Feb 14 Retrospective

I pose myself several questions every week (not having to answer all of them):

• how to improve?
• any open questions?
• any small project ideas?
• introspect: what are my feelings?

What are my feelings?

Both weeks, I’ve slipped behind in completing the full technical scope of work as laid out by the instructors. It’s an open question whether I should be learning literally everything (see: CSS), but I’d rather ask the question of: how can I learn deeper? There will always be scheduling pressures; I would like to learn a lot even despite speed.

Physically, I feel fantastic. At the beginning of this course, I set out to treat myself like an athlete. Allow me to brag a bit: I’ve been consistently sleeping 8-9 hours every night, getting reasonably healthy food, and squeezing in short calisthenics sessions in the mornings. I just got a bike, so I will soon be getting cardio in the morning as well as saving time on the commute. I take Sundays completely off from work; I use them to reset.

I’m doing my best to take this experience seriously. I’m proud of how I’ve managed my body and mind so far, mostly inspired by Daniel Gross’s talk on How to Win.

Retrospective on last week’s goals

Use Claude in “Feynman Mode”: FAIL

• Constantly felt pressure to write more code; largely didn’t reach for this tool or any sort of reflective understanding. Should probably build in structured time to review understanding (see improvements section).

Notice animal-like panic; correct: SUCCESS-ish?

• Getting better at this. This is 2 skills: noticing the panic, and applying corrective measures. Skill 2 is mostly about going for a short walk and skill 1 is mostly just practice.

Break work into smaller pieces: SUCCESS

• Planning out the necessary steps to implement a feature before starting to code has been a huge upgrade to my process. This forces better feature understanding too.
• Make more branches! Commit more often!
• One major change to my process for adding a new feature has been to add pieces frontend to backend instead of vice versa. I’ve found it’s currently easier to test visually, as clicking a button can easily output a message. I struggle to test backend endpoints. This is probably just a skill issue, as I can use curl to perform a simple endpoint test.

Abundance mindset: HARD TO EVALUATE

• Dumb of me to set something as a goal that’s hard to evaluate.

Socialize more: SUCCESS

• Huge success. Will keep this up. Feels good to get to know everyone.

How do I improve?

• Learn loop: I’ve been unimpressed with my learning speed AND with the amount of time I’ve been spending to ensure I understand the material. I’m adding a new learning loop: Plan → MVP → Learn → Optimize. I think Claude can help me 1. plan, 2. accelerate the MVP, and 3. more deeply understand the material, probably with Feynman technique. My rough estimate is to implement each day’s material as simply as possible and then spend at least 30 minutes recapping the material.
• Rationalist skills: Mark gave a great talk this past week about problem-solving skills. I most appreciated the slide where he listed out a bucket of possible techniques to apply. The rationalists are on to something! Thinking about thinking is useful! I’m gonna make a Claude skill before the lecture today that invokes a random one after every question I ask. Randomization is a time-tested strategy, like burning bones and using the cracks to select where to hunt caribou for the day.