272 Ulrich Krohs
nobody is waiting for a particular result. Time and effort (in terms of, e.g., lethal varia-
tions) required in evolution, though they may have drastic consequences on the species
level, are simply irrelevant with respect to the evolved organism—in sharp contrast to the
engineering case.
If the congruence of F- and S-modules is an outcome of evolutionary processes and
thus contingent, a conclusion ought to be drawn with respect to the theoretical treatment
of the mismatch found in large metabolic networks. The epistemic goal, then, should not
be to identify the one and only valid method of decomposing a network and trying to show
that the other way of decomposition distorts the picture. From the functional view, the
structural picture looks similarly distorted, as does the functional map from a structural
perspective. Bias is relative and can therefore be ascribed reciprocally. To discredit one
method, one had to show that it misconceives the subject of inquiry instead of demonstrat-
ing a bias with respect to some other approach. But as long as functionality is considered
as relevant in biology at all, one should allow for an integration of functional modulariza-
tion into the biological account of a network. I admit that conceptual problems do at fi rst
occur with mismatching modules. However, if biological organisms are as they seem to
be, and if physiology is still regarded as relevant to biology, then the task is to solve the
conceptual problems and develop a more differentiated account of biological networks.
15.8 Conclusion
I show that insofar as they focus almost exclusively on the benefi ts or positive fi tness
effects of modularity, present explanations of the evolution of modularity of biological
organisms are incomplete. They can neither account for modularity that originates in
integrative rather than in parcellation processes nor do they explain that modularity comes
in degrees. What is missing is the consideration of the cost of modularity, as it is known
from the fi eld of technology. I single out different kinds of costs: costs of a longer period
of development, tara costs, autonomy costs, and costs for module-wise replacement. The
list is not meant to be complete, but it must not include costs for evolutionary
processes.
What then should an explanation of the evolution of modularity look like? It must
demonstrate, for the particular case considered, that the balance of the costs and benefi ts
of modularity lies on the side of modularity, and it must estimate the expectable degree
of quasi-independence of the modules. This requires data that are neither available at
present nor easy to collect. Nevertheless, without such effort, any explanation of the evolu-
tion of modularity by selective processes, even if it is based on sophisticated mathematical
models from population genetics, is but an adaptive story: it shows that there might have
been an adaptive evolutionary path leading to the modular organization observed, without
ruling out the possibility that exactly the proposed evolutionary pathway was highly