ship, and unload a program shrinks dramatically. Unloading time can be cut from forty hours to fifteen minutes. Further agreement on the use of languages and on the necessary technical directions can speed the preliminary grooming of an imported program. The results of these agreements make a program cheaper for a teacher to order from another college and easier for him to handle once it arrives on his campus. However, an easy-to-unload program has to have been loaded with care somewhere else. This loading is not only a technical problem, it is a personnel problem. Rancher, Cowboy, Clerk Anyone working around a horse must learn to handle a pitchfork. The teacher, the programmer, and the computer center shipping clerk each view the program from a different perspective. The teacher wants the program to do the job, be it in research or in teaching. The programmer also wants the program to perform well. However, his preoccupation is with training the program, smoothing its quirks and polishing its routines so that it will perform at its best. He is often beguiled into prizing the elegance of the performance more than the accomplishment of the job. The shipping clerk knows that the program must be crated with appropriate instructions and shipped in a confined, unnatural state. The cowboy, preoccupied with the program's beauty on the open range, views this clerk with thinly veiled hostility. The clerk, preoccupied with shipping the program well, views the cowboy with ill-concealed scorn. Neither fully appreciates the other's point of view. The wise academic shares the programmer's love of the program on the open range, but realizes the need for shipment if the program is to get to other ranchers at different stations. He will tolerate the shipping clerk and perhaps even help him. The clerk will demand of the rancher that he provide a program well trained for use by others. In addition he asks for details on the program that will help his counterpart at another computer center to get it working according to the advertisement. These details include sample runs of the program and notes on the limitations of its intended use. Working with the program's handler, the clerk will crate the program with appropriate technical directions and ship it. Station to Station, Range to Range, Rancher to Rancher It is not the program 's fault if it is harnessed to the wrong job. The advertisement for the program could be a research note, published article, textbook or talk. It is from this advertisement that an alert professional first learns of a new application. Only after he has found out how he might use a program does he decide to import a copy of the original. The rancher comes down to the station to meet the train when a program arrives. He and the clerk discuss how the program should be groomed at the station and settle on a handler. The handler may be the teacher himself, one of his students, or a center employee. The wrangler will curry and brush the program and make sure that its hooves are properly shod for the region. He may help the rancher lead it home to see how it performs. During this final trial the rancher is preoccupied with checking advertisement against actual performance to see if the program lives up to expectations. When it meets with his approval, it is added to the corral, one more animal to help with the ranch work. Fast-Breeding Ranch, Fast-Breeding Range Riding a time-sharing program on a batch processing range is like riding a hobbled race horse: one has trouble understanding how it was bred to run. Riding a batch program on a time-sharing range is like riding a plow-horse: It can pull heavy loads, but slowly and only straight ahead. One method of promoting the spread of computer ideas is to improve the speed and convenience of breeding programs. This can be done a number of ways. A most common, but expensive, trick has been for a researcher with a large grant to hire a number of skilled programmers who write and modify programs to produce a string of steeds useful for his own local research. Another way to improve the convenience of writing programs is through the use of a good time-sharing system. The convenience of a fast-breeding, fast-training, interactive system allows a teacher to handle his own programs. It also means that more of his students know how to program and may be hired part-time. Finally, it means that larger programs (perhaps up to fifty lines) can be locally bred rather than imported from somewhere else through the computer center. Into the Sunset The parable makes the obvious prediction that the spread of computing ideas will be aided by: 1. Improving the shipment of the best trained middle sized programs, 2. Increasing the convenience of the local breeding of all programs, and 3. Effectively advertising good ideas to both farmers and ranchers. But the most important lesson of the parable is that anything that can vary, will vary: the program's behavior, the programmer's ability, the academic's sophistication, the cIerk's competence, the climate of research, or the needs of teaching. Because of this enormous variation, generalizations about the subject die young. A discussion of a computer program is most fruitful when it is most specific. Stories tell more than statistics.