Question about diminishing marginal returns Re: Collapse of systems

Dear networkers,

    To your recommendations I tried to have a quick look at the following
books:
      1. J.Tainter The collapse of complex societies
      2. N.Yoffee & G.L.Cowgill The collapse of ancient states and
civilizations
      3. C.Quigley The evolution of civilization
      4. K.Dark The waves of time
    Here I have few questions (perhaps very naive):
       J.Tainter's book analyzes different causes that can lead to the
collapse and tries to prove that all these causes are not the real cause.
Instead, he sees the diminishing marginal return as the real cause. I am
not quite sure that I understand the arguments because:
     When argueing against the causes like mismanagement and social
dysfunction he uses the simple argument that human societies are problem
solving organizations. I think that this argument maybe correct in the
sense that mismanagement and dysfunction are rather symptoms than causes. 
On the same footing I wonder whether the diminishing marginal return is
also a symptom of collapse as if the society is a problem solving
organization, why it should continue investing in the complexity when
the benefit declines while it can stop the mismanagement and social
dysfunctions.
     I don't find the illustrations with increasing costs of agriculture,
education, R&D,...as a reason for diminishing marginal return as 
convincing enough. The marginal return is a global indicator which should
not be a consequence of one or several parts of society. If these
illustrations tend to say that all the parts will become increasingly
costly, the problem is with the unit used to calculate the cost. If you
calculate the cost of agriculture in terms of education costs, there is a
chance that it does not increase. Instead, I think the efforts should have
concentrated on 1) the increasing overhead 2) the inability of the society
to correct the situation.

     So I found Ken Dark's argument as a deeper reason. The diminishing
marginal return is just a consequence of the information processing and
path dependent development of complexity and can be avoided by an
improvement in the information processing. 

      Here from a physicist viewpoint, I would like to see more detailed
dynamics. We have a network system model with agents and processes for the
society already. As information flows in the network at least in a
simplified picture the behaviors of the variables that can lead to the
collapse can be recognised. A proper control on these variables can reduce
the risk of collapse with periodic cycles. I guess that a deeper
understanding of the dynamics of the information processing is required.

  I don't know whether I am on a wrong or right track to study this topic.
Any advices and comments are greatly appreciated.


Aiviet