frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Passwords are okay, impulsive Internet isn't

https://www.dedoimedo.com/life/passwords-passkeys.html
3•brycewray•9mo ago

Comments

palata•9mo ago
Hmm... I see a rant against the state of software (bad software, AI diarrhea, ...) and TooBigTech having control over everything. I can agree with that, but it has nothing to do with the "passwords vs passkeys" question.

The rant against passkeys? I don't get it. Just like one can use a password manager controlled by TooBigTech or KeePass, one can use a passkey controlled by TooBigTech or a Yubikey. I find it great to authenticate directly with my Yubikey (over FIDO2) instead of using my Yubikey to decrypt a password and copying it in a form.

And then there is the part that is completely wrong about security. They say that they "can't trust their phone" so they don't want to keep the passkeys there. But that is not correct: if the passkeys are encrypted and the key is stored in a TPM, then that's effectively similar to having a security key (you have to trust the TPM, just as you have to trust the security key of course).

And then there is the nonsense:

> I can set up KeePass Portable on a USB key, run it in Linux via WINE, place it inside an encrypted VeraCrypt container, copy to any which file sharing service, if I want.

If the device where you enter the password is compromised, then the password will be compromised as soon as you enter it on that device. No matter how much you show off with your funny setup with WINE and VeraCrypt. A password manager doesn't protect against that, so passwords can be exfiltrated as they are used. Whereas a FIDO2 authentication requires the passkey every time. E.g. I need to physically touch my Yubikey for it to sign the challenge. It could be MitM, but it is visible ("I touched my Yubikey and it didn't work, what happened?").

Authenticating over FIDO2 with a security key is strictly superior to entering a password in a field, period.

'An AlphaFold 4' - Scientists marvel at DeepMind drug spin-off's new AI

https://www.scientificamerican.com/article/an-alphafold-4-scientists-marvel-at-deepmind-drug-spin...
1•helloplanets•4m ago•0 comments

AI Isn't People

https://www.todayintabs.com/p/a-i-isn-t-people
1•HotGarbage•4m ago•0 comments

Who Wins When Everyone's Writing Code?

https://predictabledialogs.com/learn/openclaw/future-of-software
2•jaikant•20m ago•2 comments

Taiwan's PSMC Joins Intel, SoftBank's ZAM alternative to HBM AI Memory

https://www.trendforce.com/news/2026/02/23/news-psmc-joins-intel-softbanks-zam-initiative-to-manu...
1•walterbell•20m ago•0 comments

Show HN: Build Your Own CLI Coding Agent in Python

https://github.com/primaprashant/alduin
1•primaprashant•20m ago•1 comments

Rust Debugging Survey 2026

https://blog.rust-lang.org/2026/02/23/rust-debugging-survey-2026/
2•umairnadeem123•22m ago•0 comments

Machine-Generated, Machine-Checked Proofs for a Verified Compiler

https://arxiv.org/abs/2602.20082
1•umairnadeem123•22m ago•0 comments

Machine gun set up close to the University of Tehran

https://www.iranintl.com/en/202602234502
2•ukblewis•22m ago•0 comments

Show HN: Describe a workflow in plain English and builds the multi-agent system

https://www.phinite.ai/
2•PhiniteAI•25m ago•3 comments

Cassandra Complex

https://en.wikipedia.org/wiki/Cassandra_(metaphor)
2•sans_souse•26m ago•0 comments

How to Organize Safely in the Age of Surveillance

https://www.wired.com/story/how-to-organize-safely-in-the-age-of-surveillance/
2•jbegley•27m ago•0 comments

Colt – Describe a browser task in English, get a Playwright script

1•Vipul_Sharma_69•29m ago•0 comments

Anthropic misanthropic toward China's AI labs

https://www.theregister.com/2026/02/24/anthropic_misanthropic_chinese_ai_labs/
1•abdelhousni•31m ago•1 comments

Show HN: Memctl.com: Open-source shared memory infrastructure for coding agents

1•meszmate•36m ago•0 comments

The Looming Taiwan Chip Disaster That Silicon Valley Has Long Ignored

https://www.nytimes.com/2026/02/24/technology/taiwan-china-chips-silicon-valley-tsmc.html
5•blatherard•37m ago•1 comments

Workaholic open source developers need to take breaks

https://www.theregister.com/2026/02/23/open_source_devs_column/
2•abdelhousni•38m ago•0 comments

Show HN: enveil – hide your .env secrets from prAIng eyes

https://github.com/GreatScott/enveil
2•parkaboy•40m ago•1 comments

Huntarr – Your passwords and your ARR stack's API keys are exposed to anyone

https://old.reddit.com/r/selfhosted/comments/1rckopd/huntarr_your_passwords_and_your_entire_arr_s...
1•donutshop•40m ago•0 comments

Why I Hate Anthropic and You Should Too

https://danielmiessler.com/blog/why-you-should-hate-anthropic
4•curmudgeon22•46m ago•0 comments

Show HN: L88 – A Local RAG System on 8GB VRAM (Need Architecture Feedback)

1•adithyadrdo•47m ago•0 comments

Compiler Education Deserves a Revolution

https://thunderseethe.dev/posts/compiler-education-deserves-a-revoluation/
2•azhenley•50m ago•1 comments

Torvalds Drops Old Linux Kconfig Option to Address Tiresome Kernel Log Spam

https://www.phoronix.com/news/Torvalds-Unseeded-Random
2•voxadam•51m ago•0 comments

FDA approves swallowable weight-loss balloon as alternative to GLP-1 drugs

https://www.businesswire.com/news/home/20260223930098/en/Allurion-Receives-U.S.-FDA-Approval
3•sizzle•51m ago•0 comments

The Mind Illuminated [pdf]

https://ia803200.us.archive.org/view_archive.php?archive=/26/items/ebook-buddhism-2/EBOOK%20BUDDH...
2•andsoitis•51m ago•1 comments

From ASPLOS to Orbit: Unikernels Twelve Years Later

https://gazagnaire.org/blog/2026-02-23-asplos-unikernels.html
1•matt_d•52m ago•1 comments

DPDP and AIF Operations: Investor Data Compliance Guide 2026

https://taghash.io/blog/dpdp-and-aif-operations-investor-data-compliance-guide-2026/
1•koolhead17•54m ago•0 comments

Show HN: ClinTrialFinder –AI-powered clinical trial matching for cancer patients

https://www.clintrialfinder.info
1•chncwang•1h ago•0 comments

Show HN: L88 – A Local RAG System on 8GB VRAM (Need Architecture Feedback)

https://github.com/Hundred-Trillion/L88-Full
1•adithyadrdo•1h ago•0 comments

Simulating the hardest Physics Problems in Python [video]

https://www.youtube.com/watch?v=M_OOwhA2fY8
1•chii•1h ago•0 comments

Show HN: We scanned 500 ClawHub skills for security risks – 10% were dangerous

1•yusufjacobs•1h ago•0 comments