Eighty/Twenty

.plan for Gordon Weakliem

View on GitHub
8 January 2025

Getting to Done

by Gordon Weakliem

I ran into Using LLMs and Cursor to become a finisher the other day (this one should have a via link, but now I don’t remember where I saw it). I made a kind of resolution this fall to start building more stuff, well before I read this, but Zohaib has a good reminder. Software Engineers are builders by nature, but we frequently end up in roles that aren’t really about building: either you get into management or management-adjacent roles, or your day-to-day is maintenance, bug fixes, and operations. That’s a problem career-wise, but really it’s not good for the soul. If you’re in software, you’re probably a builder, and those kind of roles destroy you from the inside.

Because my time has always been limited, progress on side projects had been slow in the past, and many remained unfinished as life’s events caused a loss of momentum, making them harder to resume.

That’s always been a massive problem for me - writing the code is the easy part, but the hassle around finishing has always been the problem. Luckily, it’s easier than ever. I’ve been consistently surprised and delighted at how I can overcome obstacles building on the knowledge that’s out there. There’s a million problems out there that are not addressed in a way I’d like (or are covered in ads and popups asking you to enable notifications, subscribe to a newsletter, etc). Every little irritation in daily life is a call to your inner builder to build something.

I like Zohaib’s prompting method. When I started my utilities site, I was the micro-managing Product Owner. Claude interestingly managed me back. I gave it a list of a half-dozen little utilities that I wanted to make, and Claude implemented only one of them, with a framework for expansion that was easy for me to grok. Zohaib seems to be working as more of a conversation, and I think that works better. About six months ago, I was talking to someone about prompting and he gave me a kind of long-winded exposition. I summarized his point as “don’t micromanage the LLM” and he said “Yeah! That’s exactly it!”. It’s simple advice but it’s hard for me to resist the urge to barf out a fully-realized product spec and have a conversation instead. Working with humans and LLMs present some of the same challenges.

Zohaib also has some great tips on using Cursor effectively, some of these I didn’t know and others are just good habits that I needed to be reminded of.

Have a good day. I’m off to go build something.

tags: