4 ms·

Some commenters here are saying, "I already know this stuff." Indeed. I'd be curious if people could put out a list of "advanced" mental models. For example, Ba

by delish 10y ago

Some commenters here are saying, "I already know this stuff." Indeed. I'd be curious if people could put out a list of "advanced" mental models. For example, Bayes' theorem is more advanced than Occam's razor. What's clearly more advanced than Bayes' theorem, and as useful? ET Jaynes' flavor of probability theory? I'd posit the more advanced version of active listening as, "being able to perform a bunch of kinds of therapy--freudian, rogerian, family and systems etc." Of course I don't mean you go get a license for these things. I'm positing them as difficult, generally-applicable life skills. I'm not claiming these are good examples; I think HN can come up with better ones.

VLM 10y ago
Thinking being a flux of information, and EE telecom theory having discovered all kinds of laws about flow of information, its no surprised that those models apply pretty well to the engineering tradeoffs of general mental models of thinking or thinking about information in other contexts. EE control theory class IS an entire senior year class on applying a model to something (a thermostat?) which isn't terribly hard, and then modeling and measuring its performance and finally optimizing the model which is pretty hard. Shannons law explains how good ideas, noise/distraction/bad ideas, depth of concentration or maybe total volume of information, and rate of mistakes all interrelate and how changing one (or several) will affect the others in general. There are some interesting tradeoffs in communication filter design (analog hardware or modeled in DSP) along the lines of you can freely trade smoothness in response (group delay, ripple, latency, monotonicity kinda), accuracy in response, and complexity/cost. These tradeoffs apply to everything in the world that processes things not just filter synthesis. There is some kind of chaos theory "thing" where as feedback mechanisms become more complicated, oscillation becomes inevitable and unpredictable. Doesn't matter if we're talking about high gain amplifier design or world economic models. This is aside from the general engineering mental models of a good engineer can freely exchange cost, reliability/safety, and performance. In fact it being enormously easier to exchange in those rather than expand, you can pretty much see thru transparent marketing that only mentions one or some factors. This applies to all of reality not mere structural engineering. I think the optics people could say a lot about their seemingly endless stable of aberrations. There are so many effects and interactions its surprising anything optical works at all, much less works well. Optics is almost a meta law that everything interacts with everything and constants aren't.
nhaliday 10y ago
I'm not sure about expanding on Bayes' theorem, but some other notions from ML/stats that would be good to know are overfitting/the bias-variance tradeoff and base rates. One instance where I've seen the former applied to society is the idea of research benchmarks getting stale from "overfitting". Even when researchers do cross-validation, we might still expect our exploration of the space of ML models to be skewed towards models that perform unusually well on well-known benchmarks. This was described in http://www.deeplearningbook.org/ http://www.deeplearningbook.org/ with reference to ImageNet (of course). As for the latter, pretty much every time I've seen a discussion of statistics on social or old media, 90% of the participants seem unaware that base rates matter.
bshlgrs 10y ago
I think a lot of [these ones](http://mcntyr.com/52-concepts-cognitive-toolkit/ http://mcntyr.com/52-concepts-cognitive-toolkit/) are really useful and more advanced.

by delish 10y ago

VLM 10y ago

Thinking being a flux of information, and EE telecom theory having discovered all kinds of laws about flow of information, its no surprised that those models apply pretty well to the engineering tradeoffs of general mental models of thinking or thinking about information in other contexts. EE control theory class IS an entire senior year class on applying a model to something (a thermostat?) which isn't terribly hard, and then modeling and measuring its performance and finally optimizing the model which is pretty hard. Shannons law explains how good ideas, noise/distraction/bad ideas, depth of concentration or maybe total volume of information, and rate of mistakes all interrelate and how changing one (or several) will affect the others in general. There are some interesting tradeoffs in communication filter design (analog hardware or modeled in DSP) along the lines of you can freely trade smoothness in response (group delay, ripple, latency, monotonicity kinda), accuracy in response, and complexity/cost. These tradeoffs apply to everything in the world that processes things not just filter synthesis. There is some kind of chaos theory "thing" where as feedback mechanisms become more complicated, oscillation becomes inevitable and unpredictable. Doesn't matter if we're talking about high gain amplifier design or world economic models. This is aside from the general engineering mental models of a good engineer can freely exchange cost, reliability/safety, and performance. In fact it being enormously easier to exchange in those rather than expand, you can pretty much see thru transparent marketing that only mentions one or some factors. This applies to all of reality not mere structural engineering. I think the optics people could say a lot about their seemingly endless stable of aberrations. There are so many effects and interactions its surprising anything optical works at all, much less works well. Optics is almost a meta law that everything interacts with everything and constants aren't.

nhaliday 10y ago

I'm not sure about expanding on Bayes' theorem, but some other notions from ML/stats that would be good to know are overfitting/the bias-variance tradeoff and base rates. One instance where I've seen the former applied to society is the idea of research benchmarks getting stale from "overfitting". Even when researchers do cross-validation, we might still expect our exploration of the space of ML models to be skewed towards models that perform unusually well on well-known benchmarks. This was described in http://www.deeplearningbook.org/ http://www.deeplearningbook.org/ with reference to ImageNet (of course). As for the latter, pretty much every time I've seen a discussion of statistics on social or old media, 90% of the participants seem unaware that base rates matter.

bshlgrs 10y ago

I think a lot of [these ones](http://mcntyr.com/52-concepts-cognitive-toolkit/ http://mcntyr.com/52-concepts-cognitive-toolkit/) are really useful and more advanced.