friggeri

I stopped drinking 8 or 9 years ago, and I haven’t missed it. Not that it ever was a problem, I enjoyed a glass of wine or a cocktail every now and then, but I was training for an ultra marathon and figured I’d stay dry while I trained for a couple of months. Had a glass of wine a few days after the race and felt completely hammered, at which point I realized if I wanted to drink again I’d have to “train” my body for it, and thought that was totally backwards. I haven’t found any situation where an alcoholic drink can’t be replaced by a glass of water, an NA beer or a mocktail. I’m also very lucky to be married to a wonderful woman who arrived to more or less the same conclusion and we enjoy being the boring ones who don’t drink (but never get a hangover).

I’ve been training an alphazero style model for an abstract strategy game I created 20 years ago. It’s been really fun learning about MCTS and figuring out how to optimize all parts of the pipeline to be able to train on ~millions of moves for ~hundreds of dollars.

I grew up in France and started smoking before age 15, and was a heavy smoker (1 pack a day) until I quit when I was 27 after moving to the US. Getting out of an environment where smoking was normalized is what helped me take that step.

Smoking is terrible in that it really messes with your brain, you intellectually know all the reasons why you should quit but it’s so easy to put it off. “I can quit whenever, I just don’t want to”. Until you try and get that gnawing yearning feeling.

It took me a solid year to not feel that way all the time. And another two to not feel that when having a coffee/glass of wine/standing next to someone else smoking.

13 years later I’m completely over it, to the point where I can barely tolerate being in the vicinity of someone smoking. And yet I often have dreams where I light a cigarette.

If you’re thinking of quitting, there is no better way than today. It’s likely going to be hard, but it’s one of the best things you can do to yourself. You already know that of course, but hopefully this helps.

Looking at the prompts op has shared, I'd recommend more aggressively managing/trimming the context. In general you don't give the agent a new task without /clearing the context before. This will enable the agent to be more focused on the new task, and decrease its bias (if eg. reviewing changes it has made previously).

The overall approach I now have for medium sized task is roughly:

- Ask the agent to research a particular area of the codebase that is relevant to the task at hand, listing all relevant/important files, functions, and putting all of this in a "research.md" markdown file.

- Clear the context window

- Ask the agent to put together a project plan, informed by the previously generated markdown file. Store that project plan in a new "project.md" markdown file. Depending on complexity I'll generally do multiple revs of this.

- Clear the context window

- Ask the agent to create a step by step implementation plan, leveraging the previously generated research & project files, put that in a plan.md file.

- Clear the context window

- While there are unfinished steps in plan.md:

-- While the current step needs more work

--- Ask the agent to work on the current step

--- Clear the context window

--- Ask the agent to review the changes

--- Clear the context window

-- Ask the agent to update the plan with their changes and make a commit

-- Clear the context window

I also recommend to have specialized sub agents for each of those phases (research, architecture, planning, implementation, review). Less so in terms of telling the agent what to do, but as a way to add guardrails and structure to the way they synthesize/serialize back to markdown.

In the last 50 years, software has morphed from a hobbyist pursuit, to a nerdy subculture, to a trillion+ dollars industry. This has caused a pretty significant mix shift in the software developer community: the reason driving the mean developer into this field in 2025 is very different from that of the 2015 developer, and that of the 2005 developer.

Arguably there might be more curious tinkerers nowadays, but they might represent a smaller slice of the pie.