Given a complex problem, the solver should not attempt to guess the correct solution, instead they should strive to understand the reality around the problem and deduce the solution instead.
The reason I belive in this principle is that my experience has shown that given a complex problem, the time spent on guessing will take much longer than understanding the reality. Once the reality is sufficiently understood — the solution often becomes obvious.
The interesting characteristic at play here is the understanding — the complexity of the problem is different to each solver. If the solver understands the reality around the problem and potential solutions sufficiently well, then it will be faster to guess.
This principle could then also be worded as: Understand the reality around the problem until you are able to make accurate guesses.
There is one other interesting characteristic for this principle — the difficulty of understanding versus guessing. It seems to me like guessing takes a lot less mental effort than understanding and that is the reason why guessing is usally preferred by people that don’t follow or are unaware of this principle. This, however is a fallacy, since for a truly complex problem — it would have been more beneficial to spend a couple of hours to understand, rather than a week guessing. I assume that a week of guessing is more exhausing than a couple of hours of understanding. The ratio of time spent varies per problem, but a ratio such as this is quite realistic in my experience. For certain problems, the time spent guessing can be infinite.
The last characteristic of this principle comes down to learning — the nice thing about understanding is that it reduces amount of assumptions you have and makes the solution more general. If you understand the problem and the solution, you will be able to reduce the amount of defects in your solution, since you will also be able to connect the problem to other scenarios. You will gain more visibility in a sense. That visibilty will also increase your knowledge of that certain domain. You will be able to use that knowledge, when a similar, but not exactly the same issue occurs — you will know where to look. You will also be able to teach the area that you just learned.
Given that there are many benefits to deducing instead of guessing — I think it’s safe to say, that deducing should be preferred over guessing, even if it might initially take a bit longer.