TABLE 6.1 Principles of Operation for Multicellular Organisms and Networked Computing


Multicellular Organisms

Networked Computing

Collaboration between highly specialized cells

Cells in biofilms specialize temporarily according to “quorum” cues from neighbors. Cells in “true” multicellular organisms permanently specialize (differentiate) during development. Loss of differentiation is an early sign of cancer.

Today most computers retain a large repertoire of unused general behavior susceptible to viral or worm attack. Biology suggests that more specialization and less monoculture would be advantageous (although market forces may oppose this).

Communication by polymorphic messages

Cells in multicelled organisms communicate with each other via messenger molecules, never DNA. The “meaning” of cell-to-cell messages is determined by the receiving cell, not the sender.

Executable code is the analogue of DNA. Most PCs permit easy, and hidden, download of executable code (Active-X or even exe). However, importing executable code is well known to create security risks, and secure systems minimize or eliminate this capability.

“Self” defined by a stigmergic structure

Multicelled organisms and biofilms build extracellular stigmergic structures (bone, shell, or just slime) that define the persistent self. “Selfness” resides as much in the extracellular matrix as in the cells.

Determination of self is largely ad hoc in today’s systems. However, an organization’s intranet is a stigmergic structure, as are its persistent databases.

“Self” protected by programmed cell death (PCD)

Every healthy cell in a multicelled organism is prepared to commit suicide. PCD evolved to deal with DNA replication errors, viral infection, and rogue undifferentiated cells. PCD reflects a multicellular perspective—sacrificing the individual cell for the good of the multicellular organism.

A familiar example in computing is the Blue Screen of Death, which is a programmed response to an unrecoverable error. An analogous computer should sense its own rogue behavior (e.g., download of uncertified code) and disconnect itself from the network or reboot itself periodically to give itself a clean initial state.


SOURCE: Steve Burbeck, IBM, personal communication, October 11, 2004.

concurrency, multilevel description, and object orientation, Kam et al. constructed a T-cell simulation that presents its results by displaying animated versions of the model’s Statecharts.

A second example is provided by the work of Searls. It is a common, if not inescapable, metaphor that DNA represents the language of life. In the late 1980s and early 1990s, David B. Searls and collaborators made the metaphor much more concrete, applying formal language theory to the analysis of nucleic acid sequences.76 Linguistics theory considers four levels of interpretation of text: lexical (the


D.B. Searls, “The Linguistics of DNA,” American Scientist 80:579-591, 1992. Formal language theory is a major subfield of computer science theory; it is based on Noam Chomsky’s work on linguistics in the 1950s and 1960s, especially the Chomsky hierarchy, a categorization of languages by their inherent complexity. Formal languages are at the heart of parsers and compilers, and there exists a wide range of both theoretic analysis and practical software tools for the production, transformation, and analysis of text. The main algorithmic tool of language theory is the generative grammar, a series of rules that transforms higher-level abstract units of meaning (such as “sentence” or “noun phrase”) into more concrete potential statements in a given language. Grammars can be categorized into regular, context-free, context-sensitive, and recursively enumerable, each of which requires more algorithmic complexity to recognize than the level before it.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement