And a third thing: OpenAI's thuggish behavior. Threatening a researcher with reputational ruin and demanding one of their competitors be denied proper credit is unacceptable, and we know OpenAI did both (unless Buckmaster is simply lying. which I find unlikely).
>many have "Use this data to improve" setting on.
Quite irrelevant. We are talking about academic misconduct, not IP law. It would be misconduct for OpenAI to publish without proper reference to prior art even if that prior art had been explicitly committed to the public domain 1,000 years ago.
I think you have to be careful with things like the bayati scale besides just being in tune. The quarter-tone is meant to be tension, you don't want to stay on it too much.
That your comment and article will be subject to the ranking and processing of a machine authority and there is nothing you can do about it. So you might think you’re being flagged or shadow banned because someone downvoted you or otherwise. But no, it was an automated process that did it.
If it makes the author happier they can ask dang what is the prompt used for automated HN processing.
Of this work? We're taking it all the way to shipping a distro with full hardware support on all released Macs and later to upstream Linux. We've taken significant care to respect Apple's copyright and publicly document all our processes. We don't think this will take too long, our development velocity has been pretty good, see: the GPU driver.
im not swe by trade, in fact I would love it if what you said was true because I could consolidate various kinds of businesses into one holistic infrastructure and demolish all the competitors.
Location: Charlotte, NC, USA
Remote: Yes
Willing to relocate: No
Technologies: TypeScript, Node, Python, Next.js, Stripe, SQLite/Postgres, LLM APIs (OpenAI/Anthropic), agent orchestration, distributed systems debugging
Résumé/CV: https://gg5533.github.io/Sami-Habbal-Resume.pdf
Email: samihabbal5@icloud.com
I work on the failures that only show up under real traffic — retries that cause double effects, calls with no deadline, agent loops with no ceiling, and state that disagrees with what the payment provider thinks happened.
Public and checkable: a fix of mine was merged into changedetection.io (33k stars) by the project owner — https://github.com/dgtlmoon/changedetection.io/pull/4390 — their path-containment guard had no test, and disabling it left every existing test passing. My regression tests for their LLM retry client were merged into the branch of a second, still-open PR there.
The bug I'm most known for is one I wrote myself: a lease that passed 78 tests and would still have let a slow provider call overwrite a completed order and re-charge the customer. A second model caught it before it shipped: https://github.com/GG5533/stripe-fulfilment-kit/blob/main/CA...
Looking for contract or fractional work, 10-30 hrs/week. One bounded fix at a fixed price is how I'd start with anyone.
I mean i can reference a lot of other things which are still happenin in the USA or do you know want to argue that united states of america takes good care of its people?
Description:
The arguments against AI in 2026 are the ones we made against cloud in 2011, and we lost that one. On lock-in, ownership, and running your own models.
An article written by me, feel free to message me about it here.
>At the moment, it’s unclear whether CAMP has a written agreement for 3331 17th St. The organization, however, does have a verbal agreement with all the landlords/property owners where they have murals throughout the alley.
They claim to have arrangements with all property owners. Verbal arrangements are just as binding as written ones, though generally not a good idea as they are harder to prove.
The right way to look at it is that LLMs help us to maximize the utility of the entirety of current human achievement/knowledge by discovering obscure connections in it.
And? Why do you even the comment on a social critisism comment? Just ignore it, feel happy when you buy yourself the next stupid thing while you feel offended seeing your fellow citicen.
The social critisism is, that people in the united states of america (were apple is home), let people die while also buying unironically $800 watches as a mass consumer product.
so it tops out at sonnet 4.5 as the largest model it will use. the harness gives you no choice in the model so you have to hope it doesn't decide your task isn't complex enough and auto-route you to a smaller/weaker model which you then go round and round in circles with where a larger model would've 1 shotted it
We can have real arguments, but lets be clear, comparing a $800 dollar watch you use for lisening to music and paying is not the thing to compare to something you actually need to live.
And a third thing: OpenAI's thuggish behavior. Threatening a researcher with reputational ruin and demanding one of their competitors be denied proper credit is unacceptable, and we know OpenAI did both (unless Buckmaster is simply lying. which I find unlikely).
>many have "Use this data to improve" setting on.
Quite irrelevant. We are talking about academic misconduct, not IP law. It would be misconduct for OpenAI to publish without proper reference to prior art even if that prior art had been explicitly committed to the public domain 1,000 years ago.