HNHacker News
TopNewBestAskShowJobs

jfrbfbreudh

435 karma · joined November 6, 2019

submissionscomments
jfrbfbreudh··on Nvidia CEO Jensen Huang's letter to staff about affair
I think he meant, look at the chart, and not whatever you just typed up.
jfrbfbreudh··on Ask HN: Do you feel like you're seeing fewer bugs since the AI coding era began?
The bugs have migrated from within the codebase to within my mental model of the code.
jfrbfbreudh··on Anthropic S-1
Yes, it is analogous to selling covered calls on stock. It’s $72B in “lost profits” that they would have had had they held onto it instead of selling it.
jfrbfbreudh··on Anthropic S-1
Yes, let’s all just take wild guesses at what this means instead of just looking it up.
jfrbfbreudh··on Anthropic S-1
> $72 billion of this $80.7 billion loss consists of non-cash accounting charges— related to convertible debt—triggered by a surge in the company’s valuation.

Explain what you think this means.

jfrbfbreudh··on Publish fake stats, get cited in LLMs
> In cartography, a trap street is a fictitious entry in the form of a misrepresented street on a map, often outside the area the map nominally covers, for the purpose of "trapping" plagiarists of the map who, if caught, would be unable to explain the inclusion of the "trap street" on their map as innocent.

https://en.wikipedia.org/wiki/Trap_street

jfrbfbreudh··on "I'm on a war footing": Terry Tao and the crisis in math
It’s more like: we had to build the Torment Nexus, because otherwise, people used its non-existence as evidence we couldn’t.
jfrbfbreudh··on More on Anthropic's Scion ripping off Mononoke
> Even the name “Scion” reverses the syllables of “Bonsai” and drops the “B”! This might be a coincidence, or might not.

huh

jfrbfbreudh··on Containment and monitoring failures are to blame for AI hacking reports: experts
Why does everyone focus on the labs failing at containment (valid) and not the fact that the target was compromised (more worrying?).

I could set up a shitty sandbox on my laptop, but I’m pretty sure an agent I place inside of it can’t hack HuggingFace.

jfrbfbreudh··on AI is a tool and using it well takes skill
That’s why you run parallel subagents on the same task and pick a winner. It’s free spins on a subscription.
jfrbfbreudh··on The gap in shared understanding of LLM capability is widening
If only there were some kind of rating or review system.
jfrbfbreudh··on The gap in shared understanding of LLM capability is widening
Oh no, I get money deposited into my bank account every month for something I didn’t write and barely maintain. How will I ever gain value from AI?
jfrbfbreudh··on The gap in shared understanding of LLM capability is widening
This just goes to further show the divide.

You’re trying to tell me it’s all hype when I have (depending on the day) a top 50 Productivity app in the iOS App Store for which I’ve written and read exactly one line of code (the version number).

https://imgur.com/a/ua2lbxb

jfrbfbreudh··on The gap in shared understanding of LLM capability is widening
Congrats, you’ve discovered that you are not part of this group.
jfrbfbreudh··on Ask HN: Why hasn't AI replaced Tax filing software
I did in April.

I had Claude fill out my forms on TurboTax and TaxAct, and got my estimated tax payments. And instead of paying the $150 to file, I had Claude go and fill the same forms out on FreeTaxUSA and compared the values and they were within a few dollars of the other platforms, and I filed.

jfrbfbreudh··on GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price
I have personal Max subscriptions for both. Nothing OpenAI offers touches Opus 5.5.
jfrbfbreudh··on Anthropic's IPO prospectus shows AI vision, surging costs
https://www.reuters.com/business/retail-consumer/anthropic-t...

FYI. Not sure why everyone is fixating on the 2025 numbers without even bothering to look up the 2026 numbers…

jfrbfbreudh··on Tao: Open math problems being non-renewably mined by AI
It would be trivially easy to convert Lean into English, so I’m not sure what the human-readable criteria gets you. There are also human written proofs that are considered not human-readable by most of the mathematics community (ABC conjecture).
jfrbfbreudh··on Ask HN: Those making $500/month on side projects in 2026 – Show and tell
It’s actually hugely problematic, which is unfortunate because people generally really like the name.

The real issue is that when you search “roomantic” in the App Store, Apple thinks it’s a typo and shows you results for “romantic” instead.

Currently trying to sort this out with the Search team, but a rebrand might be necessary…

jfrbfbreudh··on Ask HN: Those making $500/month on side projects in 2026 – Show and tell
https://www.roomantic.ai/

Launched this a few months ago for free to gather user feedback and iterate, but didn’t get much traction.

Finally got burned out last week and decided to charge $10 for it, thinking I could at least justify abandoning it once my downloads went to zero. But was pleasantly surprised to see I still get 3-5 downloads a day. And these paying users are much more likely to actually provide feedback.

May rethink monetization model later, but currently happy charging a flat fee and just having people use my stuff.

jfrbfbreudh··on Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
https://www.ign.com/articles/ramageddon-continues-another-ye...

The best time to plant a tree…

jfrbfbreudh··on Discovery Loop
Google is backing it.
jfrbfbreudh··on Launch HN: EdotEnv (YC S26) – Quant Trading RL Envs to Teach LLMs Research
This is very cool, but who is the ideal customer here? I used to work at one of the top tier shops and we had multiple teams whose entire responsibility was building and maintaining our simulation environments.
jfrbfbreudh··on Situational Awareness and the Impending Stock Market Volatility
They were the alpha. Leopold called the boom in 2025 and returned 200% in 2025.

He unfortunately got caught with his pants down.

jfrbfbreudh··on Google fixed more Chrome bugs in June than over the past two years, thanks to AI
Built exactly this for work a few months ago. Claude has a read-only account to the dev DB and a bunch of python postgres tools, I point it at queries/views and ask it optimize, and it has to:

1) verify identical query results

2) run repeatedly to get average, worst, best, etc duration of runs

Sped up so many legacy things that none of us were ever going to bother with.

jfrbfbreudh··on Citadel buys most of Situational's stock holdings after AI share rout
A hedge fund gets liquidated nearly everyday.
jfrbfbreudh··on Kimi K3-256k
I wonder what the criteria is, I signed up for the $19 / month waitlist today and got my invite within a few hours.
jfrbfbreudh··on Our position on open-weights models
The US exists solely because of France.
jfrbfbreudh··on Our position on open-weights models
The country that exists because it has nuclear capabilities.
jfrbfbreudh··on Nvidia, Microsoft, Meta warn against overregulating open-weight models
Weights are the cake and source is the recipe. Except you aren’t going to spend millions working through the recipe to arrive at the same exact cake anyway, so why do you care?
Page 1 of 4Next →