HN Distilled

Essential insights from Hacker News discussions

A Guide to Gen AI / LLM Vibecoding for Expert Programmers

Original Article

Hacker News Discussion

Here's a summary of the themes discussed on Hacker News regarding "vibe coding" with LLMs:

Skepticism about LLM Capabilities and Reliability

A significant portion of the discussion expresses fundamental doubts about the current capabilities and trustworthiness of LLMs for generating reliable code. Many users feel that LLMs "don't understand" in a way that humans do, leading to frustrating interactions and subtly incorrect outputs.

  • "No, you can't either. Not a single soul on this planet can review the thousands of lines of vibe coded bullshit that LLMs spit out." - iLoveOncall
  • "It takes way more time to explain, and then re-explain, and then re-re-re-explain to the LLM what I want the code to do. No, it isn't because I don't understand LLMs, it's because LLMs don't understand, period." -leptons
  • "You also seem to be missing the point that if vibe coding lets your engineers write 10x the amount of code they previously could in the same working hours, you now have to review 10x that amount." - iLoveOncall

The Importance of Rigorous Code Review

The discussion highlights the essential role of thorough code review, regardless of whether the code is written by a human or an LLM. Some argue that even experienced engineers cannot simply trust LLM-generated code without meticulous scrutiny, as subtle errors can easily slip through.

  • "The same way a person normally writes code, you are just having the AI do the grunt work. This does have the caveat that reading code is usually harder than writing it, so the total time savings is far less than what AI companies claim. You only get in danger when you don't review the code and just YOLO it into production." - jandrese
  • "ivape: You underestimate how much code some people have sifted through in their lifetime. I read through every line the LLM codes, function by function. I know exactly when it’s going sideways. You’d have to know to even know, it’s one of those things. If you don’t know then you don’t know. Put it another way, you know how some people fail open book tests? How’s that possible? The book is not enough." - ivape
  • "As a tech lead I have reviewed code written by junior engineers and written by AI, and there is a very clear difference between the two." - iLoveOncall

The Diminishing Returns of LLM Assistance Due to Review Overhead

Multiple contributors point out that the claimed efficiency gains from using LLMs for coding are often negated by the time required to understand, debug, and review the generated code. The act of "vibe coding" itself can become just as time-consuming as writing the code manually once the necessary verification is done.

  • "I found that once I spend enough time to actually fully understand what LLM wrote, I’ve burned through my efficiency gains. If that’s the case, why bother?" - exitb
  • "If you have to describe the code to the ai and then read through each line of it anyways, why not just write the code yourself?" - shortrounddev2

The "Skill" of Using LLMs and the Role of Experience

There's a debate about whether using LLMs effectively is a new "skill" or simply a different way of interacting with tools. Some believe that traditional programming experience is crucial for discerning the quality of LLM output, while others suggest that newer, less-experienced developers might adapt more readily to these new paradigms.

  • "using ai is not a "skill" that ppl are trying to make it to be." - apwellwell23
  • "I don't follow. Why wouldn't you believe that a senior engineer, that has been reviewing code for years, be any better at reviewing code?" - rychco
  • "if anyone takes the art of software programming further using LLMs, it’s going to be young inexperienced people who grow up closely observing and learning the transcendental nature of LLMs and software, not hardened industry titans joyfully cracking their whip over an army of junior devs and LLMs." - ppqqrr
  • "New generations are always leapfrogging those that came before them, so I don't find it too hard to believe even under more pessimistic opinions of LLM usefulness." - Retr0id

Reframing the LLM Workflow: Small, Reviewable Chunks

A counter-argument suggests that the problem isn't LLMs themselves, but the way they are being used. Breaking down tasks into smaller, manageable modules that can be individually verified and then integrated is proposed as a more effective approach, similar to traditional development practices.

  • "The big thing is to have the AI spit out small digestible modules, check those for correctness, and then glue them together. The same way a person normally writes code, you are just having the AI do the grunt work." - jandrese
  • "So, why would you review thousands of lines at a time? That means you didn't break the problem down appropriately. PRs should be around 100-200 lines of code, including tests. You build your tasks appropriately. Reviewing thousands of lines at a time is always a failure state." - ebiester

The Nature of Programming and AI's Impact on Creativity

Some users ponder the fundamental nature of programming in the age of LLMs, questioning whether the goal is to create static "sculptures" or more dynamic, "alive" programs. The potential for LLMs to facilitate new forms of creativity is acknowledged, though its application to typical software projects is debated.

  • "we’ve been taught to think of programs as sculptures, shaped by the creator to a fixed purpose. with LLMs, the greatest advance isn’t in being able to make larger and more detailed sculptures more quickly; it’s that you can make the sculptures alive." - ppqqrr
  • "But who wants a program to be alive? To be super clear, I love the tech behind LLMs and other transformers. But when discussing regular, run of the mill software projects that don't require AI capabilities - do you really need to have the understanding of the transcendental nature of LLMs to do that job well?" - rafterydj

Personal Exploration and Intuition Building as Key to Mastery

A recurring theme is that the best way to understand and effectively use LLMs is through hands-on experimentation and practical application. Building an intuition for what LLMs can and cannot do is seen as a crucial step towards mastering the technology.

  • "Try and use it for shit and play around with it — you’ll figure out where it sucks, where it is good by yourself." - mccoyb
  • "Couldn't agree more. The only way to really get effective at using this stuff is to spend time with it, building up intuition about what it can and cannot do. Playing with it is a great way to do that." - simonw

Distrust of "Vibe Coding" by Those Who Enjoy Programming

For those who genuinely enjoy the craft of programming, the concept of "vibe coding" is perceived as a chore, equating it to tedious management tasks rather than fulfilling creative work. The comparison to managing scrums or helping a student with a thesis, without the human element, is seen negatively.

  • "I think the article could be more accurately titled 'A Guide to Gen AI / LLM Vibecoding for Programmers who hate their job' To me, someone who actually love programming, it makes vibe coding look like hell." - GuB-42
  • "The workflow of vibe coding is the same as managing scrums or helping a student through a research thesis. Which programmer wants that?! Just hearing the word 'scrum' makes me want to run away..." - GuB-42