I want to run a writing group based on Holden Karnofsky’s Learning By Writing approach. The idea is to personally do in-depth investigations into questions you care about. I understand this as roughly reading less than you think you should and writing more than you think you can.

I’m not yet certain exactly how this should look. My first guess would be to have a ~1 hour group call once a week for 4 weeks. Each group member picks a question they want to work on over the 4 weeks. Meetings are spent giving feedback, debugging, discussing etc. 

I think Holden’s approach is powerful because you form hypotheses early and iterate on them. This helps combat the trap of continually gathering information while not really making progress

Here’s an outline:

Step 1 Pick a topic
Step 2Read and/or discuss with others (a bit)
Step 3Explain and defend my current, incredibly premature hypothesis, in writing (or conversation)
Step 4Find and list weaknesses in my case
Step 5Pick a subquestion and do more reading/discussing
Step 6Revise my claim/switch sides
Step 7Repeat steps 3-6 a bunch
Step 8Get feedback on a draft from others, and use this to keep repeating steps 3-6

I (Jordan) am personally excited to write about AI threat models regarding existential risks. I’d love for you to join me if you’re interested in similar things, though I welcome a variety of topics. Depending on the numbers, I may encourage people to run separate groups.

If you’d like to try this with me, please fill out this form

This will be most relevant for people who have in-depth knowledge of EA, x-risk, rationality etc. already.




Sorted by Click to highlight new comments since: Today at 4:23 AM

I really love this idea - although I don’t have time to do this right now, I want to express my support of it