frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

Open in hackernews

AI didn't kill Stack Overflow

https://www.infoworld.com/article/3993482/ai-didnt-kill-stack-overflow.html
14•taubek•1d ago

Comments

comrade1234•1d ago
For me, when trying to find an answer to a problem, I never went to stack overflow first and searched using their ui. Instead I always searched in google. Over time the google results got worse - putting Reddit at the top, for example, and often stack overflow would be on the next page of results.

At around the same time google's search results started getting bad I started using an llm and those results were good enough that I never tried stack overflow directly.

WalterGR•18h ago
I just add “stack exchange” to my Google searches. It works fine.
m-schuetz•1d ago
Yeah, AI didnt cause SO's decline, it merely accelerated it. I stopped using SO years before that because it felt like whenever you were asking a question, you had to spend 80% of the time justifying why you need to do it that way. And occasionally, someone would still show up and say that the question is wrong because I should do something entirely different.
smitty1e•1d ago
Sites like Stack Overflow are a DMZ[1] where one can solve problems and build skills.

Conceiving of them as a teaching/recruiting market might be a way to help matters, but keeping the focus on the strategic mission of developing people, instead of the tactical goal of maximizing profit, is the gnarly part.

[1] https://en.wikipedia.org/wiki/Demilitarized_zone

OutOfHere•1d ago
SO failed to use a structured approach to validating new questions. Users could have been asked to justify how their question is different from each of the top five closest older questions, also whenever a challenge arises, and generally given the benefit of the doubt. Instead, users met with hostile downvotes whenever posting a question even if it was an original one.

This may be hard to appreciate now, but AI is soon going to begin to struggle with self-training, leading to worse outputs and even a model collapse. For this reason, SO still is more important than ever as a source of original training material.

xtiansimon•1d ago
This piece is strange. SO’s organizing principle was _always_ on a collision course with stagnation. Plane goes up. Plane goes down.

The author rides right over or just thinks you can group Usenet with what came after—forums.

Forums were everywhere. They were single threads, and they were always going off the rails, off-topic, changing between what the OP asked and what the final best answer became.

And along came SO. Klaatu came out of the spaceship and said you will do the work, you will make questions and we will provide answers—and no Duplicated Questions. If you’re policing in this way, at some future point you will have less participation. Plane goes up. Plane goes down. The only way for this not to occur, system wide, is if there are new and better technologies which are growing rapidly.

That was always happening at SO, with or without AI

orev•1d ago
Stack Overflow harnessed the good attributes of tech people (I’m not saying that only tech people have these attributes, but I do think they’re very common in tech people): the joy of learning and the joy of sharing/helping others learn.

However that also brings with it some bad attributes of tech people (again, not exclusive to tech people, but very common): intolerance to people who don’t try to help themselves first, don’t “RTFM”, don’t try to search first, and (probably most importantly) post duplicate questions.

Unfortunately for Stack Overflow (and other tech sites), those “bad” things are also critical to what makes a community site be successful. Sites like Reddit, Twitter, Instagram, etc. are very successful and full of people posting things that are off topic, low effort, and duplicated. On Stack Overflow, only the first “your dog is so cute” comment would be allowed, and then all the rest deleted. And in the process, alienating everyone who had something deleted—those negative feelings add up eventually.

And maybe that’s the fundamental flaw of it. They’re trying to be both a social site and a version of Wikipedia where each page is the single source of truth for that particular topic. They might recover from their problems if they reduced the social aspects and leaned more into the reference material aspect of the site. But maybe their funding model won’t allow that.

fabian2k•1d ago
> For Stack Overflow, the new model, along with highly subjective ideas of “quality” opened the gates to a kind of Stanford Prison Experiment. Rather than encouraging a wide range of interactions and behaviors, moderators earned reputation by culling interactions they deemed irrelevant.

This part is just outright wrong, though I think it's not an uncommon misunderstanding. You do not get any reputation at all for removing content. Curation is almost not incentivized by reputation at all, the only exception are suggested edits (and rep gain is capped there).

If your goal is only to get more internet points the best strategy is to quickly answer every question that comes in, and not to close, downvote or delete anything. Curation actually works against you as you won't get points if the question is deleted (within a certain time limit). So gamification is doing the opposite of what the author claims here.

johnea•1d ago
I didn't see in the article, or anywhere in these comments, the reasons that Stack Overflow became a "read only" resource for me.

The article did center around user's as moderators, but I felt there was a sudden increase in moderation around the time the company was sold.

Previously I had been able to enter a question, or an answer relatively easily. But there was a sudden and sustained increase in the need to have a high enough "reputation" to be able to do either.

It seemed like the site became centered around people who spent a lot of time there, honing their points. And anyone who wasn't doing this just became unable to post questions or answers.

At that point, I would just use the site to refer to answers already in place, and no longer posted at all. From then on, there was a slow degradation of new and relevant posts.

It does still occasionally come up in a search on a technical question, and I still use it when that happens.

cadamsdotcom•1d ago
Stack Overflow is a fantastic concept: Q&A to build a knowledge base.

To really round out the Stack Overflow project the next step is to turn the knowledge into a wiki - the world’s biggest “missing manual” of community curated knowledge of how things really work.

That’d reduce the pressure to close duplicate questions. Mods who want to curate can work on the wiki instead!

Want a humanoid, open-source robot for just $3k? Hugging Face is on it

https://arstechnica.com/ai/2025/05/hugging-face-hopes-to-bring-a-humanoid-robot-to-market-for-just-3000/
1•walterbell•7m ago•0 comments

Scientists Discover Brain 'Time Cells' That Enable 'Mental Time Travel' (2021)

https://www.vice.com/en/article/scientists-discover-time-cells-in-the-brain-that-enable-mental-time-travel/
1•walterbell•8m ago•0 comments

UK to build up to 12 new attack submarines

https://www.bbc.com/news/articles/c4g2jr1m49no
1•geox•17m ago•0 comments

DuckLake with Ibis Python DataFrames

https://emilsadek.com/blog/ducklake-ibis/
1•esadek•25m ago•0 comments

Show HN: Moon Phase Algorithms for C, Lua, Awk, JavaScript, etc.

https://github.com/oliverkwebb/moonphase
1•oliverkwebb•39m ago•0 comments

The Internet of Consent

https://www.anildash.com//2025/05/27/2025-05-27-internet-of-consent/
1•Refreeze5224•44m ago•0 comments

Show HN: NextGP – a clean F1 countdown app built with Expo and React Native

https://apps.apple.com/us/app/nextgp/id6746546202
1•flakii•54m ago•0 comments

Fireside Chat with Kelsey Hightower: An Outsider's Look at Nix

https://www.youtube.com/watch?v=caxcawUCSZ8
3•kesor•56m ago•0 comments

Absurdly Good Doggo Consistency with Flux.1 Kontext

https://notesbylex.com/absurdly-good-doggo-consistency-with-flux1-kontext
2•lexandstuff•56m ago•0 comments

Run LLM models on your Android phone locally

https://github.com/google-ai-edge/gallery
2•sarusso•1h ago•0 comments

A beginner's guide to constant-time cryptography

https://www.chosenplaintext.ca/articles/beginners-guide-constant-time-cryptography.html
3•susam•1h ago•1 comments

Show HN: Wikigen.ai – an engaging way to learn [video]

https://www.youtube.com/watch?v=Ov4aE7F_DnE
2•priorityfill•1h ago•0 comments

Structured Concurrency in Robot Control

https://max.xz.ax/blog/structured-concurrency-robot-control/
1•susam•1h ago•0 comments

Scientific Computing in Rust 2025 (4-6 June 2025)

https://scientificcomputing.rs/2025/
1•susam•1h ago•0 comments

Show HN: Feedback Wanted: Gemini-Engineer Tool

https://github.com/ozanunal0/gemini-engineer
1•rtfm01•1h ago•0 comments

Why do lawyers keep using ChatGPT?

https://www.theverge.com/policy/677373/lawyers-chatgpt-hallucinations-ai
3•pseudolus•1h ago•0 comments

Show HN: Better-Experiments – open-source A/B testing library

https://github.com/0xgautam/better-experiments
2•0xgautam•1h ago•0 comments

The robot vacuums we've tested

https://www.theverge.com/22997597/best-robot-vacuum-cleaner
3•walterbell•1h ago•0 comments

Real TikTokers are pretending to be Veo 3 AI creations for fun, attention

https://arstechnica.com/ai/2025/05/real-tiktokers-are-pretending-to-be-veo-3-ai-creations-for-fun-attention/
2•pseudolus•1h ago•0 comments

OAuth 2.0 Flows

https://www.workflows.guru/resources/oauth2-flows-explained
2•ask3m•1h ago•0 comments

Codex Seraphinianus: The Weirdest Book

https://www.abebooks.com/books/rarebooks/serafini-fantasy-art-weird/codex-seraphinianus.shtml
4•Koshkin•1h ago•1 comments

Ask HN: Simple open source to remove the bright background on a web page?

1•9o1d•1h ago•3 comments

A Future of Software Development?

https://nathanlubchenco.substack.com/p/a-future-of-software-development
1•nathanlubchenco•1h ago•0 comments

Ask HN: Why are dating apps so bad? Why hasn't anyone made a good one?

11•1270018080•1h ago•22 comments

DreaMS: Self-Supervised Molecular Representations from Mass Spectra

https://www.nature.com/articles/s41587-025-02663-3
1•gnabgib•1h ago•0 comments

How to Agentically Deconstruct CC

https://southbridge-research.notion.site/How-to-agentically-deconstruct-CC-2055fec70db18150bfa6d218a91c1322
1•handfuloflight•1h ago•0 comments

A Glucose Monitor for Someone Without Diabetes: Optimal or Overkill?

https://www.wsj.com/tech/personal-tech/wearable-glucose-monitor-stelo-dexcom-6655e84d
2•bookofjoe•1h ago•2 comments

The Custodial Stablecoin Rekt Test

https://blog.trailofbits.com/2025/05/29/the-custodial-stablecoin-rekt-test/
2•wslh•1h ago•0 comments

Innovations in aging biology: highlights from the ARDD emerging science workshop

https://www.nature.com/articles/s41514-025-00193-5
2•walterbell•1h ago•0 comments

Black Death bacterium has become less lethal after genetic tweak

https://www.nature.com/articles/d41586-025-01687-8
1•geox•1h ago•0 comments