Code is cheap. Show me the talk • Hacker News Board

Re: Code is cheap. Show me the talk

#171

Post by Sophira » Sat, Jan 31, 2026, 4:08 AM UTC

Completely disagree that "software development, as it has been done for decades, is over".

AI is a useful tool, agreed - but it will always be reliant upon humans writing code.

Let's take a very simple example to prove my point. Let's say that everybody is now using AI to write code, and someone develops a new programming language with it. Cool! We have a new language now.

What we don't have, however, are any examples of how that programming language is used. Nothing that the AI can learn from.

How well do you think AI is going to handle being able to code in this new language which doesn't appear at all in its training data?

Re: Code is cheap. Show me the talk

#172

Post by overgard » Sat, Jan 31, 2026, 4:11 AM UTC

cheema33 wrote: post #152

overgard wrote: post #123

I asked Codex to write some unit tests for Redux today. At first glance it looked fine, and I continued on. I then went back to add a test by hand, and after looking more closely at the output there were like 50 wtf worthy things scattered in there. Sure they ran, but it was bad in all sorts of ways. And this was just writing something very basic. This has been my experience almost every time I use AI: superficially…

This is how you do things if you are new to this game. Get two other, different, LLMs to thoroughly review the code. If you don’t have an automated way to do all of this, you will struggle and eventually put yourself out of a job. If you do use this approach, you will get code that is better than what most software devs put out. And that gives you a good base to work with if you need to add polish to it.

I actually have used other LLMs to review the code, in the past (not today, but in the past). It's fine, but it doesn't tend to catch things like "this technically works but it's loading a footgun." For example, the redux test I was mentioning in my original post, the tests were reusing a single global store variable. It technically worked, the tests ran, and since these were the first tests I introduced in the code base there weren't any issues even though this made the tests non deterministic... but, it was a pattern that was easily going to break down the line.

To me, the solution isn't "more AI", it's "how do I use AI in a way that doesn't screw me over a few weeks/months down the line", and for me that's by making sure I understand the code it generated and trim out the things that are bad/excessive. If it's generating things I don't understand, then I need to understand them, because I have to debug it at some point.

Also, in this case it was just some unit tests, so who cares, but if this was a service that was publicly exposed on the web? I would definitely want to make sure I had a human in the loop for anything security related, and I would ABSOLUTELY want to make sure I understood it if it were handling user data.

Re: Code is cheap. Show me the talk

#173

Post by timcobb » Sat, Jan 31, 2026, 4:21 AM UTC

alt187 wrote: post #170

3kkdd wrote: post #158

Earlier quoted context omitted.

Im sick and tired of these empty posts. SHOW AN EXAMPLE OF YOU ACTUALLY DOING WHAT YOU SAY!

There's no example because OP has never done this, and never will. People lie on the internet.

I've never done this because i haven't felt compelled to do this because I want to review my own code but I imagine this works okay and isn't hard to set up by asking Claude to set this up for you...

Re: Code is cheap. Show me the talk

#174

Post by timcobb » Sat, Jan 31, 2026, 4:22 AM UTC

cheema33 wrote: post #152

overgard wrote: post #123

I asked Codex to write some unit tests for Redux today. At first glance it looked fine, and I continued on. I then went back to add a test by hand, and after looking more closely at the output there were like 50 wtf worthy things scattered in there. Sure they ran, but it was bad in all sorts of ways. And this was just writing something very basic. This has been my experience almost every time I use AI: superficially…

This is how you do things if you are new to this game. Get two other, different, LLMs to thoroughly review the code. If you don’t have an automated way to do all of this, you will struggle and eventually put yourself out of a job. If you do use this approach, you will get code that is better than what most software devs put out. And that gives you a good base to work with if you need to add polish to it.

> you will struggle and eventually put yourself out of a job.

We can have a discussion without the stakes being so high.

Re: Code is cheap. Show me the talk

#175

Post by visarga » Sat, Jan 31, 2026, 4:46 AM UTC

overgard wrote: post #123

I asked Codex to write some unit tests for Redux today. At first glance it looked fine, and I continued on. I then went back to add a test by hand, and after looking more closely at the output there were like 50 wtf worthy things scattered in there. Sure they ran, but it was bad in all sorts of ways. And this was just writing something very basic. This has been my experience almost every time I use AI: superficially…

> "write some unit tests for Redux today"

The equivalent of "draw me a dog" -> not a masterpiece!? who would have thought? You need to come up with a testing methodology, write it down, and then ask the model to go through it. It likes to make assumptions on unspecified things, so you got to be careful.

More fundamentally I think testing is becoming the core component we need to think about. We should not vibe-check AI code, we should code-check it. Of course it will write the actual test code, but your main priority is to think about "how do I test this?"

You can only know the value of a code up to the level of its testing. You can't commit your eyes into the repo, so don't do "LGTM" vibe-testing of AI code, it's walking a motorcycle.

Re: Code is cheap. Show me the talk

#176

Post by simonw » Sat, Jan 31, 2026, 5:22 AM UTC

krupan wrote: post #164

simonw wrote: post #15

Earlier quoted context omitted.

I'm not sure you're actually in disagreement with the author of this piece at all. They didn't say that software engineering is over - they said: > Software development, as it has been done for decades, is over. You argue that writing code is 10-20% of the craft. That's the point they are making too! They're framing the rest of it as the "talking", which is now even more important than it was before thanks to the wri…

When we say generating code is only a small percentage, that does not imply that the rest is just talking. Simon, you were part of a relatively small fast moving project in Django and the news website it powered, with from I understand a pretty small team. Have you worked as part of a team of 10, 20, 100, 1000 engineers? It's different.

"Talking" here doesn't literally mean talking. It means figuring out the scope of the problem, researching solutions, communicating with stakeholders, debating architecture, building exploratory prototypes, breaking down projects - it's all the stuff that isn't writing the code.

I've worked at various sizes of organization. Most notably I joined Eventbrite when they were less than 100 developers and stayed while they grew to around 1,000.

Re: Code is cheap. Show me the talk

#177

Post by acemarke » Sat, Jan 31, 2026, 5:22 AM UTC

overgard wrote: post #123

I asked Codex to write some unit tests for Redux today. At first glance it looked fine, and I continued on. I then went back to add a test by hand, and after looking more closely at the output there were like 50 wtf worthy things scattered in there. Sure they ran, but it was bad in all sorts of ways. And this was just writing something very basic. This has been my experience almost every time I use AI: superficially…

Hi, I'm the primary Redux maintainer. I'd love to see some examples of what got generated! (Doubt there's anything we could do to _influence_ this, but curious what happened here.)

FWIW we do have our docs on testing approaches here, and have recommended a more integrated-style approach to testing for a while:

- https://redux.js.org/usage/writing-tests

Re: Code is cheap. Show me the talk

#178

Post by graemep » Sat, Jan 31, 2026, 5:39 AM UTC

jdjeeee wrote: post #84

bheadmaster wrote: post #79

Earlier quoted context omitted.

This is actually a great way to foster the learning spirit in the age of AI. Even if the student uses AI to arrive at an answer, they will still need to, at the very least, ask the AI to give it an explanation that will teach them how it arrived to the solution.

No this is not the way we want learning to be - just like how students are banned from using calculators until they have mastered the foundational thinking.

There is research that shows that banning calculators impedes the learning of maths. It is certainly not obvious to me that calculators will have a negative effect - I certainly always allowed my kids to use them.

LLMs are trickier and use needs to be restricted to stop cheating, just as my kids had restrictions on what calculators they could use in some exams. That does not mean they are all bad or even net bad if used correctly.

Re: Code is cheap. Show me the talk

#179

Post by graemep » Sat, Jan 31, 2026, 5:47 AM UTC

icedchai wrote: post #138

enlyth wrote: post #48

Earlier quoted context omitted.

A software engineer with an LLM is still infinitely more powerful than a commoner with an LLM. The engineer can debug, guide, change approaches, and give very specific instructions if they know what needs to be done. The commoner can only hammer the prompt repeatedly with "this doesn't work can you fix it". So yes, our jobs are changing rapidly, but this doesn't strike me as being obsolete any time soon.

Yep, I've seen a couple of folks pretending to be junior PMs, thinking they can replace developers entirely. The problem is, they can't write a spec. They can define a feature at a very high level, on a good day. They resort to asking one AI to write them a spec that they feed to another. It's slop all the way down.

People have tried that with everything from COBOL to low code. Its even succeeded in some problem domains (e.g. thing people code with spreadsheet formula) but there is no general solution that replaces programmers entirely.

Re: Code is cheap. Show me the talk

#180

Post by dmitrygr » Sat, Jan 31, 2026, 6:16 AM UTC

> One can no longer know whether such a repository was “vibe” coded by a non-technical person who has never written a single line of code, or an experienced developer, who may or may not have used LLM assistance

Usually one can tell quite easily, actually...