An Experiment on Problem Solving with Delayed Computer Responses

Abstract

Four subjects solved problems on line with the Lincoln Laboratory Reckoner. Controlled delays were interposed between output commands and the resulting outputs, which were via typewriter, cathoderay tube (CRT), or line printer; other commands were executed promptly and without reply. The mean delay was 1, 4, 16, or 64 s, with individual delays varying widely and unpredictably about the mean. Data on nine problems per delay per subject showed that at longer delays the rate at which commands were given decreased, especially the rate of output commands, and that the problems were solved with fewer outputs in about the same time. With two subjects, there was a marked effect of delay on the relative frequency of use of the three output devices.