frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

Open in hackernews

Some Reddit users just love to disagree, says new AI-powered troll-spotting algo

https://theconversation.com/some-reddit-users-just-love-to-disagree-new-ai-powered-troll-spotting-algorithm-finds-255879
3•msephton•4h ago

Comments

pavel_lishin•3h ago
NO WE DON'T!
wewxjfq•2h ago
I'm a disagreer and I have to disagree with the authors: disagreeing is not trolling, it's countering trolls. Trolls usually flood the zone with comments that seem elaborate but are ultimately low effort. They know persistence is key, so they don't bother much with replies that counter their claims; rather, they focus on those in agreement to manufacture consent.

It might look sinister if someone drops a reply and leaves, but that alone tells you nothing. In my experience, there is always Brandolini's law at play:

> The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.

If a reply takes a lot of effort and if you suspect the other side isn't arguing in good faith, one reply already feels like a waste of time and there's no reason to go for round two.

[1] https://en.wikipedia.org/wiki/Brandolini%27s_law

msephton•2h ago
That's an interesting law.

Here's another data point:

"Online political debate isn’t inherently toxic, a new study of Reddit commenters finds. Instead, it becomes toxic because of the kind of commenters who opt in." https://insight.kellogg.northwestern.edu/article/trolls-pois...

wewxjfq•1h ago
I think they should cast a wider net. Toxicity might discover trolls in the original sense, i.e. individuals seeking a dopamine rush. The trolls I have in mind try to steer the narrative on certain topics with comments that aren't rude or inflammatory per se and, like I said, even seem elaborate on the surface.* But if it's done for spin, it's still toxic nonetheless.

What the researchers say about disengagement also applies to these type of trolls. I mentioned persistence, it's what ultimately makes countering these users futile. Anyone who realizes that tomorrow there will be a new thread with the same narrative eventually disengages.

*) More often I see accounts posting long comments in such a short amount of time that I doubt a human was naturally engaging in an online conversation. I would be ashamed to admit how long it took me to write my two comments in this thread and then I see accounts posting much longer comments back-to-back in under two minutes.

Using MathWorks Model-Based Design to Build the Tesla Roadster

https://www.mathworks.com/company/technical-articles/using-model-based-design-to-build-the-tesla-roadster.html
1•felineflock•57s ago•0 comments

Machine learning without true labels and noisy heuristics

https://emiruz.com/post/2025-05-11-soft-labels/
1•usgroup•1m ago•0 comments

How to Reach Out on LinkedIn

https://krul.app/blog/how-to-reach-out-on-linkedin/
1•oczek•1m ago•0 comments

Demonstrably Secure Software Supply Chains with Nix

https://nixcademy.com/posts/secure-supply-chain-with-nix/
1•todsacerdoti•1m ago•0 comments

Show HN: Rocko ReFi

https://rocko.co/learn/how-to-refinance-a-defi-loan/
1•vinniejames•2m ago•0 comments

Skydance Discovers DEI vs. "Public Interest" Catch 22

https://nypost.com/2025/05/09/media/paramount-skydance-must-scrap-dei-to-pass-fccs-public-interest-litmus-test-sources/
1•josefritzishere•2m ago•1 comments

Show HN: Search to Play – Discover and track video games (like Letterboxd)

https://searchtoplay.com/
1•guar47•2m ago•0 comments

Explainable Artificial Intelligence

https://en.wikipedia.org/wiki/Explainable_artificial_intelligence
1•cainxinth•3m ago•0 comments

Initial USA Unemployment Claims

https://fred.stlouisfed.org/series/ICSA
1•mooreds•4m ago•0 comments

Seven Wonders of Biology

https://www.asimov.press/p/seven-wonders-of-biology
1•mailyk•5m ago•0 comments

A Day in a Democracy

https://www.texasobserver.org/a-day-in-a-democracy/
1•mooreds•6m ago•0 comments

Why LLM Memory Isn't Enough for Your AI App (and What to Use Instead)

https://www.recallio.ai/all-posts/why-llm-memory-isn-t-enough-for-your-ai-app-(and-what-to-use-instead)
1•designorbit•7m ago•0 comments

Org Charts of AI Startups Are Built to Stay Small, Flexible

https://www.bloomberg.com/news/articles/2025-05-08/org-charts-of-ai-startups-are-built-to-stay-small-flexible
1•kjhughes•7m ago•1 comments

Open source Seed-Coder-8B model instruct

https://seedcoder.org/
1•zoudong376•9m ago•1 comments

The Fastest Way yet to Color Graphs

https://www.quantamagazine.org/the-fastest-way-yet-to-color-graphs-20250512/
2•isaacfrond•12m ago•0 comments

Can the Anthropic computer use tool be used to play the Chrome T-Rex Runner game

1•largepuma•13m ago•0 comments

Why Vibe When You Can Fly Machtiani Is Now Open Source

https://github.com/tursomari/machtiani
2•todsacerdoti•14m ago•0 comments

USCO "pre-publication" Report is flawed both procedurally and substantively

https://chatgptiseatingtheworld.com/2025/05/12/opinion-why-the-copyright-offices-pre-publication-report-is-flawed-both-procedurally-and-substantively/
1•sophrocyne•15m ago•0 comments

The New Amazon Robot That Can Feel What It Touches [video]

https://www.youtube.com/watch?v=2X4CU3jmw-g
1•ksec•16m ago•0 comments

"The Wolves Came"

https://s-usih.org/2025/05/the-wolves-came-richard-steigmann-gall-on-daniel-steinmetz-jenkinss-did-it-happen-here-perspectives-on-fascism-and-america/
1•Tomte•17m ago•0 comments

DDD: The missing piece in AI app generation?

https://app.swoft.ai
1•Uigatai•17m ago•1 comments

Avoid Each_with_object (Generally)

https://clayshentrup.medium.com/avoid-each-with-object-generally-89e3b2800b38
1•todsacerdoti•23m ago•0 comments

Neural Software – Future of AI [video]

https://www.youtube.com/watch?v=Z-XYhq0190E
1•jasonjmcghee•23m ago•0 comments

The Great Displacement Is Already Well Underway

https://shawnfromportland.substack.com/p/the-great-displacement-is-already
3•JSLegendDev•23m ago•2 comments

An OS for modern work powered by AI agents [video]

https://www.youtube.com/watch?v=2nG92EtDBqM
1•smurthy123•24m ago•1 comments

Cuid2: Next Generation GUIDs

https://github.com/paralleldrive/cuid2
2•rane•24m ago•1 comments

Ask HN: Do You Prepare for Job Interviews? If So, How?

1•dovab•26m ago•0 comments

Reviving a Modular Cargo Bike Design from the 1930s

https://www.core77.com/posts/136773/Reviving-a-Modular-Cargo-Bike-Design-from-the-1930s
2•surprisetalk•26m ago•0 comments

Warlocks-style multiplayer browser game (server authoritative, no installs)

https://warlockjs.pages.dev/
1•iCeGaming•30m ago•1 comments

Ethnicolr_v2 – Replication materials and ms for name to race models

https://github.com/appeler/ethnicolr_v2
1•goji_berries•30m ago•0 comments