The Bandwidth Famine Data transmission first became a topic of general discussion in the computer field in 1960. Earlier, there had been military interest centered on the SAGE project for air defense, but that project had been kept strictly under wraps and few people knew about it. My own interest in data transmission began when I attended a conference in Delft, The Netherlands, in September 1960.

The meeting was organized by the Institute of Radio Engineers, which was then the leading institute concerned with computer hardware. A few years later it was to join forces with the American Institute of Electrical Engineers to form the IEEE. Interest in data transmission grew rapidly and received a stimulus from the development of time sharing in 1963. Since then, there has been steady progress towards better reliability and wider bandwidth FAQ.

Initially, there were doubts about the adequacy for data transmission of contemporary telephone lines. It was feared the lines would turn out to be too noisy and too liable to interruption. At Delft, statistics were quoted for noise on telephone lines and there was much speculation about its various causes.

It was clear that a good deal of the noise was caused by maintenance work being done in cable ducts and elsewhere. It was noted that the Hawaii cable, which lay undisturbed at the bottom of the ocean, was relatively noise-free. One speaker remarked that if only the maintenance workers would keep out of the way, the situation would be improved!

However, it soon appeared that early fears about the inadequacy of telephone lines were exaggerated and that, by the use of error control techniques based on redundancy, even noisy lines were perfectly usable, and could in fact be made to yield as low an error rate as was necessary. The effect of noise was to reduce the rate of transmission, not to make the circuit unusable.

Telephone lines have improved over the years, and mechanical switching has given way to electronic switching. A modern adaptive modem can squeeze a great deal of bandwidth out of a voice-grade line. Further improvements may be expected. These will be welcome for data, but probably not be noticed by ordinary telephone users. For a time, people were generally comfortable with the bandwidth provided by a voice-grade line. By the late 1960s, however, the need for wider bandwidths was being felt. The ARPANET used lines having a data rate of 56 Kbits/sec.

The demand for higher data rates continued and was more than could be met. The result was a bandwidth famine which has continued with some recent relaxation to the present day. It has been a true famine in the sense that it has not been possible to acquire enough bandwidth for all purposes, no matter how much one was prepared to pay.

Until the middle 1970s, the computer community was content to use the telephone company's offerings for local areas as well as for wide areas. Around this time the Ethernet was announced and interest in rings for data communication was revived. From the beginning, both approaches emphasized wide bandwidths, which were achieved as a natural consequence of using computer engineering techniques in place of telecommunication techniques.