Sunny

← All weeks

What I’ve been reading (W13, 2025)

When people give you feedback (notes) on something · We get distracted because working on things we care about is often scary

–

  1. When people give you feedback (notes) on something, when they tell you what’s wrong, they’re usually right. But when they tell you how to fix it, they’re wrong https://www.youtube.com/shorts/NHHZSNw9J2o
  2. We get distracted because working on things we care about is often scary. It makes us face our limits. Maybe we’re not as good as we hoped, or things won’t turn out how we want. When you’re trying to write something difficult or do something meaningful, there’s no promise it will go well. That uncertainty can feel overwhelming. But the internet feels wide open. It lets you move from one thing to the next without pressure or risk. It gives you a break from reality, and that’s a huge relief https://www.theatlantic.com/family/archive/2021/08/oliver-burkeman-advice-time-productivity/619723/
  3. Addiction isn’t always about the usual vices like gambling, porn, alcohol, social media or video games. At its core, it’s a deep attachment that throws you off balance. Sometimes, it hides in positive things, like working hard, staying disciplined or chasing a creative goal. Instead of listening to all your parts and attending to their various needs, you start overindulging one or two at the expense of everything else https://elainewrites.substack.com/p/isolation
  4. Lawyers, doctors, and architects in small towns and big cities share a profession but live different day-to-day lives shaped by their goals and problems. No one expects the small-town lawyer to start using the same tools or follow the same processes as someone at a big firm in New York. But in software, developers often talk about what they’re doing without saying anything about their own context, and others read it and think, we need to do what they’re doing, as if their situations were the same https://thorstenball.com/blog/2020/09/15/the-context-in-which-we-build-software
  5. If the LLM keeps doing things you don’t like, you need to change the vibe it’s picking up. That could mean tweaking the prompt or rules you’re giving but also changing how your code is written. Since the LLM tries to guess what comes next based on what it sees, how your code is written matters. The model’s training, your prompt, and your code together shape how it behaves. You can’t change the training, and your code has a much more significant influence than the prompt, so it’ll have the biggest impact https://ezyang.github.io/ai-blindspots/culture-eats-strategy