Top down design - Bottom up testing

With design and evolution processes being defined as essentially the same thing, just in different contexts, this opens up analysis of evolutionary processes more in the way we would analyse say software design techniques. Conversely, processes that are primarily seen as design or creativity (say new consumer products, movies, tools, software etc.) can be analysed more in line with how evolution is analysed. An important concept is that complex units (animal species, mobile phones, complex software) can be conceptually broken down into independently acting sub-units (individual animals, camera part of the phone, subroutines or software objects) to get a perspective of how the sub-unit affects the design/evolution success of the unit, or even the class of units.

There can be contexts in which because there is interaction between different levels, one cannot come to reliable conclusions. An individual animal's success may have situations which don't confer success for the species. If camera's are banned from certain situations then the whole mobile phone is devalued for that context.

When looking at countries, the way to test if a law is going to work for the benefit of the world, the ideal way to test it is to have two countries with all laws and circumstances identical except for the law change, and see which country works better, and then assume that this will transpose to it being better for the whole world. Dynamics of having prohibition countries next to non-prohibition countries will elicit all sorts of interfacial behaviour which can swamp the effect you are trying to test- e.g., it is easy for someone in prohibitionist Gujarat to point to the alcohol-fuelled chaos of the coastal enclaves that are part of Goa, Daman and Diu as evidence that prohibition is a good thing, but those enclaves are the way they are because of Gujaratis evading prohibition. This is the kind of example I am talking about for where there is interaction between different levels of abstraction. Things like mendelian traits in genetics have very little of this kind of interaction between the levels. A trait that is good for an individual is most commonly good for the whole species to have.

5)Complex units can be defined as a set of multiple complex subunits for Marconomic analysis. In this there is no one "level" more special than another. It is a way of modelling what is really happening, that will be useful in bounded conditions where there is little interaction between the levels. Different aspects of analysis will be better dealt with at different levels. For instance, The Panda's Thumb, talks at length about "units of evolution" being the Gene, the Individual, or groups of individuals etc. Dawkins famously in his book points to "the Selfish Gene" as being the primary unit, while others rate the individual as the unit, others favour group selection theories. Marconomics states that all these are valid models that will give the right dynamics (and answers) in different contexts.