Ask HN: Are you too getting addicted to the dev workflow of coding with agents?

41 points by gchamonlive ↗ HN
It's becoming an extremely dopaminergic work loop where I define roughly the scope of my task and meticulously explore and divide the problem space into smaller chunks, then iterating over them with the agent. Rinse and repeat.

Each execution prompt after a long planning session feels like opening a lootbox when I used to play Counter Strike.

It's really fun to code like that, it's like riding a bike after a lifetime of only knowing how to run. But I'm really wary that's addictive for me. Wonder if there are more people here that feel like this too.

45 comments

[ 3.0 ms ] story [ 69.0 ms ] thread
> Each execution prompt after a long planning session feels like opening a lootbox when I used to play Counter Strike.

The "uncertain reward" nature of LLM usage makes it a skinner box, yes.

Unironically, the descendant of Claude Code is the metaverse/holodeck/next minecraft.

It will look nothing like those things, but it will be obvious in retrospect.

For better and worse.

I've heard similar things from many people know, but I don't feel like this at all. I don't find coding with Claude any more or less addictive than without. I do find coding with claude slightly more fun, but mostly because brainstorming with someone/something feels less lonely than writing code alone. I wonder where the discrepancy comes from.

Seeing the final result of a feature doesn't really give me any dopamine. Maybe because I'm mostly working on projects I know how to do. When I give it a prompt I already know what the result should look like, so I'm not really surprised by anything it produces.

I'm also in this situation where I mostly work on stuff I know how to do manually.

For me it's always magic to see it work, even if it's a tiny change I'd need to do. To be able to ask "add an opt-in flag for this part of the script" and see it work, updating documentation and asking follow-up questions when instructions are vague it's impressive...

(comment deleted)
I give less shits seeing how sloppy the quality bar is now
(comment deleted)
I got somewhat addicted to the planning phase to the point I started getting task paralysis because I was hell bent on creating the perfect plan.

Everything can be optimized, performance can be improved, you can always think of more edge cases and user stories to cover everything, but after a point that just becomes procrastination in the form of chasing perfection. It's also hell if you've got even the slightest bit of ADHD, rapidly leading to task paralysis with the sheer scale of the plan.

Now I sit with a notebook sketch out everything I am thinking about and then condense it to a planning prompt and then once the plan aligns with my representation of the task, I start implementing.

I totally get where you're coming from with the planning loop. It can be a real trap because it always feels like you're making progress. The notebook idea is actually pretty clever. It lets you chill for a bit before you start involving the agent.
The problem is Claude often kinda grinds to a halt. So I find myself getting context switching a LOT as I’m waiting for it to do something. I’m alright at context switching, but it’s really fucking tiring doing it so constantly.
I had two days straight of coding sessions with codex and I haven't seen a single hitch in the weekend. With Gemini though, which I use for work, I get this exact problem. It's tiring and unproductive if the feedback loop is too short from submitting a prompt and seeing the results
I’ve always been addicted to coding and building, this just makes it easier to get my fix…
(comment deleted)
Yes, sometimes I have found it hard to sleep if I'm close to building something I want to build.

I think there is kind of a meme going around about multitaskers doing very well with vibe coding, and I can see it. Although, as someone who has the opposite problem, it can be tiring if I try to do more than two things at once.

I tried it once, to do two tasks at once, and it was very confusing, having to juggle two contexts, even though you have a bit of idle time waiting for the agent to finish his turn. I haven't experimented with subagents, but I prefer to interact with a single agent at a time so I can give the output more attention.
whenever i give them a chance, i usually give them a codebase I've been working on for a while (so most bugs have been fixed), after I found a new bug, and ask them to "find all bugs".

They always end up praising me for the high quality code and howdthey found exactly ZERO manifest bugs in the code, and this must be the work of a skilled senior developer owing to the code's polish.

Then I point out the bug I had just discovered.... "you're exactly right!"

Not really. I use claude at work and I wouldn’t use at home because a) the same plan would cost $200/month… and I don’t enjoy paying that amount of money for a toy, b) it’s not open source, so I cannot trust antrohpic to give them access to my stuff
I also love it. Finally, I am no longer constrained by syntax errors or forgotten API details. I can focus on the feature. It's like taking programming to a higher level - programming in English (instead of Java).
That's it for me too. I'm churning through a multi-year backlog and can finally implement my ideas for which there just wasn't time before.
I feel this sneaking up on me. I've only recently allowed Claude to actually edit some files directly, rather than just show me suggested edits. It could certainly be addictive to just hit enter while code magically appears, thinking "oh yeah, I totally would have done it like that".
No worries. You are in the right place. This is how others feel as well and how software engineering will feel for new generations so yeah the bicycle comparison fits well.
the dopamine hits are real. being an ex addict i guess for me its a turn off because i know this is basically the same thing (for me). i dont mind using AI, but i ended up cancelling my subscriptions because it touch a bad memory for me. I'd advise people caution. Like anything that hits dopamine up frequently, your mind adapts quick to expect and 'need' such hits.

its very personal if its good or bad i suppose. (not a psychologist so honestly dont know if its really similar. just expressing my personal feeling about it)

I don't think so, at least for me manual development was the same, giving a lot of dopamine boosts. The only difference is that I get the same amount of dopamine when I generate a full page instead of a single component. Nonetheless it can be a real issue because it takes away from the enjoyment of working on details in the codebase.
What’s the negative here that makes this an addiction and not just good?
I had to actively force myself to leave home because I was getting consumed by it in the weekend. "Just another fix", "just this one last polish"...

It's MMO all over again. And I'm not complaining, but I want to raise awareness for me and other that maybe this isn't such an innocent activity as we think. Coding with agents might just be too addictive for some.

Nope. Pairing with Claude Code makes programming smoother for me and makes it easier and faster to check different solutions (i.e. "sketch out this approach and let's see how that works out"), but there are no lootbox moments. The LLM does exactly what I tell it in pretty small chunks.

How I work today is still very similar to how I worked in 2023, but now I'm typing a TON of English and very little Ruby. But the overall vibe is nicer and starting on difficult stuff is significantly easier.

I had lots of woah moments where I'd plan changes and the agent would actually find inconsistencies in my understanding and complexities I wasn't considering. It's also been very good on keeping surrounding documentation and other code references updates. Maybe in time I'll get used to it and it'll just be another something I'll take for granted, but right now it feels just like gambling for me LOL
Yes, for the last few months I rather watch LLM generate code than to play video games