:2BSpecial issue on computer software. :3BSci. Am. 251, 3 (Sept. 1984). In addition to the usual “Departments” for which the Scientific American is well known, the September 1984 issue is devoted entirely to computer software and applications. A. K. Dewdney, in his regular Computer Recreation column, examines the current state of perception theory. His very readable article shows the status of the subject and also points out the impasse into which its development has fallen. The quality of presentation is much superior to that in the more specialized books with which the reviewer is familiar.

Alan Kay leads off the special material with a good survey of the way in which recent software packages seek to be user friendly. He presents a clear picture of the steps which are necessary to plan and implement a new software package. The popular icon technique is much in evidence.

Niklaus Wirth, the perpetrator of PASCAL, describes Structures and Algorithms. Structured programming enthusiasts will no doubt derive pleasure from this material, but the general public will find it difficult to understand. A like remark may be made regarding the next article, by L. C. Tesler, on programming languages. There is an obvious bias in favor of structured concepts but, as usual, the examples which purport to show the inferiority of such languages as BASIC reveal either an unfamiliarity with them, or a lack of enthusiasm for them. Experienced programmers would be very unlikely to do things in the way described. As a matter of record, the BASIC program for adding up the odd terms of an array would not run in the form presented.

Operating systems are well described by Denning and Brown. The treatment is general, although there is some discussion of UNIX. This is perhaps a pity since many of the readers of Scientific American are more likely to encounter CP/M or one of its clones.

It is impossible not to mention the fact that the center pages of the journal are filled with a quantity of glossy advertising for the Apple Macintosh. In the light of this, the reviewer feels compelled to remark that the BASIC available on this 16-bit machine runs about four times as slowly as that available on most cheap 8-bit machines. In addition to this, an hour of experiment with the Mouse produced the same sense of loathing as one feels for the domestic pest of the same name]

Terry Winograd contributes an extremely good article on computers in linguistics. He makes clear the methods used for syntactic analysis and illuminates the real difficulties involved in the resolution of such ambiguous phrases as “She dropped the plate on the table and broke it,” or “The chickens are ready to eat.”

A. van Dam discusses software for graphics. The accent here is on creative art and some of the illustrations are spectacular. Surprisingly, there is not much discussion of CAD/CAM, although this subject is covered to some extent in the later article on process control by A. Z. Spector.

Naturally, there is the required presentation on Information Management, in this case by M. Lesk. The analysis of database operation and concept searching is well done, as is the material on such geographical applications as route finding.

Stephen Wolfram describes “strange attractors” in some detail, a subject which will be quite familiar to readers of Scientific American. He then discusses fractal theory and the interrelationship of these topics with such things as turbulence. This is a well-written article and much easier to understand than Mandelbrot’s book [1] which is cited in the van Dam article. Perhaps the most interesting feature of Wolfram’s paper is the way in which he leads up to a comprehensible account of Gödel’s work.

The last article, by D. B. Lenat, is titled “Computer Software and Intelligent Systems.” It expounds the virtues of expert systems and gives a rather superficial description of how they work. Surprisingly, there is no discussion of LISP and similiar aids. The reader must be content with a very illuminating advertisement for MACSYMA, which gives more information in one page than many authors do in ten.

In summary, this is an excellently produced and superbly illustrated issue of a respected journal. Despite the inevitable superficiality of some of the material, all people involved in the use of computer systems will find something of interest. It can be thoroughly recommended.