Here is the actual paper describing the determination of the level of complexity of a task and/or interaction and how to try to simplify it in a practical manner.
The complexity reduction could be obtained by mean systems decomposition by componente parts (by subsystems, by functions and (or) in the time). The subystems must generate solutions options and the all objects (all functiosn or all times partitions) systems must to conciliate decisiones by parts according some structure. You can find concrete solutions schemes and its practical applications in my publications disponible in my Researgate.net profile
Completing the previous answer. This is almost the only way ofr reducing complexity and ccnciliate aposteriori the decisions of the parta for solving the original problem You can find some ideas, by example, in "Analysis and Synthesis of Engineering Systems" in my Researchgate profile.
Completing the previous answers: When the problem is related to to territorial direction probleens the decomposition needs in it partition using fuzzy thecnics, but frequently the object partition have natural character while ,p.e. decomosing the design of a car or another machine. In that case there are a lot of other complex problem related to decomposition for designing, process planning, fabrication, etc. and the partition in the time and by functions for production planning, etc.
In my modest opinion the problem consists not in avoiding complexity, but in taking it in account solving complex problems by parts using various kind of tools.