3 ideas
Break a problem down to its fundamental physical truths and reason upward, rather than copying what others do with minor variations. When you analyze a system by its base components — like calculating a battery's raw material cost instead of accepting the market price — you discover that constraints assumed to be fixed are often just inherited conventions.
A venture is worth pursuing if the expected value is high enough, even when the most probable single outcome is failure. Importance and goodness of an outcome can outweigh low odds of success, so the rational decision is sometimes to attempt something you expect to lose.
The rate of improvement depends on how fast and how honestly you cycle through feedback on what you build. Actively seeking negative feedback from others — especially from those who tell you what's wrong rather than what's right — accelerates iteration because errors get corrected before they compound.