Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Agentic Coding Is a Trap

larsfaye.com

Post reply on HN
Print view View original

131–140 of 403 posts

Re: Agentic Coding Is a Trap

#131

Post by mempko »

Writing code is not the hard part of software development. This is coming from someone who has programmed for 30 years, writing an average of 100k+ lines a year.

The sooner programmers start thinking about modeling the domain, user mental models, architecture and data structures and less focus on the mechanics of writing code, the better.

Writing code is the EASY part. LLMs have basically solved the easiest part of software development. They however are bad at all the stuff I mentioned. LLMs don't have a point of view, you do as a software developer.

Re: Agentic Coding Is a Trap

#132

Post by dilyevsky »

kelnos wrote: post #64
ex-aws-dude wrote: post #9

The thing is the code quality is still ultimately up to you Nothing stopping you from iterating with the agent till the code is the exact same quality that you yourself would write

IME, it's faster and less frustrating to just write the code myself, if the goal is to get code to my quality standards.

Respectfully, unless someone is really really bad at articulating what the quality standards are or works with a very niche stack that is definitely not the case anymore with SOTA models

Re: Agentic Coding Is a Trap

#133

Post by throwaway041207 »

byzantinegene wrote: post #116
fnordpiglet wrote: post #51

Interestingly I’ve learned more about languages and systems and tools I use in the last few years working with agentic coding than I did in 35 years of artisanal programming. I am still vastly superior at making decisions about systems and techniques and approaches than the agentic tools, but they are like a really really well read intern who knows a great deal of detail about errata but have very little experience.…

you have 35 years of experience and have already built up the learning capability and general framework to acquire new knowledge. you know how to use agentic coding as a tool to supplement your work. the juniors who start today don't have that, they overrely on agentic coding and do not know what they don't know

IMO, by the time todays juniors would have 5-10 years of expected experience, the entire field will be something different altogether. Language choice distribution will collapse (if not change altogether), whole new modalities of monitoring and progressive delivery guardrails will come into play, essentially creating a 24/7 incremental rollout of pure agentic code, correctness will be determined by a mix of language features and self-monitoring by models in production and automated testing against production snapshots in pre-production, and deep debugging will the be province of a select group of engineers and there will be a pathway to those roles for juniors, but those roles will be coveted and difficult to break into (and probably will require education and maybe even informal accreditation).

Re: Agentic Coding Is a Trap

#134

Post by bitwize »

marcus_holmes wrote: post #18
bitwize wrote: post #3

This is exactly the same problem of "mechanical engineers' job is to design parts, not machine them, so we'll take training on machines out of the mech eng curriculum." Result: fresh mech eng grads do not know how to properly design parts because they have no idea how they are machined.

How do they solve this for mechanical engineering? Or is it an ongoing problem?

I don't know if they've addressed it. But ~15 years ago, my father was mentoring some college students and noticed that while they had been taught to machine a block (i.e., the rudiments of machining), they had no idea how to design appropriate tolerances for e.g., a gear because they hadn't ever made anything that complicated. So it was an issue back then. Presumably these details are learned on the job through trial and error, or by oversight from a more senior engineer who understands the requirements better. But in the past it was understood you'd start learning them from actually building parts as part of the curriculum.

Re: Agentic Coding Is a Trap

#135

Post by CGamesPlay »

byzantinegene wrote: post #116
fnordpiglet wrote: post #51

Interestingly I’ve learned more about languages and systems and tools I use in the last few years working with agentic coding than I did in 35 years of artisanal programming. I am still vastly superior at making decisions about systems and techniques and approaches than the agentic tools, but they are like a really really well read intern who knows a great deal of detail about errata but have very little experience.…

you have 35 years of experience and have already built up the learning capability and general framework to acquire new knowledge. you know how to use agentic coding as a tool to supplement your work. the juniors who start today don't have that, they overrely on agentic coding and do not know what they don't know

Exactly this. We need to be more precise than blanket statements like "agentic coding is a trap" and start figuring out what a "tasteful" application of agentic coding looks like. ChatGPT is destroying liberal arts curriculums because students can choose to not do anything of the thinking themselves and produce mediocre work that passes the bar. I think the same problem is showing itself with agentic coding, just with more directly measurable consequences (because the pile of software ends up failing in a more spectacular way than the pile of bad writing).

Re: Agentic Coding Is a Trap

#136

Post by mempko »

EGreg wrote: post #62

Agents are a first-generation technology. They propose and act at the same time. I recommend you read https://safebots.ai/agents.html

I believe agents are the wrong abstraction. I came up with a superset called an Abject. See https://abject.world

Seems safebox went after a subset.

Re: Agentic Coding Is a Trap

#137

Post by hypeatei »

monksy wrote: post #20

I kind of think this article misses the mark a little. There is skill loss from heavy AI use. But I want to acknowledge the awkward elephant in the room. AI Is making people too fast. I don't mean that a faster output is bad. It's a faster output and code rather than a full understanding and experience in producing the code. It's rewarding people who try to talk about business value rather than the people that are bu…

> We're in a cesspit of business driven development

It's not just businesses doing it either, I regularly see big PRs get merged on open source projects that seem fine on the surface but contain a 1000 paper cuts worth of bugs (not critical, but just enough to annoy you)

On top of that, the code wasn't idiomatic C++ (for this specific project) and the LLM completely ignored available APIs. Sure, it can be fixed, and maintainers should've caught it, but the amount of code being generated requires so much energy on everyone's behalf.

Re: Agentic Coding Is a Trap

#138

Post by bitwize »

dboreham wrote: post #6
bitwize wrote: post #3

This is exactly the same problem of "mechanical engineers' job is to design parts, not machine them, so we'll take training on machines out of the mech eng curriculum." Result: fresh mech eng grads do not know how to properly design parts because they have no idea how they are machined.

Surely they're made on CNC machines now? (well, since the 1970s)

Lathe, mill, CNC, or matter transmuter, doesn't matter. Effective design only becomes possible with intimate knowledge of how it is built.

Re: Agentic Coding Is a Trap

#139

Post by echelon »

byzantinegene wrote: post #116
fnordpiglet wrote: post #51

Interestingly I’ve learned more about languages and systems and tools I use in the last few years working with agentic coding than I did in 35 years of artisanal programming. I am still vastly superior at making decisions about systems and techniques and approaches than the agentic tools, but they are like a really really well read intern who knows a great deal of detail about errata but have very little experience.…

you have 35 years of experience and have already built up the learning capability and general framework to acquire new knowledge. you know how to use agentic coding as a tool to supplement your work. the juniors who start today don't have that, they overrely on agentic coding and do not know what they don't know

> the juniors who start today don't have that, they overrely on agentic coding and do not know what they don't know

Y'all need to stop worrying about the kids.

They're smarter than us and will run circles around us.

They're going to look at us like dinosaurs and they're going to solve problems of scale and scope 10x or more than what we ever did.

Hate to "old man yells at cloud" this, but so many people are falling into the trap because of personal biases.

While the fear that "smartphones might make kids less computer literate" is true, that's because PCs are not as necessary as they once were. The kids that turn into engineers are fine and are every bit as capable.

Re: Agentic Coding Is a Trap

#140

Post by faangguyindia »

I don't use AI for everything, but I use AI to make repeatable, auditable workflows. For DevOps, I don't just ask AI to go on my production server and fix all issues. I ask it to write scripts, which I audit, then I dry run, then I test, and finally approve and run on production.

I was just looking through HN search for "show HN", and I saw many fitness and calorie tracking apps.

A lot of them disappeared just after a few months of launch; a few of them survived a year, then died out as their domain name expired.

People are making things, but they are not reaching their "audience".

I created https://macrocodex.app/, launched on 16 Mar 2026, and reached 10,000+ monthly active users.

Fitness/Calorie tracking is a competitive space where there are tons of apps and services.

I could never have built such an app because I do not know how to design pages; I can talk to a designer, but from past experience, it takes them a long time to understand what the market wants and projects. And companies with small budgets find it very difficult to find a good guy.

Many of my projects never got shipped because I dreaded making landing pages, icons, UI, etc.

I am not saying we did a very good job with AI on landing pages or UI at all; that's not an area of my expertise; the domain knowledge is, but the fact that many people find it useful, I think I’ve succeeded.

I've even put a ticket system in the app for support and received a few bug reports, which I resolved.

Here's the latency of my other service: https://prnt.sc/6474F4gba_he

I no longer use managed services in AWS, and my costs are very low; this enables me to offer my apps and services for free to many users.

Post reply on HN