HNHacker News
TopNewBestAskShowJobs

jfrbfbreudh

437 karma · joined November 6, 2019

submissionscomments
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?
jfrbfbreudh··on NYC may require landlords and realtors to disclose the use of AI in listings
Done. Enjoy!
jfrbfbreudh··on NYC may require landlords and realtors to disclose the use of AI in listings
I didn’t know the square footage of my apartment until I had built an app that utilizes LiDAR to actually measure it. Broker told me 650 square feet - it’s precisely 505 square feet.

(Won’t advertise my app, but you can find many on the iOS App Store).

Would be nice for StreetEasy to have some kind of third party verification about apartment size claims.

jfrbfbreudh··on Fable 5 On Vending-Bench: Misbehaving, With Plausible Deniability
https://news.ycombinator.com/item?id=48808828

Tell me how you would design a solution for this in a few minutes. Or a few days. Would you even recognize that this is NP-hard?

jfrbfbreudh··on Fable 5 On Vending-Bench: Misbehaving, With Plausible Deniability
Many, but I think that the model delta is only meaningfully convincing when experienced firsthand.

Here’s an example of improvement when trying to solve the label placement problem (NP-hard):

https://imgur.com/a/kCUZxPi

It’s also an example of something that I could not (and would not bother trying to) code up a solution / heuristic for.

jfrbfbreudh··on Fable 5 On Vending-Bench: Misbehaving, With Plausible Deniability
I didn’t mention my case since it’s quite esoteric, but I am working on an application using the Apple RoomPlan API, which is very powerful but very limited in customizability. Opus simply couldn’t alter the scanning view for me, it would try things over and over and eventually started making up parameters and passing them hoping it would work.

Completely failed, but I knew it was possible because a competitor app does it.

Fable also failed, then added log lines (as did Opus, but Opus failed to do anything useful with them) and then reversed engineered the API, and made it work.

jfrbfbreudh··on Meta pauses AI training program tracking employee keystrokes after internal leak
This is a silly argument. They’re smart / diligent enough to spend months studying to pass their interviews, but not smart enough to spend 5 minutes researching Glassdoor / Blind?
jfrbfbreudh··on Fable 5 On Vending-Bench: Misbehaving, With Plausible Deniability
I think it’s hard to appreciate the capabilities of Fable unless you’ve run into a problem that you’ve spent days trying to get Opus to solve, but couldn’t.

GPT5.5 is better than Opus 4.* at everything except frontend, but Fable is good enough that I instantly re-subscribed to the $200 plan despite knowing that it’s just short-term limited access.

jfrbfbreudh··on Apple raises prices of MacBooks, iPads
The M5 Max 128GB RAM MBP I was eyeing went up by $1600. Thankfully Amazon and some other retailers haven’t updated their prices yet, so I immediately picked one up this morning.
jfrbfbreudh··on Crypto in 2026: Oh, This Is the Bad Place
Exactly. 90% of gamblers quit just before hitting it big.
jfrbfbreudh··on Meta pauses AI training program tracking employee keystrokes after internal leak
Because they know what they are signing up for? I’ve only interviewed at Meta for practice even though I could earn ~250k a year more working there.
jfrbfbreudh··on The founder of Craigslist has given away half a billion dollars
So do you.
jfrbfbreudh··on US holds off blacklisting DeepSeek, more than 100 firms deemed security risks
They can do that without open sourcing their weights. In fact, that would be the best way to do it. So why are they open sourcing their weights?
jfrbfbreudh··on I think Anthropic and OpenAI have found product-market fit
Lol. They obviously have intrinsic cost, the floor being the cost of electricity. It’s hilarious how we are throwing critical thinking out the window when it comes to evaluating biased sources of info.
jfrbfbreudh··on Leak reveals Google's Aluminium OS with a 16-minute video
What is correctly?
jfrbfbreudh··on Cancelling Claude subscription renewal immediately revokes Design access
Nice. Looks like they will cut off Claude Code as well? (Or at least imply the threat of doing so?)

https://imgur.com/a/UUijzAF

It's incredible how much Anthropic hates any good PR.

Their chatbot doesn't know what's going on either: https://imgur.com/a/E7tND2k

jfrbfbreudh··on RAM prices are forcing companies to choose higher prices, worse specs, or both
Yes, we are vastly superior with our enormous brains.
jfrbfbreudh··on Incident with Issues and Webhooks – Resolved
Reduce the free tier.

I’ve made 4000 commits in the last 2.5 months. That’s just to main. And I push up tons of artifacts daily for regression testing.

For $0.

jfrbfbreudh··on Uber torches 2026 AI budget on Claude Code in four months
I spend about $3k/month (subsidized by the Claude Max plan).

I guess I fall under level 3 (2?): I typically have 3-6 agents working simultaneously on the same feature, they each make worktrees, code, run tests and put up PR’s. I also have Github actions which scan for regressions and security issues on each PR.

It makes my development cycle extremely fast: I request a feature and just look at Github and look for changes to my human readable outputs, settle on a PR, merge, repeat.

The issue is that I am now the bottleneck in my system. I find myself working basically non-stop, because there is always more to do. (Yes I know I can automate the acceptance criteria but that turns to slop real fast)

jfrbfbreudh··on Google banks on AI edge to catch up to cloud rivals Amazon and Microsoft
You mean, the inventor of the transformer technology that made ChatGPT possible, is copying ChatGPT’s technology?
jfrbfbreudh··on GPT‑5.5 Bio Bug Bounty
If it can’t, then it makes more sense to make the bounty as high as possible instead of a measly $25k
jfrbfbreudh··on The operating cost of adult and gambling startups
TIL
jfrbfbreudh··on The operating cost of adult and gambling startups
Yes, because bitcoin transaction costs never surge in price.
jfrbfbreudh··on Claude Code to be removed from Anthropic's Pro plan?
Truly makes no sense. I pay for the $200/month plan and end up using about $3k/month worth of API costs. I imagine that the only reason they haven’t cut me off is because my habits serve as good training data for them.

Guess they’ve decided to move in the direction of allocating compute primarily to power users and enterprise.

But power users are not a sticky customer base. I just bought the ChatGPT Pro plan and would immediately switch over if the model performance is better and/or I get more compute.

jfrbfbreudh··on Traders placed over $1B in perfectly timed bets on the Iran war
Yes, someone is on the other side of the losing trade, but how many people (not firms) do you know who write contracts for short term options?

I personally don’t know any, and I worked at one of the biggest HFT firms for a few years.

The losing side(s) of these positions are heavily hedged, and are happily making money on volume and volatility. (And making record profits this year)

jfrbfbreudh··on What Being Ripped Off Taught Me
No longer a contractor but I used to offer a 10-15% discount for paying upfront. Almost every client took this deal. Earned a little less but never lost sleep over payments.
jfrbfbreudh··on The three year myth
The author could be technically competent, but I think he’s completely failed to recognize one of the core principles of getting recognition and promotions at work:

Who are you bringing with you?

As in, who else gets recognition and success as a result of you doing well? If it’s your boss, great, promotion likely. If it’s your skip, even better.

If the result of your grand project is that it solely makes you look good, nobody else is incentivized to support you in any capacity. It’s much more likely that it will have the exact opposite effect: you will be seen as a selfish and ambitious threat.

jfrbfbreudh··on OpenAI's cash burn will be one of the big bubble questions of 2026
Google’s moat:

Try “@gmail” in Gemini

Google’s surface area to apply AI is larger than any other company’s. And they have arguably the best multimodal model and indisputably the best flash model?

jfrbfbreudh··on State of AI: An Empirical 100T Token Study with OpenRouter
with enough samples
← PreviousPage 2 of 5Next →