“Sometimes magic is just someone spending more time on something than anyone else might reasonably expect.” —Teller
hypothesis: a big part of doing a job well is just being willing to spend a huge amount of time absorbing all of the relevant context and building a mental model of the problem. reading stuff, talking to people, looking into the reasoning behind claims that other people just take at face value, looking at related stuff, memorizing the rough order of magnitude of important numbers, etc. if you absorb 10x more context than someone else, then you can be hugely more effective than they are to an extent that looks almost like magic
scrolling is a maladaptive perversion of the instinct to pack your context window with relevant info. websites like x dot com (the everything app) create the sense that you are absorbing important info. but actually most of it is useless.
Unseeing requires spending time to think about problem from various perspective, mentally separate it to pieces, and then realize something like “actually the aspect A of the problem is unrelated to the aspect B; we could simply apply the A-fix, and B is just a red herring”, especially in situations where everyone goes like “I see that this is related to B, and there is a standard B-fix for that, and… uhm, we will need to think about the details later”.
a lot of the time, there will be a strong temptation to just read the summary or the headline, which tells a nice straightforward story. this works maybe half of the time, but the other half of the time there is some really important caveat to the headline, or complication in the story, or huge simplifying assumption that is being made, or even perhaps the entire thing is bullshit. actually knowing this stuff is super important!
a lot of the time, there will be a strong temptation to just read the summary or the headline
Sometimes, you boss will yell at you to stop wasting time when the answer is right there in the headline. If he feels very generous, he might tell you to read a book about the 80:20 rule.
This tracks with what I’ve experienced. Similar to scrolling, I’ve noticed a maladaptive behavior in myself in which, when given a task, I’ll try and have an LLM pick up all the context and do the task, because it feels faster (why absorb all context and give to LLM when LLM can do all?)
This can feel very productive. Spending time absorbing context and building models feels less so, but after some time (usually a day, sometimes much longer if there’s a lot to absorb), these feelings switch: if I left it to an AI I feel lost with whatever it implemented, while if I did the learning I feel rewarded & clear on how to do the task.
I believe the fully general version of this hypothesis is captured by the modern attitude called “maxxing”. I.e. “optimizing something without too much thought or effort”, “compensating some lack of quality with quantity”.
So we can elegantly simplify what you’re saying: a big part of jobmaxxing is just contextmaxxing and modelmaxxing + peoplemaxxing and reasonmaxxing.
hypothesis: a big part of doing a job well is just being willing to spend a huge amount of time absorbing all of the relevant context and building a mental model of the problem. reading stuff, talking to people, looking into the reasoning behind claims that other people just take at face value, looking at related stuff, memorizing the rough order of magnitude of important numbers, etc. if you absorb 10x more context than someone else, then you can be hugely more effective than they are to an extent that looks almost like magic
scrolling is a maladaptive perversion of the instinct to pack your context window with relevant info. websites like x dot com (the everything app) create the sense that you are absorbing important info. but actually most of it is useless.
I think of this as being related to unseeing, even if they aren’t exactly the same.
Unseeing requires spending time to think about problem from various perspective, mentally separate it to pieces, and then realize something like “actually the aspect A of the problem is unrelated to the aspect B; we could simply apply the A-fix, and B is just a red herring”, especially in situations where everyone goes like “I see that this is related to B, and there is a standard B-fix for that, and… uhm, we will need to think about the details later”.
a lot of the time, there will be a strong temptation to just read the summary or the headline, which tells a nice straightforward story. this works maybe half of the time, but the other half of the time there is some really important caveat to the headline, or complication in the story, or huge simplifying assumption that is being made, or even perhaps the entire thing is bullshit. actually knowing this stuff is super important!
Sometimes, you boss will yell at you to stop wasting time when the answer is right there in the headline. If he feels very generous, he might tell you to read a book about the 80:20 rule.
This tracks with what I’ve experienced. Similar to scrolling, I’ve noticed a maladaptive behavior in myself in which, when given a task, I’ll try and have an LLM pick up all the context and do the task, because it feels faster (why absorb all context and give to LLM when LLM can do all?)
This can feel very productive. Spending time absorbing context and building models feels less so, but after some time (usually a day, sometimes much longer if there’s a lot to absorb), these feelings switch: if I left it to an AI I feel lost with whatever it implemented, while if I did the learning I feel rewarded & clear on how to do the task.
I believe the fully general version of this hypothesis is captured by the modern attitude called “maxxing”. I.e. “optimizing something without too much thought or effort”, “compensating some lack of quality with quantity”.
So we can elegantly simplify what you’re saying: a big part of jobmaxxing is just contextmaxxing and modelmaxxing + peoplemaxxing and reasonmaxxing.