Version 7 Dysfunctions of highly parallel real-time machines as ‘developmental disorders’: Security concerns and a Caveat Emptor Rodrick Wallace Division of Epidemiology The New York State Psychiatric Institute ∗ September 15, 2009 Abstract A cognitive paradigm for gene expression in developmental biology that is based on rigorous application of the asymp- totic limit theorems of information theory can be adapted to highly parallel real-time computing. The coming Brave New World of massively parallel ‘autonomic’ and ‘Self-X’ machines driven by the explosion of multiple core and molecular com- puting technologies will not be spared patterns of canonical and idiosyncratic failure analogous to the developmental dis- orders affecting organisms that have had the relentless benefit of a billion years of evolutionary pruning. This paper provides a warning both to potential users of these machines and, given that many such disorders can be induced by external agents, to those concerned with larger scale matters of homeland se- curity. Key Words: developmental disorders, cognition, highly parallel computation, homeland security, information theory, real time. 1 Introduction A cognitive paradigm for gene expression has emerged in which contextual factors determine the behavior of what Co- hen calls a ‘reactive system’, not at all a deterministic, or even stochastic, mechanical process (e.g., Cohen, 2006; Cohen and Harel, 2007; Wallace and Wallace, 2008, 2009). The different highly formal approaches are much in the spirit of the pioneer- ing efforts of Maturana and Varela (1980, 1992) who foresaw the essential role that cognitive process must play across a broad spectrum of biological phenomena. A relatively simple model of cognitive process as an in- formation source – a generalized ‘language’ – permits use of Dretske’s (1994) insight that any cognitive phenomenon must be constrained by the limit theorems of information theory, ∗ Address correspondence to: R. Wallace, 549 W. 123 St., 16F, New York, NY, 10027. rdwall@ix.netcom.com. Affiliation for identification only. in the same sense that sums of stochastic variables are con- strained by the Central Limit Theorem. The viewpoint per- mits a new formal approach to gene expression and its dys- functions (e.g., Wallace and Wallace, 2009, 2010). We have already applied something like this to highly parallel com- putation (Wallace 2006, 2008a, 2009, 2010), and now extend that work to illustrate the broad range of pathological behav- iors such machines may express, taking a perspective much like that of developmental biology. Machine failure becomes, from this viewpoint, analogous to developmental dysfunction, in the sense of Wallace (2008b), who argues that ecosystem resilience theory permits novel exploration of comorbidity in developmental psychiatric and chronic physical disorders. Structured psychosocial stress, chemical agents, and similar noxious exposures, can write dis- torted images of themselves onto child growth, and, if suffi- ciently powerful, adult development as well, inducing a punc- tuated life course trajectory to characteristic forms of comor- bid mind/body dysfunction. For an individual, within the linked network of broadly cognitive physiological and mental subsystems, this occurs in a manner recognizably similar to resilience domain shifts affecting a stressed ecosystem, sug- gesting that reversal or palliation may often be exceedingly difficult. We will suggest similar problems must necessarily afflict highly parallel ‘autonomic’ or ‘Self-X’ machines, particularly those acting in real time. The current stampede by indus- try toward using such devices to govern our transportation, power, and communication systems, to control individual ve- hicles under fly-by-wire protocols, run nuclear reactors, chem- ical plants, oil refineries, and so on, raises important issues of homeland security, given the relative ease of inducing develop- mental disorders in the highly parallel systems that determine gene expression. We begin by recapitulating recent models of development, following Wallace and Wallace (2008, 2009). Subsequent sec- tions significantly expand that work. 1