Man-machine interaction, combining the best of the twin worlds of computer precision and human perception, should become the fastest growth area of computer usage during the next decade. 4. GLOBAL EQUIPMENT UPGRADING Machine intelligence during the next decade will be available at very low cost, lower, in many cases, than the cost of the other devices within the same piece of equipment. This implies a global upgrading of current "dumb" equipment into "intelligent" equipment, at a small incremental cost. There could also be a corresponding downward price change, but never below the dumb equipment minimum. Calculators with advanced features (trigonometric functions, memory, and programmability) will only be slightly more expensive than, and will therefore replace, the minimum four-function variety; this trend is quite visible today. Much more importantly, intelligent terminals, housing small computers as subsets, may not cost much more than those with a bare keyboard. This upgrading will be more striking in optical display terminals, where an added intelligent buffer will greatly enhance their ability to handle colors and complex picture-processing algorithms. Micro computers should approach, and in some respects exceed, current minicomputers in sophistication and performance. Whether this will significantly lower the price of small machines will depend on the attached equipment. For larger machines, peripheral equipment has long been the hardware cost-determining factor. Upgrading here should lift the system into a new performance category. We can expect the electronic memory to grow in size by more than one order of magnitude; working hand-inhand with a very fast cache memory, the combined effect is a superlarge, superfast memory, capable of extremely complex management chores, not the least of which is the [image]Micon MCM data terminal measures about 9" square and runs on rechargeable batteries effective handling of electronic disks and other storage devices to form powerful virtual storage systems. There will be upgrading of the I/O and communication interface. Large machines will communicate freely through networks, satellite, or packet radio. When a terminal deals with a "central machine," the latter may actually be a collection of computers reacting correctly under a uniform communication protocol. It is assumed that most of the outstanding software problems and bottlenecks will diminish through the added LSI computing power, and new systematic programming practices. New software will pay particular attention to data management algorithms and the human interface. 5. LOWERING THE USER-MACHINE BARRIERS It has been estimated that the cost to program and debug a line of code ÷ the cost to execute the line has now reached the astronomical value of 100 million (3). Clearly in a typical installation, the most expensive component is the human cost, which should now be minimized at the expense of machine time. indeed, human convenience should be maximized whenever possible. The relationship between the programmer and the machine has seen ups and downs. ln the early days of computing, users had physical contact with the machine in order to push the appropriate buttons, but had to state their needs through the unwieldly machine code. The advent of FORTRAN and other procedural languages permitted programming on human terms, but the user was soon ejected from the machine room and had to communicate through a batch-centered job-control language. The advent of terminals and time-sharing has helped the user to reassert himself, under the desirable illusion of direct machine involvement. But there still remain complex sign-on procedures, difficult control statements varying from layer to layer, incomprehensible error messages, unexplained delays, also unexpected system crashes, destroying the work of innocent users. The intelligent terminal, provided with powerful monitoring programs, can go far to serve as go-between, much as a resourceful receptionist mediating between an executive and a visitor. The work includes expanding simple sign-on codes into the proper format, explaining unusual happenings, catching and fixing simple errors, keeping statistics, recoding and storing locally for safekeeping security and economy. Small jobs can certainly be handled locally, from start to finish. With sharply lowered machine cost, interpretive computing on terminals will become common for small problems, especially for students. The conventional compiling process introduces an extra layer of problem transformation into the job, and is a source of misunderstanding. On the other hand, it is easy to learn the use of interpreters. Further, on a terminal every interpretive step can be monitored in terms directly meaningful to the programmer. Compiling and batch processing can be reserved for time-consuming programs, as an economic measure. Optimum interpretation, involving the real-time balancing between interpreting and compiling, should become a reality. 6. DATA BASES FOR NONPROGRAMMER USERS The computer, far from freeing the average citizen from drudgery, actually generates some resentment in him, because he has no direct use of the computer, yet is often the recipient of its less-desirable by-products, such as wrong bills and junk mail.