========================================================================= Date: Sun, 6 Feb 1994 18:25:00 CST Reply-To: TK0JUT2@MVS.CSO.NIU.EDU Sender: CU-DIGEST list From: "(Jim Thomas)" Subject: Cu Digest, #6.13 CONTENTS, #6.13 (Feb 6, 1994) File 2--"DIGITAL WOES" by Wiener (Book Review) ---------------------------------------------------------------------- Date: 31 Jan 1994 16:26:51 -0600 From: ROBERTS@DECUS.CA(Rob Slade, Ed. DECrypt & ComNet, VARUG rep, Subject: File 2--"DIGITAL WOES" by Wiener (Book Review) BKDGTLWO.RVW 931223 Addison-Wesley Publishing Co. Heather Rignanesi, Marketing, x340, 73171.657@Compuserve.com P.O. Box 520 26 Prince Andrew Place Don Mills, Ontario M3C 2T8 416-447-5101 fax: 416-443-0948 or Tiffany Moore, Publicity tiffanym@aw.com Bob Donegon bobd@aw.com John Wait, Editor, Corporate and Professional Publishing johnw@aw.com Tom Stone, Editor, Higher Education Division tomsto@aw.com 1 Jacob Way Reading, MA 01867-9984 800-822-6339 617-944-3700 Fax: (617) 944-7273 5851 Guion Road Indianapolis, IN 46254 800-447-2226 "Digital Woes", Wiener, 1993, 0-201-62609-8, U$22.95/C$29.95 lauren@reed.edu When reviewing books on technical topics, one quickly learns to dread the work of those who do not actually practice in the field. (Yes, we are told that Wiener is a technical writer. They may very well be professionals, but the overwhelming majority are not technical professionals.) With this prejudice firmly in place, it came as a delightful surprise to find that "Digital Woes" is an accurate, well-researched, and thoroughly engaging treatment of the subject of software risks. Chapter one is a list of specific examples of software failures, large and small. The stories are thoroughly documented and well told. The choice of examples is careful, and useful as well, covering a variety of problems. One could, of course, add to the list. In the virus field programs are extremely limited in function and rarely exceed 3000 bytes in length, yet almost every viral strain shows some programming pathology; most of the damage seems to be done by mistake. The user interfaces of antivirals are subject to hot debate, perhaps more importantly than in other systems because of the risks involved in misunderstanding. In regard to decision support, I recall the assumption, on the part of Excel, that everyone wants to use linear forecasting. Everyone involved in technical fields will be able to add other specific examples. For those uninvolved, Wiener's work is quite sufficient and convincing. Chapter two is an explanation of why software contains bugs, and why software errors are so deadly. Techies will feel somewhat uncomfortable with the lack of jargon, but persevere. Initially, I thought she had missed the point of the difference between analogue and digital systems--until I realized I was in the middle of a complete and clear explanation that never had to use the word "analog". (Technopeasants will, of course, appreciate the lack of jargon. Rest assured that the same ease of reading and clarity of language holds throughout the book.) Chapter three examines the various means used to try to ensure the reliability of software--usually with a depressing lack of success. As with all who have worked in the field, I can relate to the comments regarding the difficulty of testing. At one point I uncovered a bug in the third minor variant of the fourth major release of the fifth generation of a communications program. Apparently I was the first person on staff who had ever wanted to keep a running log between sessions--and the functions I used combined to completely lock up the computer. Most RISKS-FORUM readers will by now be nodding and muttering, "So what else is new". However, Wiener here proves herself capable of some valuable and original contributions beyond the pronouncements of those working in the field. Noting that she is familiar with programmers who have never, in twenty years of work, had their code incorporated into a delivered product, she raises the issue of what this type of work environment does to the psyche of the worker. My grandfather carved the wooden decorations in our church, and, fifty years after his death, I can still point that out. However, in a career of analysis, training and support, I can point to little beyond an amount of Internet bandwidth consumed. (Many would say "wasted".) To the ephemeral nature of the craft, though, one must add the legacy of constant failure. Martin Seligman's "Learned Helplessness" points out the danger quite clearly. A similar thought was voiced some years ago over the impact on developing youth of the then new video games, and the fact that you could advance through levels but never, ultimately, win. These children are grown now. You may know them as "Generation X". Chapter four deals with means to prevent failure. Actually most of the material discusses recovery--assuming that the system will eventually fail, how to ensure that the failure causes the least damage. Chapter five is entitled "Big Plans" and looks at various proposed new technologies and the risks inherent in them. In this discussion Wiener warns against those who are overly thrilled with the promises of the new technology. I agree, but I would caution that public debate is also dominated by those strident with fear. The arguments of both sides tend to entrench to defeat the opposition, while the public, itself, sits bemused in the middle without knowing whom to believe. It is a major strength of Wiener's work that the field is explored thoroughly and in an unbiased manner. Many books which try to present an objective view of a controversial problem tend to trail off into meaningless weasel-words, but the final chapter here concerns "The Wise Use of Smart Stuff." Wiener lists a good set of criteria to use in evaluating a proposed system. The one item I would recommend be toned down is the axiom that personal care be excluded. I keep an old Berke Breathed "Bloom County" cartoon in my office wherein Opus, the Penguin, berates a computer for depriving him of his humanity until the bemused machine attempts to confirm that Opus is human. The perceived coldness of our institutions is often illusory. I once worked in a geriatric hospital and thought it a shame that our culture did not keep aging parents at home. Until, that is, I lived in a culture that did, and found that the "technology" of our hospitals provided more human contact to the old folks than did the "organic" home care. I also note that the belittled ELIZA is the only program to have passed the Turing test so far. A limited, unexpected, and hilarious pass, perhaps, but a pass nonetheless. I note, as I am reviewing this book, a press release by a headhunting agency that half of all executives are computer illiterate. The survey method is extremely suspect, and I assume these figures are so kind as to be ridiculous. I would heartily recommend this work to technical and non-technical workers alike. Particularly, though, I recommend it to those executives who are the ones to make the ultimate decisions on major projects. Please re-read it after the next vendor demo you attend. copyright Robert M. Slade, 1993 BKDGTLWO.RVW 931223 Postscriptum - my wife agrees with Peter Denning that I tend to editorialize in my reviews. This is likely true. "Digital Woes", however, deals with a topic which has prompted many editorials--and deals with it well. Permission granted to distribute with unedited copies of the Digest