230 Wybo Houkes
intentionalist framework; in both cases, the conceptual border between organisms and
artifacts is open, and intentionality and selection coexist. Finally, in section 13.4, I present
an argument against the confl ict images of this section, based on similarities between both
cases; I offer an alternative, open-border image; and I describe briefl y which tasks this
image leaves for philosophical analysis.
13.2 First Case: Evolutionary Design in Electronics
Over the past decade electrical engineers have become increasingly interested in the pos-
sibilities of designing circuitry through processes that resemble those that produce biologi-
cal items. Attempts to construct such design processes are known under such names as
“hardware evolution,” “bio-inspired systems,” and—the name I adopt here—“evolutionary
design” (ED). The idea of “growing designs” that are capable of adaptive self-reproduction
is certainly much older than the 1990s, but in the latter half of the decade it has rapidly
developed beyond the visionary stage: there are now many research teams and several
conference series devoted to ED, and there is signifi cant interest from industry.^3 Simultane-
ously there is a growing trend among biologists to simulate evolutionary processes, for
instance, by building computer models. Some of the results in these artifi cial-life programs
are presented at the same conferences and published in the same journals as those of the
engineering research just mentioned.
13.2.1 Aim and Approach
ED is motivated by the hope of designing circuitry quickly, innovatively, and without
continuous designer interference. The guiding idea in the fi eld is that through defi ning a
set of eligible components, a procedure or algorithm for constructing circuits from these
components, and a fi tness measure for evaluating the constructed circuits, the construction
and evaluation of circuits can be fully automated.^4 Moreover, researchers hope that the
results might outperform traditional design solutions; they believe that ED processes may
explore fruitful portions of the component or circuit “design space” that are ignored by
human designers.^5 The following hypotheses are representative of this goal:
(H1) Conventional design methods can only work within constrained regions of design space. Most
of the whole design space is never considered. (H2) Evolutionary algorithms can explore some of
the regions in design space that are beyond the scope of conventional methods. In principle, this
raises the possibility that designs can be found that are in some sense better. (Thompson, Layzell,
and Zebulum 1999: 167)
Given this ambition of supplementing or outperforming conventional methods, most
researchers in this fi eld attempt to develop ED processes without preprogramming them
for success through, for example, carefully selecting the design space of components,
defi ning a very strict fi tness measure, or removing unwanted elements from the space of