I actually like a lot of these. I wish the forum had collapsible sections or bullet points so that curious readers could expand sections we wish to learn more about.
Extremify
I do some version of this a lot and call it “limiting-case analysis”. The idea is that you want to find the most generalisable aspects of the pattern you’re analysing, and that often means setting some variables to 0 or ∞ or even removing them altogether.
“The art of doing mathematics consists in finding that special case which contains all the germs of generality.”—David Hilbert
Notice which aspects are parameters. If they’re set to some specific value, generalise them by turning them into variables instead.
Play the generalised game of whatever you’re trying to do. Chess boards with an infinite number or unconstrained variable of squares are more general than 8x8.
Social proof
This is so crucial for motivation. If we’re worried about how our ideas will be received, our brains will refuse to innovate.
However, I would caution against the need for social validation. If you rely on others to check your ideas for you, you’ll have less incentive to try to check them yourself. At the start, it may just be true that others are better at judging your ideas than you are, but I’d still recommend trusting your own judgment because it won’t get practice otherwise.
“I was used to just having everybody else being wrong and obviously wrong, and I think that’s important. I think you need the faith in science to be willing to work on stuff just because it’s obviously right even though everybody else says it’s nonsense.”—Geoffrey Hinton
Plus, you need to be able to use your ideas to reach more distant nodes on the search tree without slowing down for validation at every step. It’s better to have the option to internalise the entire process. This gives you a much shorter feedback loop that gets better over time.
YourBias.Is
You advise me to “familiarise myself with common cognitive biases” so that I can learn to avoid them. I agree of course, but I think there’s important nuance. If you defer to empirical experiments and solid statistics to form your beliefs about how you’re biased, you may learn to be statistically without understanding why you’re correct.
Imo, the reason you should consult literature and statistics about biases is mostly just so that you can learn to recognise how they work internally via introspection. I think that’s the only realistic way to learn to mitigate them.
I actually like a lot of these. I wish the forum had collapsible sections or bullet points so that curious readers could expand sections we wish to learn more about.
Extremify
I do some version of this a lot and call it “limiting-case analysis”. The idea is that you want to find the most generalisable aspects of the pattern you’re analysing, and that often means setting some variables to 0 or ∞ or even removing them altogether.
“The art of doing mathematics consists in finding that special case which contains all the germs of generality.”—David Hilbert
Notice which aspects are parameters. If they’re set to some specific value, generalise them by turning them into variables instead.
Play the generalised game of whatever you’re trying to do. Chess boards with an infinite number or unconstrained variable of squares are more general than 8x8.
Social proof
This is so crucial for motivation. If we’re worried about how our ideas will be received, our brains will refuse to innovate.
However, I would caution against the need for social validation. If you rely on others to check your ideas for you, you’ll have less incentive to try to check them yourself. At the start, it may just be true that others are better at judging your ideas than you are, but I’d still recommend trusting your own judgment because it won’t get practice otherwise.
“I was used to just having everybody else being wrong and obviously wrong, and I think that’s important. I think you need the faith in science to be willing to work on stuff just because it’s obviously right even though everybody else says it’s nonsense.”—Geoffrey Hinton
Plus, you need to be able to use your ideas to reach more distant nodes on the search tree without slowing down for validation at every step. It’s better to have the option to internalise the entire process. This gives you a much shorter feedback loop that gets better over time.
YourBias.Is
You advise me to “familiarise myself with common cognitive biases” so that I can learn to avoid them. I agree of course, but I think there’s important nuance. If you defer to empirical experiments and solid statistics to form your beliefs about how you’re biased, you may learn to be statistically without understanding why you’re correct.
Imo, the reason you should consult literature and statistics about biases is mostly just so that you can learn to recognise how they work internally via introspection. I think that’s the only realistic way to learn to mitigate them.