- American Physical Society Sites
- Meetings & Events
- Policy & Advocacy
- Careers In Physics
- About APS
- Become a Member
Rebecca Slayton, The MIT Press, Cambridge, Massachusetts, 226 pages, plus end notes and references. ISBN 978-0-262-01944-6, hardback, no price stated.
As the title suggests, this book is not an easy read. Although it is not “technical” in the sense physicists are used to (that is, using mathematics to explain concepts), it is very technical as a historical reference book. The author’s goal is to document and explain the complex and “messy” (author’s term) interdependencies between high-level physics and engineering advisors to the U.S. government, and the emerging and evolving fields of computer hardware and software engineering, in the context of efforts to develop air and missile defense systems for the country. This obviously is not an easy task, and the author cites a tremendous amount of reference material and first-person interviews to buttress her arguments. I can foresee this book being used as a textbook for a graduate course in the history of technical policy-making in the U.S.
With such wide-ranging goals dealing with both political and technical subjects simultaneously over a 63-year time period, the author has a challenging task indeed. One method she uses to address this multiplicity of subjects is repetition of key points as she jumps back and forth between the three narratives of missile defense, policy advice, and evolution of software engineering. A major theme throughout the book is the impossibility of creating “perfectly reliable” software for extremely complex systems, and the way physicist and engineer advisors to the Federal Government consistently underestimated (or were oblivious to) this problem over most of the 50 years from WWII to the mid-1990s. A companion theme is how the search for provably reliable software for missile defenses during this period drove government, industry and academic software developers to evolve the discipline of Software Engineering. The author does an admirable job of showing the difficulties inherent in creating a technical discipline from scratch, as was the case with computer programming between the late 1940s and the late 1990s. Within this theme, the differing viewpoints on the reliability issue held by those who stood to profit from government contracts or who were “true believers” (my term) in missile defense, versus those who felt they were “objectively” analyzing the reliability problem are thoroughly reported and carefully referenced. A consensus that seems to emerge from the Software Engineering Community, is that the reliability of the software for most systems of “arbitrary complexity” can be improved to “almost perfect” through continuous repair of glitches, even (perhaps especially) after the system enters use. The author quotes expert claims that this real-time improvement was/is not possible for nuclear missile defense systems, because those systems are intended for one-time use, and cannot be realistically tested beforehand. While that may have been true in the past, I am not sure it is generally true today, given the variety of missile defenses (for different types of threats) the U.S. has deployed or has in R&D at present.
Another key point the author argues effectively is that complex technological systems of all kinds, not just missile defense systems, by their nature, have social and political components. These components introduce “arbitrary complexity” into the software of these systems, in the sense that in the real world, final, fixed requirements for the software are impossible to define. This is because these requirements are ever changing with the whims of politics, history and technology itself. The author shows that the professional computing community itself is divided over the eventual outcome for this issue: Will “perfectly reliable” software forever be out of reach due to fundamental constraints, or can best practices, continual testing and hard work create reliable software for at least some complex systems?
One addition that would improve the book is a glossary of acronyms. The author is faithful to define every acronym when it is first used, but may not repeat a given acronym for 20 pages or more, and by then the reader has forgotten its meaning. I have spoken in Government acronyms for 50 years, and still would have appreciated a listing for some of the acronyms the author used. The book’s detailed index helps in this regard, but a separate, single glossary would be an improvement.
While historians are the primary audience for this book, it does contain a cautionary lesson for physicists. When providing analysis and advice on complex, multi-disciplinary systems, we should never assume that the physics is the most difficult or limiting component. Other, possibly newer, disciplines/technologies/human factors may limit a system more than the physics.
Ronald I. Miller
DoD/DIA/Missile & Space Intelligence Center (Retired)