Techniques for arguing for a particular logic. Suppose your logic is valid, what must have been true and how that might contradict to existing constraints. If there a hidden explanation for the observation in addition to this line of logic. Is there a premise for this logic turns out to be invalid.
Constraints and system. How a constraint manifest itself in a system. Sometimes it is not as straight forward as our intuition would allow us to perceive. But there might be things that we can prove, such as inequality, distribution, expected returns.
Looking at things from its smallest unit, a small portion of time, a small portion of control. How those small portions would manifest itself into macro parameters.
Bootstrap is much more important than perfection.
If the optimal solution is not feasible, what are the alternatives? For instance, if I can’t determine the optimal choice for one step, can I study the optional solutions for a sequence of actions? This is a rather important leap for controls involve stochastidy.
Reasoning Backwards. The first time I was this style of thinking was in game theory, 100 coins, 5 people, how do you distribute the coins such that you would gain the most votes. The constraints is that the voting is conducted from number 5 person to you. Turns out, there is a clear structure if you reason backward. This idea is used in control as well. Rolling Cost To Go is the idea of carrying information in the terminal state all the way back to the starting point of the trajectory. It requires an understanding of the sytem model. The Q learning stufff, on the other hand, is a bootstrap structure such that the terminal state information get to propogate backwards along the time line. Ferminization.
Key element. Thinking is most interesting when it concerns a complex system. There are steps that are much more important than does others. Almost like, there are steps that would lead to phase shift while others are just incremental improvement. How to figure out what are the key elements? For instance, when it comes to skillsets, there are some that are more critical compare to the others. For instance, language skills, the ability to learn, the ability to abstract etc. They are the skillsets sthat would lead up to other things and therefore should spend more time on. The same is true for a project. It is much more than just following a script and work things one with labor, but the first step of every project should be the attack on the most critical step.
Scenario Planning. This is way to basic, I would not elaborate here.
See the truth, nothing but the truth. This is a skillset that is underrated. There are lot of social norms which pretend to be the truth, but they are just that, social norms based on the economical development of that time. Ususally, there is a natrual state of things and forces that prevent things to be in its natural states.
when a theory breaks down, can’t explain the observed reality, how to handle it? Are there assumptions can be adjustted? or is there a logical flaw?