ST:1999
by Neil Harris
Computers that understand the spoken word? Modems with baud rates above 64, 000? Scanners that not only sense images but interpret them as well? Neil Harris, formerly Atari Computer's Director of Product Marketing, speculates where present computer technology will take us by the turn of the century--and beyond.
Outside the computer-using community, people have peculiar ideas about what computers are and what they can do. They picture computers as something out of science fiction, like HAL 9000, the brilliant but deranged system in 2001: A Space Odyssey; or, more optimistically, the Enterprise computer in Star Trek.
I'm convinced that the computer industry is working toward the Star Trek concept of a computer--one that does what people think a computer ought to do: it listens to and understands the spoken word, obeys commands with unflinching dedication, accesses a seemingly infinite amount of information and controls real-world devices directly. This is the industry's ultimate goal, and one worth pursuing.
START offered me this opportunity to speculate on the direction home computers will take in the next decade. Not an easy task--who in 1979 would have guessed where we'd be now? Of course, I can't be 100 percent correct with my predictions but I do know that all the concepts discussed here are now under development. And given the everchanging nature of the computer industry, some of these developments may be available sooner than anyone thinks.
Still the ST
The 1999 model of computer from Atari will still be called the ST, short for "16/32." But this signifies the number of processors, not the number of bits handled by a single processor. Thirty-two processor chips drive the input/output for speech recognition, simulated 3-D graphics and intelligent communications with other computers and computer-driven devices. The other 16 processors handle the internal computing tasks. Does this seem like a lot? Remember when people thought 64K of RAM was more than enough?
Processor chips are the brains inside computers. Every year they get faster, and innovations like RISC (Reduced Instruction Set Computer) architecture seem likely to continue this trend indefinitely.
Then the laws of physics intrude
We are rapidly approaching the limit of how closely we can pack the electron channels inside a processor chip. If the chips shrink much more than current technology, thus packing these channels even more closely, the electrons will "jump the tracks" from circuit to circuit and the chips will be useless.
Another significant problem is the bottleneck between a processor and its memory. The "bus bandwidth," the speed at which data moves between the processor and RAM, is a serious limit to system performance. If the bus is not fast enough the processor has to wait until the data arrives (known as a wait state). The efficient use of on-chip registers and memory cache can reduce this waiting time, but not infinitely.
Parallel processing can effectively bypass the present limits in both processor speed and bus bandwidth. Tasks are divided among several processors with separate memory for each, so that there can be multiple memory accesses at the same time.
By processing in parallel, computers can think very much like humans. You and I share the luxury of having minds that can process and control many bits of information simultaneously. Even with multi-tasking, however, today's computers can still do only one thing at a time. Such a quantum leap in processor power as is afforded by parallel processing will have dramatic effects on the capabilities of future systems.
Future Sights ond Sounds: Simulations Systems
Computer animation is limited by the power of the processor. As video resolution improves, much more "horsepower" will be needed to make things happen on the screen. Custom chips can help with some kinds of operations but not for the most complex and interesting ones.
Tomorrow's computer will simulate all sorts of real-world sights and sounds. Imagine your monitor displaying the image of a person speaking perfect English, with lips precisely synchronized to the spoken words and all other motions amazingly realistic. It's not a far-fetched notion and it could happen soon.
But more important than talking (since today's computers can do that already even though they tend to sound like Irish-Martian hybrids), computers will understand spoken commands. What percentage of the population is really comfortable with a typewriter-style keyboard? The term user-friendly will take on a whole new meaning.
Quickened Communications
Electronic mail is one of the most appealing features in telecomputing. I can send you a letter and you'll receive it a moment later. But there's a catch: you have to be online with a service like CompuServe or GEnie to collect your mail. If you're there when I send it, you get it then. If not, you get it when you decide to look. Clearly a system needs to be developed that will deliver your electronic mail regardless of whether or not you're present to receive it.
The answer is in the works in the form of the Integrated Services Digital Network (ISDN). The telephone companies have already begun to change over from their existing analog lines and switches to ISDN. Large corporations are also making the change and within the next 10 years you'll have ISDN service right in your home.
ISDN is an ideal answer to the telecommunications problem because it lets voice and digital data share the same lines--yet remain independent of each other. A typical modem operates at 1200 or 2400 baud. More costly models can handle 9600 and up to 19,200 baud. ISDN gives you a high-speed, 64,000-baud line along with a voice line and a slow-data line (equivalent to present modem speeds).
ISDN will break down the barriers between computers. At 64 kilobaud you'll be able to do things you can only dream of today, such as seeing real-time graphics between computers. And you won't have to go online with a bulletin board to collect your mail because ISDN hooks directly to your system, just like your present telephone line.
Databases: Dateline 1999
New storage devices available in the next few years will give us near-instant access to huge volumes of data. Atari's CD-ROM, for instance, presently holds about 550 million bytes of information but it is likely that storage densities and access speeds will greatly improve in the next decade. Combine this with the kind of fast access to outside information that ISDN provides and the dream of a superintelligent computer becomes all the more real. What is not in your computer will be found in someone else's.
Of course, problems arise with this proliferation of information. Masses of data must be organized in order to extract meaning from it. Enter hypertext, a software technology now in its infancy. Hypertext is a way to build links between pieces of textual information. The data you use doesn't even have to reside inside your system--a hypertext link can point anywhere. This is a very efficient way to organize data on CDROM disks and it is absolutely essential as a means to organize information residing elsewhere on a network.
Data Entry Made Tolerable
One reason office computer systems are more common today than home computer systems is that businesses can hire people to handle the day-to-day chore of data input. Not so at home. My personal financial records are not on my computer because I refuse to type in all that information. I need technology to do this job for me.
Imagine an intelligent scanner, one that not only senses an image but interprets it as well. Entering your checkbook transactions into the system will be as simple as feeding paper into a photocopier is today. The scanner reads the data and your computer is smart enough to know what to do with it. In fact, this same device will also serve as a laser printer, copier and fax machine. Of all the predictions made in this article, this is the one which is likely to happen at the earliest time--the necessary technology is already available on the hardware side and is developing fast on the software side.
Home Computing Beyond The Year 2000
In the twenty-first century, the ST will listen as well as speak, read as well as write, send and receive information without close supervision and even reason, after a fashion. This is a tall order for today's engineers but clearly foreseeable using present and emerging technologies. That computer from Star Trek is not far from the home.
And beyond that? Drawing from science fiction as well as computer science, there are many possibilities. Imagine, as cyberpunk novelist William Gibson does, a world in which you just don't navigate through computer networks, you experience them as multi-sensory data as well. Imagine even further cellular computers evolving from DNA research or "biological modems" that can transmit data directy to the brain.
If it all sounds too far-fetched, remember: So was that computer on your desk not too long ago.
For several years, Neil Harris was the acknowledged "voice" of Atari Computers. His most recent position with Atari was Director of Product Marketing. In September 1988, however, he resigned from Atari to take the position of Manager, Product Marketing with General Electric Information Services (GEnie) in Rockville, Maryland.