About complexity

It seems to me that the complexity of any system (solution) that doesn't contain unnecessary redundancy is constant given the set of requirements is has to fulfill. It is only possible to redefine that complexity - to move it from one level to other but one cannot reduce it. So what are the possibilities to tackle with complexity? Basically we can either abstract or partition.

Above description applies to all knowledge domains for example in encyclopedias it is rarely possible to present the whole information in a linear way. Instead people create chapters or articles and in order to keep information coherent invent special terms for referring to related domain concepts. Now the problem with this approach is that for retrieving some new piece of information first we need to be able to find appropriate article and secondly we need to be able to understand the abstractions (domain language) used in that article. Otherwise the mechanisms used to manage complexity will start working against us cause we still need to browse through large amount of data and it will be hard to re-create consistent understanding of whatever we were looking for.

If we take this back to software world it means that we need to have set of subsystems or modules and we need to use consistent and commonly known language for describing concepts in both problem and solution space. If we are not aware of the names of patterns or design approaches widely used we make the use of abstractions much less effective. Also if we are not able to pay enough attention to splitting the responsibilities of the system we will put other developers into situation where they have no "table of contents" of the system and hence need to browse through large volumes of code every time they need to find something.