1.4.4.Objectives and nonobjectives

Because this volume is a work of theoretical applied science (Appendix A), some objectives appropriate to pure and applied experimental studies, or to pure theoretical studies, are inappropriate here. The following paragraphs outline both familiar objectives that are neglected and less-familiar objectives that are pursued. Appendix A describes these issues in more depth.

a. Not describing new principles and natural phenomena. An enormous literature describes new natural phenomena, but this volume describes the implications of known phenomena for new technologies.

b.Seldom formulating exact physical models. In analyzing functional systems, estimates should be accurate or conservative, but need not be exact. The goal is to gain a quantitative understanding of the relationship between structure and function, not to formulate exact physical models for their own sakes.

c.Seldom describing immediate objectives. Most of the scientific literature discusses either past achievements or objectives achievable using existing techniques. Few publications examine objectives that require substantial preparatory development (space science and high-energy physics produce many of the exceptions). Constructing the physical systems discussed in Chapters 8-14 and 16, however, is well beyond current capabilities. In this volume, only Chapter 15 describes objectives suited to the constraints of present laboratory technique. Our ability to model has outstripped our ability to make, and these studies exploit that fact to analyze ambitious objectives.

d. Not portraying specific future developments. This volume describes systems that can deliver performance orders of magnitude beyond that possible with current technology. Nonetheless, as research in this area expands, better designs will in most instances supersede these systems prior to their realization. Accordingly, they should be regarded not as portrayals of specific future developments, but merely as examples of what can be done.

e. Seldom seeking an optimal design in the conventional sense. In mature fields of technology, competitive pressures encourage a search for designs that are nearly optimal. In the exploratory phase of design, however, the more modest goal of workability suffices. A design can depart from optimality either (1) by being overdesigned and inefficient, or (2) by being underdesigned and unreliable. Here, (1) is acceptable, (2) is not.

f. Seldom specifying complete detail in complex systems. Given an established technology base, a designer can often describe a system at a high level of abstraction (Section 1.4.2) with confidence that compatible sets of components can be specified before placing the unit into production. This is easier if the product can be overdesigned and inefficient, because margins of safety in the initial design can then compensate for uncertainties in component performance. The approach pursued in this volume amounts to the analytical (as distinct from physical) development of a technology base with capabilities known within some error margins. The later chapters describe systems at various levels of abstraction, likewise using margins of safety to compensate for uncertainties in component performance.

g. Favoring false-negative over false-positive errors in analysis. In modeling and analyzing proposed designs, one would ideally distinguish workable from unworkable designs with perfect accuracy. But since models are never exact and accurate, errors of some sort are inevitable. These errors can be of two kinds: False-positive evaluations wrongly accept an unworkable design; false-negative evaluations wrongly reject a workable design. In exploring a new domain of technology, conclusions regarding the feasibility of systems rest on conclusions regarding the feasibility of subsystems, forming a hierarchical structure of analysis that can have several layers. A substantial rate of false-positive assessments at a subsystem level would make false-positive assessments at the system level quite likely: designs that rely on unworkable subsystems will not work, and may be beyond repair. False-negative assessments, in contrast, are relatively benign. Correcting a false-negative assessment of a rejected subsystem cannot invalidate the analysis of a system whose design omits that subsystem; indeed, correcting this error merely expands the list of workable designs. It is desirable to minimize errors of both kinds, but where uncertainties remain, it is better to bias analytical models and criteria toward safe, false-negative conclusions. This strategy guides the following chapters.

h. Describing technological systems of novel kinds and capabilities.The objective of this volume, then, is to provide an analytical framework adequate for designing nanomechanical systems (Chapters 3 to 10), and to begin to exploit that framework by designing systems capable of processing both information (mechanical nanocomputers) and matter (molecular manufacturing systems). A further objective is to show how these systems can be implemented, starting from our present technology base (Chapters 15 and 16). Achieving these objectives can define fruitful goals for experimentation, simulation, and software development, and can provide a better basis for understanding human capabilities and choices in the coming years.

1.5.Overview of following chapters

In a work of this length in a new field, the relationship among topics can easily become lost in a mass of detail. This section provides a chapter-by-chapter overview, describing relationships and indicating areas where a detailed analysis merely shows that a simpler analysis is sufficient.