frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Micro-Front Ends in 2026: Architecture Win or Enterprise Tax?

https://iocombats.com/blogs/micro-frontends-in-2026
1•ghazikhan205•48s ago•0 comments

Japanese rice is the most expensive in the world

https://www.cnn.com/2026/02/07/travel/this-is-the-worlds-most-expensive-rice-but-what-does-it-tas...
1•mooreds•1m ago•0 comments

These White-Collar Workers Actually Made the Switch to a Trade

https://www.wsj.com/lifestyle/careers/white-collar-mid-career-trades-caca4b5f
1•impish9208•1m ago•1 comments

The Wonder Drug That's Plaguing Sports

https://www.nytimes.com/2026/02/02/us/ostarine-olympics-doping.html
1•mooreds•1m ago•0 comments

Show HN: Which chef knife steels are good? Data from 540 Reddit tread

https://new.knife.day/blog/reddit-steel-sentiment-analysis
1•p-s-v•1m ago•0 comments

Federated Credential Management (FedCM)

https://ciamweekly.substack.com/p/federated-credential-management-fedcm
1•mooreds•2m ago•0 comments

Token-to-Credit Conversion: Avoiding Floating-Point Errors in AI Billing Systems

https://app.writtte.com/read/kZ8Kj6R
1•lasgawe•2m ago•1 comments

The Story of Heroku (2022)

https://leerob.com/heroku
1•tosh•2m ago•0 comments

Obey the Testing Goat

https://www.obeythetestinggoat.com/
1•mkl95•3m ago•0 comments

Claude Opus 4.6 extends LLM pareto frontier

https://michaelshi.me/pareto/
1•mikeshi42•3m ago•0 comments

Brute Force Colors (2022)

https://arnaud-carre.github.io/2022-12-30-amiga-ham/
1•erickhill•6m ago•0 comments

Google Translate apparently vulnerable to prompt injection

https://www.lesswrong.com/posts/tAh2keDNEEHMXvLvz/prompt-injection-in-google-translate-reveals-ba...
1•julkali•6m ago•0 comments

(Bsky thread) "This turns the maintainer into an unwitting vibe coder"

https://bsky.app/profile/fullmoon.id/post/3meadfaulhk2s
1•todsacerdoti•7m ago•0 comments

Software development is undergoing a Renaissance in front of our eyes

https://twitter.com/gdb/status/2019566641491963946
1•tosh•8m ago•0 comments

Can you beat ensloppification? I made a quiz for Wikipedia's Signs of AI Writing

https://tryward.app/aiquiz
1•bennydog224•9m ago•1 comments

Spec-Driven Design with Kiro: Lessons from Seddle

https://medium.com/@dustin_44710/spec-driven-design-with-kiro-lessons-from-seddle-9320ef18a61f
1•nslog•9m ago•0 comments

Agents need good developer experience too

https://modal.com/blog/agents-devex
1•birdculture•10m ago•0 comments

The Dark Factory

https://twitter.com/i/status/2020161285376082326
1•Ozzie_osman•10m ago•0 comments

Free data transfer out to internet when moving out of AWS (2024)

https://aws.amazon.com/blogs/aws/free-data-transfer-out-to-internet-when-moving-out-of-aws/
1•tosh•11m ago•0 comments

Interop 2025: A Year of Convergence

https://webkit.org/blog/17808/interop-2025-review/
1•alwillis•13m ago•0 comments

Prejudice Against Leprosy

https://text.npr.org/g-s1-108321
1•hi41•14m ago•0 comments

Slint: Cross Platform UI Library

https://slint.dev/
1•Palmik•18m ago•0 comments

AI and Education: Generative AI and the Future of Critical Thinking

https://www.youtube.com/watch?v=k7PvscqGD24
1•nyc111•18m ago•0 comments

Maple Mono: Smooth your coding flow

https://font.subf.dev/en/
1•signa11•19m ago•0 comments

Moltbook isn't real but it can still hurt you

https://12gramsofcarbon.com/p/tech-things-moltbook-isnt-real-but
1•theahura•22m ago•0 comments

Take Back the Em Dash–and Your Voice

https://spin.atomicobject.com/take-back-em-dash/
1•ingve•23m ago•0 comments

Show HN: 289x speedup over MLP using Spectral Graphs

https://zenodo.org/login/?next=%2Fme%2Fuploads%3Fq%3D%26f%3Dshared_with_me%25253Afalse%26l%3Dlist...
1•andrespi•24m ago•0 comments

Teaching Mathematics

https://www.karlin.mff.cuni.cz/~spurny/doc/articles/arnold.htm
2•samuel246•26m ago•0 comments

3D Printed Microfluidic Multiplexing [video]

https://www.youtube.com/watch?v=VZ2ZcOzLnGg
2•downboots•27m ago•0 comments

Abstractions Are in the Eye of the Beholder

https://software.rajivprab.com/2019/08/29/abstractions-are-in-the-eye-of-the-beholder/
2•whack•27m ago•0 comments
Open in hackernews

Stop Hacklore – An Open Letter

https://www.hacklore.org/letter
9•todsacerdoti•2mo ago

Comments

ianpenney•2mo ago
As someone who takes this seriously I’m only working jobs where your ass will get fired immediately for clicking that link in the clearly dumb email.

If leadership won’t bring the same rigour of safety culture - which is mandated by legislation - to security? Don’t bother, just move on.

OkayPhysicist•2mo ago
Traveling to the wrong webpage pwning you is a piece of hacklore so outdated they replaced it with the updated QR code version. Clicking a link has not been a dangerous activity for years. When the rare browser exploit is discovered, it's patched immediately.
gnabgib•2mo ago
I must be misunderstanding you because phishing happens weekly with huge consequences. It's not browser exploits, it's an email that looks legit enough with an incorrect URL or a page that's so convincingly identical to PayPal you feed it the information. Just this week:

SitusAMC https://www.situsamc.com/databreach

Harvard University https://www.bleepingcomputer.com/news/security/harvard-unive...

Iberia Airline https://www.bleepingcomputer.com/news/security/iberia-disclo...

Salesforce via gainsight https://status.salesforce.com/generalmessages/20000233

OkayPhysicist•2mo ago
Phishing is tricking someone into providing confidential information to a malicious party/site. "Don't click on suspicious links" is, IMO, an overreaction that fails to teach people the core lesson that is "Always confirm that you're providing sensitive information to the party you think you are".

Online, we've made it exceptionally easy to make those sorts of checks: a website, served over HTTPS, is coming from the url. Other systems are so, so much worse about this. Any system where unauthorized impersonation is possible is a technical failure, and the fault for abuse of that unauthorized impersonation is on the providers and designers of that system. Like phone calls. Or email.

People tend to be pretty good at differentiating between "this person can be trusted with sensitive information", and "I shouldn't trust this stranger". What they need are the tools to determine who they're talking to.