In the same way that a probability represents your state of information about an event or action your assessment of the complexity of a task or problem is a function of your state of information and relevant expertise.
I think this has implications for models like cynefin where the assignment to a domain (simple, complicated, complex, chaotic) is presented as objective but I believe this assessment if a function of your information and understanding.
One implication for designing software is to use familiar UI metaphors--which are viewed as simpler and easier to use.
One implication for competitive analysis is that you cannot assume that your competitors bring the same information and level of understanding to a problem that you do. In best case you have more information and expertise. In the case you need to anticipate, you have less information and expertise, and what looks complex or chaotic to you may be complicated to a competitor.