The incredible thing (to me) is that this isn’t even remotely a new thing: it’s reviewing pull requests vs writing your own code. We all know how different that feels!

I'm the same way. LLMs are still somewhat useful as a way to start a greenfield project, or as a very hyper-custom google search to have it explain something to me exactly how I'd like it explained, or generate examples hyper-tuned for the problem at hand, but that's hardly as transformative or revolutionary as everyone is making Claude Code out to be. I loathe the tone these things take with me and hate how much extra bullshit I didn't ask for they always add to the output.

When I do have it one-shot a complete problem, I never copy paste from it. I type it all out myself. I didn't pay hundreds of dollars for a mechanical keyboard, tuned to make every keypress a joy, to push code around with a fucking mouse.

Three things I can suggest to try, having struggled with something similiar:

1. Look at it as a completely different discipline, dont consider it leverage for coding - it's it's own thing.

2. Try using it on something you just want to exist, not something you want to build or are interested in understanding.

3. Make the "jumps" smaller. Don't oneshot the project. Do the thinking yourself, and treat it as a junior programmer: "Let's now add react components for the profile section and mount them. Dont wire them up yet" instead of "Build the profile section". This also helps finding the right speed so that you can keep up with what's happening in the codebase

> For me there’s no flow state possible with LLM “coding”.

I would argue that it's the same question as whether it's possible to get into a flow state when being the "navigator" in a pair-programming session. I feel you and agree that it's not quite the same flow state as typing the code yourself, but when a session with a human programmer or Claude Code is going well for me, I am definitely in something quite close to flow myself, and I can spend hours in the back and forth. But as others in this thread said, it's about the size of the tasks you give it.

I have both; for embedded and backend I prefer entering code; once in the flow, I produce results faster and feel more confident everything is correct. for frontend (except games), i find everything annoying and a waste of time manually, as do all my colleagues. LLMs really made this excellent for our team and myself. I like doing UX, but I like drawing it with a pen and paper and then do experiments with controls/components until it works. This is now all super fast (I usually can just take photo of my drawings and claude makes it work) and we get excellent end results that clients love.

I could imagine a world where LLM coding was fun. It would sound like "imagine a game, like Galaxians but using tractor trailers, and as a first person shooter." And it pumps out a draft and you say, "No, let's try it again with an army of bagpipers."

In other words, getting to be the "ideas guy", but without sounding like a dipstick who can't do anything.

I don't think we're anywhere near that point yet. Instead we're at the same point where we are with self-driving: not doing anything but on constant alert.

I feel the same way often but I find it to be very similar to coding. Whether coding or prompting when I’m doing rote, boring work I find it tedious. When I am solving a hard problem or designing something interesting I am engaged.

My app is fairly mature with well established patterns, etc. When I’m adding “just CRUD” as part of a feature it’s very tedious to prompt agents, reviewing code, rinse & repeat. Were I actually writing the code by hand I would probably be less productive and just as bored/unsatisfied.

I spent a decent amount of time today designing a very robust bulk upload API (compliance fintech, lots of considerations to be had) for customers who can’t do a batch job. When it was finished I was very pleased with the result and had performance tests and everything.

You're not alone. I definitely feel like this is / will be a major adaptation required for software engineers going forward. I don't have any solutions to offer you - but I will say that the state that's enabled by fast feedback loops wasn't always the case. For most of my career build times were much, much longer than they are today, as an example. We had to work around that to maintain flow, and we'll have to work around this, now.

This.

To me, using an LLMs is more like having a team of ghostwriters writing your novel. Sure, you "built" your novel but it feels entirely different to writing it yourself.

I feel differently! My background isn't programming, so I frequently feel inhibited by coding. I've used it for over a decade but always as a secondary tool. Its fun for me to have a line of reasoning, and be able to toy with and analyze a series of questions faster than I used to be able to.

Well are you the super developer than never run into issues, challenges? For me and I think most developers, coding is like a continuous stream of problems you need to solve. For me a LLM is very useful, because I can now develop much faster. Don't have to think which sorting algoritm should be used or which trigonometric function I need for a specific case. My LLM buddy solves most of those issues.

I like writing. I hate editing.

Coding with an LLM seems like it’s often more editing in service of less writing.

I get this is a very simplistic way of looking at it and when done right it can produce solutions, even novel solutions, that maybe you wouldn’t have on your own. Or maybe it speeds up a part of the writing that is otherwise slow and painful. But I don’t know, as somebody who doesn’t really code every time I hear people talk about it that’s what it sounds like to me.

[dead]