Review: Libicki

31 October 1995.

Review of Martin C. Libicki
Information Technology Standards

Review by Gio Wiederhold
Category K.1 The Computer Industry
Keywords: Standards, Operating Systems, Languages, Networks,

A work on standards, as presented here, should be available to all software developers concerned with the interaction of their work with that of others. This book complements provides an effective basis for material presented in serials, as the ACM Standards View magazine and in specialized articles. The importance of standards in our professional life is increasing, as the scale and distribution of computing is widening. Libicki covers six major areas of concern to the computing industry in breadth and includes topics as their development, their benefits and drawbacks, the federal role, and their futures. Libicki also sketches their histories, both technical and political, that have led to their current state and makes prognoses on their growth and failures. The futility of prescribing standards against the stream of public needs is frequently evident.

This book clarifies the standards development process and may encourage broader input from the communities actually affected by the standards. Lack of early, intelligent analysis has hindered standards development at times, since complex protocols are subject to inconsistencies and gaps of coverage. This book should encourage increased academic participation in the standards evaluation process by illustrating crucial and fascinating issues. It is still unclear how much such participation will help the process or the individuals.

Because of its breadth this book cannot provide detail for the various standards thrusts but Libicki manages to describe crucial conceptual features lucidly, while providing more than adequate references to background material through extensive end-of-chapter notes. There is no consolidated bibliography. All of the many acronyms one encounters in these standards seem to be defined and indexed, helping the reference function of this book, and leading the reader rapidly to relevant high-level concepts.

The six major topic chapters each focus on one particular standard, but spend about half of their pages on competing or related standards. The writing is factual, but occasional annecdotes and quotes liven up what is traditionally considered a dull topic. An early note, that "the world of computers, like Gaul, can be divided into three parts: mainframes, workstations, and PCs", illustrates the level of writing, ... few standards documents would quote Cesar.

The major sections cover

1. UNIX as the primary workstation operating system, with tentacles into mainframe and PC worlds. The freeware, LINUX, a recent arrival, has taken over some of the role of GNU UNIX as described, and its interoperability with Windows will skew some of the prognoses.

2. The OSI Openn Systems Interconnection reference model illustrates the problems of a prescriptive standard, which incorporates more layers than are needed for products. The competing internet standard, TCP/IP, showed that "anarchy works", but only if each independent actor is connected to the others by excellent interface standards.

3. CALS (Continuous Acquistion and Life-cycle Support) is the 1985 DoD initiative that requires that all documents for military equipment being acquired should be in machine-readable form. A great plan, but the standards were not mature, so that instead of technical drawings as CAD files, their image output, using facsimile (fax) standards, are being submitted. For such images compression became cricial, but does little to enhance life-cycle support. The CALS initiative has also affected the development of many standards in graphics and engineering design, with the end-objective of supporting concurrent engineering. In efforts as STEP (for the exchange of product model data) where "the momentum causes subcommittees to be formed by the month". Yet, little has been finalized and commercial acceptance for most CALS-sponsored efforts is in doubt.

4. Ada as prescriptive standards effort illustrates many of the problems with standards. The comparison with C and C++ also makes it clear that a language which is not desired by its end-users, the programmers and the teachers, will not succeed in environments where programmers are not tightly managed. The lack of effective compilers and software components did not help either: "Ada drives like a Volvo and is roughly as hard to find parts for."

5. ISDN as the basis for Integrated Services on Digital Networks sufferred from an initially weak prescriptive standard and less than expected demand for digital telephony. Improvements in analog technology keep whittling away at the customer end, while at the high-end engineering solutions not requiring end-to-end standardization reduce the need for the ISDN standard.

6. Digital Library Technology, exploiting the spread of communication, has adopted by acclaim a simple definition (HTML) of the structured mark-up language (SGML). It now finds HTML inadquate and is pushing forwards on many fronts. SGML, as a meta-standard, does not provide interoperation unless agreement in the specific document definitions (DTDs) is reached. Neither did SGML definers foresee the variety of media to be shipped over the web. The impact of recent developements, as Hot Java, which actually ships interpreters over the net with the material to be displayed, occurred to recently to be included.

An additional chapter addresses, less satisfactorily, cryptography (DES and successors), chip design (VHDL), geographic data standards (SDTS), and the Japanese standard microprocessor architecture (TRON). It does illustrate well the leverage expected, but often not obtained, by control of standards.

Any book which covers this broad a topic in information systems must deal with the rapidly changing world. Since the book was written AT&T UNIX has been resold by Novell to Santa Cruz Operations, leading to more consolidation and altering some prognoses. AT&T itself is now being divided into three parts without a Cesar, and no descriptions on the fate of the parts can yet be written. Standards for charging users for digital library materials by use, subscriptions, or printing, do not exist, but are needed. This book itself would benefit greatly if the author were able to keep the valuable material in the volume current and on-line. As a basis for understanding the standards process, the current paper volume is hard to beat.