Artificial Intelligence
-
AI Changes Everything - Armin Ronacher
Armin seems very AI-positive. He now uses Claude Code and finds it significantly increases productivity by alternating between giving instructions and reviewing work. His optimistic outlook sees AI enhancing human capabilities rather than replacing people, potentially democratizing knowledge and accelerating innovation.
While I can agree that AI is likely an irreversible change, I don’t share his view that it won’t replace people or that the job of programmer or designer won’t go away. I think that largely using AI to write code will change the job of a programmer completely, and the skills required will be completely different.
-
GenAI Criticism and Moral Quandaries - Armin Ronacher
This is a follow-up post by Armin in in response to a post by Glyph. While Glyph is skeptical, Armin sees significant potential AI code generation when used properly. Armin seems to think that code review with AI is enjoyable and programming with AI assistance is still satisfying but I’m not so sure. I think programming will be totally different and we’ll see a lot of churn with programmers as many who don’t enjoy the new paradigm switch out to other jobs.
-
Using Claude Code to Migrate Wyvern Management Scripts - Steve Yegge, Gene Kim
Steve Yegge and Gene Kim have a vibe-coding sessions of Steve’s game Wyvernn. They use Claude to migrate the game’s management scripts to Kotlin.
While I think that many folks will feel like they are in control during vibe-coding sessions, much like folks using auto-pilot mode in a Tesla, folks won’t be paying nearly as much attention as they think they are.
During the video, Gene says “I assume you are being vigilant about what change is being made?” and Steve responds “Oh yeah, I know my code really well… I am reading all this code as we talk” like he understands everything that’s going on, but then like a second later Claude asks if it should run the
kwyverncommand that just wrote and he’s like “do you want to proceed? Yeeaaah, I don’t know what this means but sure!”. But Steve! I thought you were reading all that code!
Programming
-
AI-assisted coding for teams that can’t get away with vibes - Atharva Raykar
Atharva provides a lot of helpful and practical tips on how to effectively use AI.
The section on “Implications on how AI changes craft” particularly stood out to me. He notes that rewrites are now very cheap, and writing abstractions is now a lot less important. While, many programmers already overvalue abstractions to begin with, I think these kind of changes will result in code that is much less readable for humans.
-
How I program with Agents - David Crawshaw
David Crawshaw (ex(?)-CTO of Tailscale) discusses his approach to programming with AI agents. He provides a few examples of how he’s used them in the past. One example that stood out was where the AI agent got confused at how used a non-standard way of encoding JSON in PostgreSQL database tables. Their solution was to add comments to the code to help the AI agent but that programmers would likely skip over. I feel like we’ll see a lot more code that includes instructions for AI agents in the future, which will make comments even more likely to be ignored by actual human programmers.
Personal Development
-
25 things I know at 25 - Atharva Raykar
Atharva is the same author of article AI-assisted coding for teams that can’t get away with vibes listed above.
This is a really great list of life lessons for someone still young enough to be this optimistic. I find myself to be a lot more cynical these days. I wish I was this self-aware when I was 25. I may not be now.
Some of the lessons I wish I was better at:
- When trying to fix your health, start with fixing your sleep: I think I need a lot more sleep than is really practical. I’ve been working on this for a while but it remains elusive.
- Never ever read the news. Pick up a history book instead: Since COVID, I’ve found it impossible to avoid the news. TBQH, It really feels like everything is burning down all around us.
- The world is a malleable place for those who can think in systems. Learning to think in these terms will turn you into an optimist, whether you like it or not: I really wish this was true for me. I think it’s entirely possible for observed metrics (the aspects of life best improved by systems) to get better, but lived experience to get worse.