Doug Carlston - Software People. An Insider’s Look at the Personal Computer Software Industry

True speech recognition of very large vocabularies might be a difficult problem, but it also offers the opportunity for the next truly major increase in practical functionality for computers. When it is no longer necessary to know programming, or even simple keyboard commands, when one can actually control a computer’s operations by talking to it, using a reasonably large vocabulary, then the true computerization of the world’s population will take place. The need for a voice-controlled computer is perhaps nowhere more widely perceived than it is in Japan, where voice recognition is believed to be the key to the creation of a workable Japanese wordprocessing system. The Japanese written language has so many characters that the present word-processing systems are very slow and clumsy; speech-based word-processing systems would improve these capabilities.

Just as humans are used to taking in information in a visual form, our species has many thousands of years of experience communicating information in an aural form. Once computers are able to capture and process our normal spoken communications, our ability to feed information to the computer will increase suddenly, in both speed and quantity.

Besides the capacity for high-bandwidth displays, nearfuture information storage devices based on videodisk technologies offer the opportunity to supply enormous amounts of data to people in a software format. A typical floppy disk today can store a few hundred kilobytes—equivalent to a few hundred pages of information. A more recent rigid-disk technology can store millions of bytes, equivalent to thousands of pages of information. But when storage technology goes optical, the unit of storage can get as small as a wavelength of light, which means that a platter the size of a long-playing record can hold hundreds of billions of bytes of information— including not only data like payroll records and bank statements but also books, symphonies, movies, lectures.

When cheap storage reaches the gigabyte level, as it is bound to do within the next few years, it will be possible to put entire libraries—huge amounts of video, audio, and text information—into the hands of individuals. Right now, these massive storage devices are used for certain specialized applications. The repair manual for a high-performance aircraft, for example, can be a half million pages long. A 747 would have difficulty loading and lifting its own repair manual! The same information can be handily carried (and accessed) by one repair person, however, if it is encoded on videodisks.

The way in which software is transferred from place to place is bound to change as well. A great deal of software will be brought to consumers by electronic means—via the telephone or television cable. Recent efforts by a variety of companies to make their mark by providing electronic distribution services have not met with the huge success predicted for them. But it is inevitable that electronic distribution will eventually succeed in areas where the software has a very high service component, as with a payroll program that needs to be changed every time the tax laws change at the city, state, or Federal level, or where large databases are involved.

Some of the most impressive accomplishments of the next few years may not even depend upon changing technology, however. Instead, they may derive from the ability of talented young men and women to push existing technologies to new limits. The most consistent efforts so far have been in the research conducted on artificial intelligence, the results of which are only beginning to emerge into the commercial markets after decades in the laboratories and thinktanks. The most prominent current example of this spinoff is the advent of “expert systems,” programs capable of transferring expertise from one human to another, which are already used commercially to locate valuable mineral deposits, and experimentally to diagnose diseases! Although true “thinking machines” might still be far off (or not possible), advances in many areas of artificial intelligence research have already guaranteed revolutions to come in microcomputer applications. And it is worth noting that artificial intelligence research is where hackers have tended to produce their most brilliant efforts (such as Richard Greenblatt’s pioneering chess program and Carl Hewitt’s ultra-high-level computer language named HACKER).

It is hard and often foolhardy to speculate about some of the more futuristic aspects of artificial intelligence research. But it is reasonable to expect that programming efforts based on new knowledge gained from past and current research soon will permit computers to respond more and more similarly to natural human communication. The area of natural language understanding—whereby computers would not only recognize words but would be capable of figuring out what sentences mean —is considerably more difficult than voice recognition, since understanding involves knowing a great deal about the world. (How can you tell the difference between “I am going to see” and “I am going to sea”? When you say “I saw that oil can leak,” do you mean that you watched oil seeping out of a can or that you realized such an event can happen?) But techniques for emulating humanlike communication, while falling short of true understanding, are bound to bring computing to an ever-wider population.

The nature of communication, the way in which people transfer information and knowledge, comes into play when programmers attempt to create software capable of emulating that communication. As it is, the recent effort of trying to figure out how to get a machine to respond in a humanlike manner to human input has already taught psychologists a great deal about how human thought and speech are organized. It has demonstrated how difficult it will be to program even the most advanced computers to emulate the kinds of feats normally accomplished by the human brain. It will surely be no loss to the public if the next great achievements in software come first from the social rather than the physical sciences.

As we all know by now, technological advances bring problems as well as opportunities. Our increasing reliance on the electronic storage and distribution of information opens doors for a whole new criminal class who lack the daring to engage in the kind of crimes that require their physical presence. Long-distance, telephone-mediated, computer-assisted thefts of information from data banks, including telephone and bank credit card numbers and personal, confidential credit information, are already occurring on a massive scale. Some of the software tricks for burrowing into bodies of data and removing information have been tagged with names like “worm programs” and are passed on from one hacker to the next as tokens of programming prowess. Indeed, to many people the word hacker has only this one, recent connotation of software thievery.

The use of programming talent for criminal purposes increases the need for law-enforcement people to have software skills, and so we see another effect the software revolution is having on our society—the rise of a new literate class. Just as the spread of literacy was largely responsible for the rearrangement of the class structure of medieval Europe, the rise of a new class whose skills are needed and valued throughout our society (and entry into which is not restricted by age, sex, or race) may help to break down traditional employment and class barriers.

Of course, there is nothing in the nature of the technology itself which guarantees that the new class will come into being in a strictly egalitarian manner. So far, boys have been far more interested in learning about computers than girls have been, and upper-middle-class white children have had far more access to computers than lower-class or nonwhite children. The impact of the software revolution could therefore serve to increase rather than decrease class and racial differences, particularly if public schools are unable to afford to become more involved in fostering computer literacy. The inevitable decrease in prices will make “low end” computers more widely available, but the latest, most capable personal computers will always tend to be more expensive.