yes, I think you both agree with each other. good thinking is an inherently slow process. the way to get fast is to do some caching, if you already explored th

yes, I think you both agree with each other. good thinking is an inherently slow process. the way to get fast is to do some caching, if you already explored the domain and stored the answers for it you can just remember the information. the problem is when the caching is done wrong. you explored only a subset but thought you explored everything. the other kind of fast thinking is when you go bullshiter route and act like an LLM you fast interpolate between known data-points without system2 validation and give plausible looking answers with full confidence, you'd be amazed by how many people get fooled by this.