This was painful to read. I feel like three—maybe five—paragraphs would be sufficient to tell your story. As it is, I kept wondering why I’m learning jargon like “level 3 insights” or what bits of information an entire section added. My version:
Hacking Business School
Goodharting “insights”.
In business school, in-class conversations constitute a significant portion of your grade. These are scored roughly by:
-1: You said something that actively hurt the learning process (e.g. blatant falsehoolds or -isms).
0: You said nothing, or nothing useful.
1: You brought up points from the readings.
2: You did some analysis on the readings (e.g. expected cost of policy A vs. policy B).
3: An insight; something nonobvious, and nonobvious how to arrive there (this is what makes it different from a mere analysis).
Obviously we wanted the highest grades, but we also didn’t want to (1) spend many more hours searching for insights, or (2) rely on luck to generate them for us. Just because it is not obvious how to arrive at insights to the rest of the class—or even the professor—does not mean there is not a formula to find them. We started by noting down which comments received ’3’s, then Goodharted on this objective.
We found three kinds of insights: those based on prior knowledge, inductive reasoning, and deep deductive reasoning, where deep means going farther than most people would assume necessary. Unfortunately, as students, we didn’t have a wealth of prior experience to draw upon, and inductive reasoning is hard, like NP-hard, so only the third kind of insight seemed like a plausible path. However, searching deeper than everyone else for insights would obviously require more work, which we’re quite opposed to.
But then… we stumbled upon a solution. Any time someone stops a search is an opportunity to go a little deeper. If someone says, “let’s assume for now...” and the rest of the class is nodding along, well let’s check that assumption! Most of the time, our assumptions will be right, but even if they’re wrong 10% of the time, we can expect to find a “nonobvious insight”.
For example, one class our professor began an analysis, and in a leading-the-witness style asked us, “Which airline routes depreciate planes more: long or short flights?” The knee-jerk response from most of the class was longer routes—because the plane is being used more (per flight). No one’s brain registered the “per flight”, until I questioned if being “used more” really means it depreciates faster. Suddenly, an insight was generated: maybe it depreciates more in the air on longer routes, but shorter routes have more frequent takeoff and landing—which might actually make them worse for wear and tear!
This simple trick, to question everything, ended up being the best insight I made in that class.
This was painful to read. I feel like three—maybe five—paragraphs would be sufficient to tell your story. As it is, I kept wondering why I’m learning jargon like “level 3 insights” or what bits of information an entire section added. My version:
Hacking Business School
Goodharting “insights”.
In business school, in-class conversations constitute a significant portion of your grade. These are scored roughly by:
-1: You said something that actively hurt the learning process (e.g. blatant falsehoolds or -isms).
0: You said nothing, or nothing useful.
1: You brought up points from the readings.
2: You did some analysis on the readings (e.g. expected cost of policy A vs. policy B).
3: An insight; something nonobvious, and nonobvious how to arrive there (this is what makes it different from a mere analysis).
Obviously we wanted the highest grades, but we also didn’t want to (1) spend many more hours searching for insights, or (2) rely on luck to generate them for us. Just because it is not obvious how to arrive at insights to the rest of the class—or even the professor—does not mean there is not a formula to find them. We started by noting down which comments received ’3’s, then Goodharted on this objective.
We found three kinds of insights: those based on prior knowledge, inductive reasoning, and deep deductive reasoning, where deep means going farther than most people would assume necessary. Unfortunately, as students, we didn’t have a wealth of prior experience to draw upon, and inductive reasoning is hard, like NP-hard, so only the third kind of insight seemed like a plausible path. However, searching deeper than everyone else for insights would obviously require more work, which we’re quite opposed to.
But then… we stumbled upon a solution. Any time someone stops a search is an opportunity to go a little deeper. If someone says, “let’s assume for now...” and the rest of the class is nodding along, well let’s check that assumption! Most of the time, our assumptions will be right, but even if they’re wrong 10% of the time, we can expect to find a “nonobvious insight”.
For example, one class our professor began an analysis, and in a leading-the-witness style asked us, “Which airline routes depreciate planes more: long or short flights?” The knee-jerk response from most of the class was longer routes—because the plane is being used more (per flight). No one’s brain registered the “per flight”, until I questioned if being “used more” really means it depreciates faster. Suddenly, an insight was generated: maybe it depreciates more in the air on longer routes, but shorter routes have more frequent takeoff and landing—which might actually make them worse for wear and tear!
This simple trick, to question everything, ended up being the best insight I made in that class.