Now that I have less time in my hand, I find myself getting more picky with what I learn. I think most of us have some intuitions on what we pick to learn, but I haven’t seen anything that has been codified. This is my rough attempt to make my own intuitions more explicit.
Here I define learning as the act of acquiring knowledge or skills, and I’m keeping it pretty broad. It also encompasses ways of learning that are informal (checking direct messages or social media), reflective (reading fiction) as well as entertaining (watching a series).
I think there are some benefits in knowing how to prioritise learning well:
My guess is that our intuitions are generally good enough, but if you’re feeling frustrated or indecisive when making a choice to learn, this could be helpful.
There are a lot of things to learn about EA and it’s helpful to pick the ones that are most impactful.
In general, you want to prioritise learning something when they are:
Relevant to your goal
Provides high information value (i.e. reduce a lot of uncertainty)
Might make you pivot to a new goal
Trustworthy
Urgent
Highly motivating or exciting for you
Easy and not time consuming
Highly likely to be achievable
Factor
Example
Goal relevance
How is learning about it relevant in achieving your goals?
If you have a high priority goal to complete a research project on insect welfare, learning about insect biology is more goal-relevant than learning about painting.
Information value
How much uncertainty can you reduce by learning about it?
You are deciding between shipping feature A or feature B for your product. If you’re still feeling uncertain after testing A with a group of users, it might be worth testing B too.
Note: It might seem that this factor is related to “goal relevance”. However, you can technically learn something with a high value of information (reduce a lot of uncertainty), but it doesn’t help with your goals.
Pivot potential
How learning about it could potentially change your goals? What are your riskiest assumptions?
You are hoping to travel to a conference in a few months, but might face travel restrictions. You set a weekly Google Alert for this. You might not change your goals immediately, but there might be a non-trivial chance you need to.
Note: Technically you could have a meta-goal for prioritising goals, but I’m putting this here just in case.
Trust level
Before engaging with the material, how much can you trust the material based on prior information you have about it (e.g. author, platform, etc)?
The prime minister in your country is hoping to make a televised announcement regarding the pandemic. They seem to have a track record of misleading people, but the ministry of health has a good track record. There are some incentives in place that make lying costly, so it’s somewhat more likely that this coming announcement is trustworthy.
Urgency
How helpful is learning about it now versus later?
If you know you’ll be outdoors in a few hours, it’s urgent to learn about the day’s weather.
Motivation
How motivated are you when it comes to learning about this?
You might feel more curious or excited when it comes to predicting AI take-off speeds, so you’re pretty motivated already.
Challenge
How much time and effort do you need to learn this? How challenging is the material?
You might feel naturally motivated, but reading about this specific article on AI take-off speeds requires a lot of time and focus.
Probability of success
How likely will you complete learning this?
You have three children who are currently sick, and require a lot of your attention. You think it’s not likely you’ll complete learning this until they’re better.
Note: “probability of success” seems like a function of “motivation” and “challenge”. Again, just putting it here so it’s explicit.
Sometimes you can’t avoid learning something despite scoring badly on some of these factors, especially for “motivation”, “challenge”, and “probability of success”. Here are some ways that can help:
If you are reading a tough-to-understand book, read summaries and reviews of the book instead. It’s possible that this is a more cost-effective way to learn instead of reading the entire book.[1][2][3]
If building up a habit of consuming daily news is tedious, read newsletters that give 5 minute summaries. Small accessible habits help improve consistency.
If you’re new to a topic, a good rule of thumb is to start with a breadth-first approach. There are larger diminishing returns when learning something in-depth.
If you’re testing a new feature/product but want to gain insights that are trustworthy, test it cheaply first. There is some trade-off between “trust level” and “challenge”, but it’s more likely than not that a cheap test could be pretty insightful enough.
If you want to increase the trustworthiness of what you learn, writing seems like a good thing to do too.[4][5]
If you have a tough final exam to take in a few months and you want to increase your chances of success, there are pretty well-researched study principles (e.g. space repetition, chunking, etc) that are helpful. [6]
I’m looking to improve how this decision making tool works. If you have any suggestions, please let me know!
There’s too much to learn
Now that I have less time in my hand, I find myself getting more picky with what I learn. I think most of us have some intuitions on what we pick to learn, but I haven’t seen anything that has been codified. This is my rough attempt to make my own intuitions more explicit.
Here I define learning as the act of acquiring knowledge or skills, and I’m keeping it pretty broad. It also encompasses ways of learning that are informal (checking direct messages or social media), reflective (reading fiction) as well as entertaining (watching a series).
I think there are some benefits in knowing how to prioritise learning well:
My guess is that our intuitions are generally good enough, but if you’re feeling frustrated or indecisive when making a choice to learn, this could be helpful.
There are a lot of things to learn about EA and it’s helpful to pick the ones that are most impactful.
In general, you want to prioritise learning something when they are:
Relevant to your goal
Provides high information value (i.e. reduce a lot of uncertainty)
Might make you pivot to a new goal
Trustworthy
Urgent
Highly motivating or exciting for you
Easy and not time consuming
Highly likely to be achievable
Goal relevance
How is learning about it relevant in achieving your goals?
Information value
How much uncertainty can you reduce by learning about it?
You are deciding between shipping feature A or feature B for your product. If you’re still feeling uncertain after testing A with a group of users, it might be worth testing B too.
Note: It might seem that this factor is related to “goal relevance”. However, you can technically learn something with a high value of information (reduce a lot of uncertainty), but it doesn’t help with your goals.
Pivot potential
How learning about it could potentially change your goals? What are your riskiest assumptions?
You are hoping to travel to a conference in a few months, but might face travel restrictions. You set a weekly Google Alert for this. You might not change your goals immediately, but there might be a non-trivial chance you need to.
Note: Technically you could have a meta-goal for prioritising goals, but I’m putting this here just in case.
Trust level
Before engaging with the material, how much can you trust the material based on prior information you have about it (e.g. author, platform, etc)?
Urgency
How helpful is learning about it now versus later?
Motivation
How motivated are you when it comes to learning about this?
Challenge
How much time and effort do you need to learn this? How challenging is the material?
Probability of success
How likely will you complete learning this?
You have three children who are currently sick, and require a lot of your attention. You think it’s not likely you’ll complete learning this until they’re better.
Note: “probability of success” seems like a function of “motivation” and “challenge”. Again, just putting it here so it’s explicit.
Sometimes you can’t avoid learning something despite scoring badly on some of these factors, especially for “motivation”, “challenge”, and “probability of success”. Here are some ways that can help:
If you are reading a tough-to-understand book, read summaries and reviews of the book instead. It’s possible that this is a more cost-effective way to learn instead of reading the entire book.[1][2][3]
If building up a habit of consuming daily news is tedious, read newsletters that give 5 minute summaries. Small accessible habits help improve consistency.
If you’re new to a topic, a good rule of thumb is to start with a breadth-first approach. There are larger diminishing returns when learning something in-depth.
If you’re testing a new feature/product but want to gain insights that are trustworthy, test it cheaply first. There is some trade-off between “trust level” and “challenge”, but it’s more likely than not that a cheap test could be pretty insightful enough.
If you want to increase the trustworthiness of what you learn, writing seems like a good thing to do too.[4][5]
If you have a tough final exam to take in a few months and you want to increase your chances of success, there are pretty well-researched study principles (e.g. space repetition, chunking, etc) that are helpful. [6]
I’m looking to improve how this decision making tool works. If you have any suggestions, please let me know!
https://www.cold-takes.com/reading-books-vs-engaging-with-them/
https://www.cold-takes.com/honesty-about-reading/
https://www.lesswrong.com/posts/72QA8qk9g6wZNDWeS/on-not-reading-papers
http://www.paulgraham.com/words.htm
https://www.cold-takes.com/learning-by-writing/
https://www.coursera.org/learn/learning-how-to-learn