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

LLMs work best when the user defines their acceptance criteria first

blog.katanaquant.com

Post reply on HN
Print view View original

111–120 of 460 posts

Re: LLMs work best when the user defines their acceptance criteria first

#111

Post by danparsonson »

unlikelytomato wrote: post #70
pornel wrote: post #36

Their default solution is to keep digging. It has a compounding effect of generating more and more code. If they implement something with a not-so-great approach, they'll keep adding workarounds or redundant code every time they run into limitations later. If you tell them the code is slow, they'll try to add optimized fast paths (more code), specialized routines (more code), custom data structures (even more code).…

This is why I'm confused when people say it isn't ready to replace most of the programmer workforce.

Yeah that describes most legacy codebases I've worked on XD

Re: LLMs work best when the user defines their acceptance criteria first

#112

Post by LarsDu88 »

2god3 wrote: post #72
LarsDu88 wrote: post #58

Earlier quoted context omitted.

This is about to change very soon. Unlike many other domains (such as greenfield scientific discovery), most coding problems for which we can write tests and benchmarks are "verifiable domains". This means an LLM can autogenerated millions of code problem prompts, attempt millions of solutions (both working and non-working), and from the working solutions, penalize answers that have poor performance. The resulting sy…

Lets suppose whatever you say is true. If Im the govt, Id be foaming at the mouth - those projects that used to require enormous funding now will supposedly require much less. Hmmm, what to do? Oh I know. Lets invest in Digital ID-like projects. Fun.

It is true. Here is the publication going over how to generate this type of dataset and finetune: https://arxiv.org/pdf/2506.14245

I don't think you grasp my statement. LLMs will exceed humans greatly for any domain that is easy to computationally verify such as math and code. For areas not amenable to deterministic computations such as human biology, or experimental particle physics, progress will be slower

Re: LLMs work best when the user defines their acceptance criteria first

#113

Post by Foobar8568 »

unlikelytomato wrote: post #70
pornel wrote: post #36

Their default solution is to keep digging. It has a compounding effect of generating more and more code. If they implement something with a not-so-great approach, they'll keep adding workarounds or redundant code every time they run into limitations later. If you tell them the code is slow, they'll try to add optimized fast paths (more code), specialized routines (more code), custom data structures (even more code).…

This is why I'm confused when people say it isn't ready to replace most of the programmer workforce.

LLM code is higher quality than any codes I have seen in my 20 years in F500. So yeah you need to "guide" it, and ensure that it will not bypass all the security guidance for ex...But at least you are in control, although the cognitive load is much higher as well than just "blind trust of what is delivered".

But I can see the carnage with offshoring+LLM, or "most employees", including so call software engineer + LLM.

Re: LLMs work best when the user defines their acceptance criteria first

#114

Post by Foobar8568 »

Mavvie wrote: post #101
stingraycharles wrote: post #46

Earlier quoted context omitted.

> If you ask to unify the duplication, it'll say "No problem, here's a brand new metamock abstract adapter framework that has a superset of all feature sets, plus two new metamock drivers for the older and the newer code! Let me know if you want me to write tests for the new adapters." Nevermind the fact that it only migrated 3 out of 5 duplicated sections, and hasn’t deleted any now-dead code.

Sounds like my coworkers.

That's the reality nobody really wants to say.

Re: LLMs work best when the user defines their acceptance criteria first

#116

Post by ehnto »

ozozozd wrote: post #92
D-Machine wrote: post #90

This article is great. And the blog-article headline is interesting, but wrong. LLM's don't in general write plausible code (as a rule) either. They just write code that is (semantically) similar to code (clusters) seen in its training data, and which haven't been fenced off by RLHF / RLVR. This isn't that hard to remember, and is a correct enough simplification of what generative LLMs actually do, without resorting…

Exactly. It’s also easy to find yourself in the out-of-distribution territory. Just ask for some tree-sitter queries and watch Gemini 3, Opus 4.5 and GLM 5 hallucinate new directives.

I think this could be the key difference in how people are experiencing the tools. Using Claude in industries full of proprietary code is a totally different experience to writing some React components, or framework code in C#, PHP or Java. It's shockingly good at the later, but as you get into proprietary frameworks or newer problem domains it feels like AI in 2023 again, even with the benefit of the agentic harnesses and context augments like memory etc.

Re: LLMs work best when the user defines their acceptance criteria first

#117

Post by red75prime »

tartoran wrote: post #33
lukeify wrote: post #6

Most humans also write plausible code.

LLMs piggyback on human knowledge encoded in all the texts they were trained on without understanding what they're doing. Humans would execute that code and validate it. From plausible it'd becomes hey, it does this and this is what I want. LLMs skip that part, they really have no understanding other than the statistical patterns they infer from their training and they really don't need any for what they are.

Could we stop using vague terms like “understanding” when talking about LLMs and machine learning? You don't know what understanding is. You only know how it feels to understand something.

It's better to describe what you can do that LLMs currently can't.

Re: LLMs work best when the user defines their acceptance criteria first

#118

Post by jqpabc123 »

graphememes wrote: post #69

bad input > bad output idk what to say, just because it's rust doesn't mean it's performant, or that you asked for it to be performant. yes, llms can produce bad code, they can also produce good code, just like people

yes, llms can produce bad code, they can also produce good code, just like people

Over time, you develop a feel for which human coders tend to be consistently "good" or "bad". And you can eliminate the "bad".

With an LLM, output quality is like a box of chocolates, you never know what you're going to get. It varies based on what you ask and what is in it's training data --- which you have no way to examine in advance.

You can't fire an LLM for producing bad code. If you could, you would have to fire them all because they all do it in an unpredictable manner.

Re: LLMs work best when the user defines their acceptance criteria first

#119

Post by otabdeveloper4 »

Implicated wrote: post #88
pornel wrote: post #36

Their default solution is to keep digging. It has a compounding effect of generating more and more code. If they implement something with a not-so-great approach, they'll keep adding workarounds or redundant code every time they run into limitations later. If you tell them the code is slow, they'll try to add optimized fast paths (more code), specialized routines (more code), custom data structures (even more code).…

Not trying to be snarky, with all due respect... this is a skill issue. It's a tool. It's a wildly effective and capable tool. I don't know how or why I have such a wildly different experience than so many that describe their experiences in a similar manner... but... nearly every time I come to the same conclusion that the input determines the output. > If they implement something with a not-so-great approach, they'l…

It is not a tool. It is an oracle.

It can be a tool, for specific niche problems: summarization, extraction, source-to-source translation -- if post-trained properly.

But that isn't what y'all are doing, you're engaging in "replace all the meatsacks AGI ftw" nonsense.

Re: LLMs work best when the user defines their acceptance criteria first

#120

Post by thesz »

Foobar8568 wrote: post #113
unlikelytomato wrote: post #70

Earlier quoted context omitted.

This is why I'm confused when people say it isn't ready to replace most of the programmer workforce.

LLM code is higher quality than any codes I have seen in my 20 years in F500. So yeah you need to "guide" it, and ensure that it will not bypass all the security guidance for ex...But at least you are in control, although the cognitive load is much higher as well than just "blind trust of what is delivered". But I can see the carnage with offshoring+LLM, or "most employees", including so call software engineer + LLM.

  > LLM code is higher quality than any codes I have seen in my 20 years in F500.
"Any codes"?
Post reply on HN