Laws of Software Engineering

[nelsonic]

https://lawsofsoftwareengineering.com/
via: https://news.ycombinator.com/item?id=47847179
Top comment is 100% correct. ✅

[ndrean]

Lots of good common sense. The last one is interesting "Cunningham's Law " "The best way to get the correct answer on the Internet is not to ask a question, it's to post the wrong answer.".
When you baptise something, it looks serious.
I am very curious about Stackoverflow. I would guess that it is dying or seriously slowed down since the era of AI but I have no data.

[nelsonic]

"Cunningham's Law" is genius if you can get it to work without being downvoted. 😆
Reminds me of: https://xkcd.com/386/

StackOverflow is a ghost town; some say it's "almost dead":
https://blog.pragmaticengineer.com/stack-overflow-is-almost-dead/

I still use it occasionally because I still use Google to search for obscure errors that LLMs are useless for.
But I don't expect many claude/cursor-addicted devs to use a "Website" anymore. 🙄
If you recall, SO was bought by Prosus in 2021 near their peak: https://meta.stackoverflow.com/a/390488/1148249
The original creators/owners were prescient to sell it as it's now probably worth 90% less. 📉

[ndrean]

So you documented the common sense intuition 😀 (if we can trust these figures).

Talking about Stackoverflow, I never saw the point of these language surveys they produced.
https://survey.stackoverflow.co/2025
How cool is that to use a desired and admired language? 🤔

I believe it would be more useful to talk about the scope vs tech: where is programming going nowadays? and why do you chose a given tech for this scope. This is more valuable for me. But talking about perceptions makes no sense in the tech world.

[ndrean]

A post of J.Valim about software engineering. The intuition pushes indeed to believe that "software quality" has not improved with AI although I am not sure there are yet facts to prove this statement. Number of identified threats or issues going down etc, these kind of things.

At this point, this is just irresponsible.

Yes, coding agents are leading to an increase of software production, but we are not seeing a similar push or increase in software quality.

If Anthropic focuses on safety and it believes software engineering is going away, then it… https://t.co/kE0rReU5st

— José Valim (@josevalim) April 27, 2026

<script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>

[nelsonic]

Yeah, saw his tweet ("post"?) 100% agree.
There are many objective metrics that can be used to measure code quality. e.g: codacy.com/code-quality-metrics

There has always been tons of poor quality code before Ai
because most organisations/people don't value quality.

Most companies only care about the "Velocity" https://en.wikipedia.org/wiki/Velocity_(software_development)
i.e. the speed of progress at which the features are delivered.
Not the correctness (quality) of the code and whether the tests fully cover edge cases or malicious input.

All that has happened is organisations that don't value quality,
are letting their codebase be plagued with Ai-written crap in the name of "Velocity".
It might meet their immediate needs to ship features,
but will be unmaintainable because no human understands it.
It's the definition of technical debt.

Ref:
https://www.reddit.com/r/theprimeagen/comments/1qfci6q/i_shipped_code_i_dont_understand_and_i_bet_you/
https://www.reddit.com/r/technology/comments/1ptpc95/aigenerated_code_contains_more_bugs_and_errors/
Though this thread is from 4 months ago, which is years in Ai,
I don't see how it will improve, do you? 🤷‍♂️

[ndrean]

Well, even Jose Valim seems to push feelings rather than facts, and as you said, 4 months is like years. Improve unmaintainable ? It is a key word, but by humans?? 😀. I am not sure unless it is really important stuff in which case human input might be still important. Then testing is even more important as it might be the ultimate thing you can firmly grab on. You not only use Codex to test Claude generated code, but we need to really check all the tests you want! But of course I have no crystal ball, otherwise I wouldn't even be here discussing this 😀 so I am thinking of stepping out of this business.

[nelsonic]

“stepping out of this business” is exactly what Scam Altman & the Ai Bros want you and all other senior/experienced engineers to do. 🪦
They want to devalue your unique human experience/perspective/taste and instead make everyone dependent on their flawed tools.

https://llmdeathcount.com 😢

Out of curiosity, what do you think you would do instead of being a Software Engineer? 💭

[ndrean]

I try to never underestimate the power of credulity of people because many are lazy and need certainty to stop thinking. Current models intrinsically hallucinate: the perfect candidate to be a kinda next religion but atomised, tailored to you. AI is perfect to fill in all the gaps, holes and uncertainties we all have. I feel like a guru now 😄
That said, I am not sure if https://llmdeathcount.com/ is not itself propaganda: does the police really investigates or has access to some chat records?? Plausible and therefore suspicious.

[nelsonic]

Ai religion for sure. 🙃

Sadly, LLM Deaths are very real. 😢

There’s an active lawsuit from his parents which wouldn’t exist without evidence, see:
https://en.wikipedia.org/wiki/Raine_v._OpenAI