end of this year could be a step in that direction if it comes about.
5. Larger/cheaper diskettes
Competitive pressure from large-scale RAM, the introduction of quadruple density diskettes and cheaper electronics should lead to single-density mini-floppy prices in the $200 to $300 range in the next two or three years, with higher density diskettes available at somewhat higher prices. In this area, as I have said, much depends upon manufacturing volume and design ingenuity, making predictions difficult. Diskettes currently serve three functions: online memory expansion through data and program overlays, non-volatile online storage of programs and data, and offline interchange and archiving of data. RAM can eventually take over the first function; magnetic bubble or non-removable disks may accommodate the second function, but the third function is the unavoidable province of floppy disks.
6. Lower-cost printers
Matrix printers are mechanically simpler than electric typewriters. Great strides have been made in reducing printer prices since 1975; maybe they can be further reduced. The prices of electronic calculator printers point the way.
7. Flat-screen displays
Don't hold your breath until they arrive! Anyone who has followed electroluminescent, plasma, light-emitting diode, electrophoretic and liquid crystal technologies. knows that CRTs are well-entrenched. Flat panel displays may come but at first, they will be small, expensive and fraught with some early problems.
8. Magnetic bubble memories and charge coupled devices
Again, don't hold your breath. Charge coupled devices have been in limited production off and on but haven't been cost-competitive with RAM/ Magnetic bubble memory is a nascent technology, and may not be cheap enough for personal computers for two or three years or even longer.
9. Graphic input and output.
Graphics tablets and plotters are now available for less than $1,000 but major price reductions are still needed to bring them into the home. Joysticks and cursor controls are probably the near-term techniques for low-cost graphics input. A facsimile-based graphics printer might be one approach to a low-cost graphics hard-copy device. The only low-cost color printer available is a Polaroid camera. Ink-jet printers might eventually ddrop to a low price but right now, they're very expensive.
Telecommunications prices will probably drop with electronics prices now that AT&T must allow direct access of appropriate equipment to its telephone lines without renting a Direct Access Arrangement coupler form Ma Bell.
Looking ahead now to potential products and devices, here are some ideas.
1. The marriage of radio with digital electronics
A transistor radio could be equipped with a microprocessor which would store news and weather reports and display them on command on a 16-character alphanumeric display. Or messages could be transmitted at 120-characters-per-second over long-distance telephone lines between one personal computer and another to cut phone costs. Britain and Japan are using Teletext topiggyback alphanumeric information in TV broadcasts, a concept requiring digital technology. The February, 1979, issue of Byte Magazine describes a similar concept called Digicast for multiplexing information into FM broadcasts. Another possible cevide of this type is a pulsed marine radio distress beacon which could punch out a compressed alphanumeric identification code and approximate latitude and longitude coordinates at much higher pulsed power than a present-day continuous-broadcast distress beacon.
2. Artificial intelligence and household robots
Now that personal computers are available, amateurs may make rapid strides in artificial intelligence and household robots, since these are challenging and popular subjects. They are also subjects which haven't received lavish funding because of limited immediately-practical applications and the real-time demands which this work has placed on large computers. This work would profit from much larger and faster personal computers, as technology makes them available.
3. Electronic schedule pads
An extension of the pocket alarm clock, this device would store your appointment schedule and sound a tone when a particular event were coming up, along with a printed display of what the event was.
4. True pocket translators
Bilingual electronic dictionaries are now available. Pocket translators with partial translation should come in the 1980s, and eventually, far in the future, perhaps, will come voice input and output.
5. Word processing systems
These are still expensive for home and small-office use, but low cost home systems selling for less than $1,000 could become a reality in the latter 1980s. Eventually, far in the future, perhaps, the true voice-writer will arrive, capable of transcribing continuous speech.
6. Cartoon and graphic animation aids
As personal computer speeds increase, cartoon animation and dynamic graphics can appear in the home, probably in the latter 1980s. Eventually, far in the future, interactive computer 'movies" and simulations might be possible, using computer interpolation and extrapolation of stored imagery and information.
7. Music composition and performance aids
One useful device for music writers would be an audio/graphics system which assists in the transcription of music to paper. Another useful concept would be the chording, accompaniment and synthesis of music, with the drudgery performed by the computer and the creativity provided by the composer.
Access to stored data bases at places like libraries and airlines may become a reality, as may electronic interaction with catalog stores, banks, etc. Homes information retrieval may be used for encyclopedia functions, perhaps using video/digital disks.