Hacker Newsnew | past | comments | ask | show | jobs | submit | ncommentslogin

Following the same pattern as youtube-dl. GitHub DMCA is increasingly a chokepoint for open-source scraping tools, even when they don't circumvent any technical protection.

Codeberg is a pragmatic choice — Forgejo-based, hosted in Germany, different legal framework, and no business incentive to comply with marginal takedown requests.

The broader trend is worth watching. Web scraping tools are in a strange legal gray area: the web was built to be machine-readable, but platforms increasingly want to control which machines can read it.


It's kind of amazing how much of those old games was actual logic instead of data.

Feels like they were closer to programs, while modern games are closer to datasets.


That is what I mean, If you think learning “skills” to make visual art for CG, is just learning software or a particular “tech” or “technique”, you cannot make it work in the world of AI… If you just want to make random stuff, just use AI. What I was saying is that the skill of visual art including good pixel art, comes from exposure and learning visual vocab and grammar. Sorry for the previous long winded post (habit from teaching design and CG in Ivy)

Hm, I can actually try the training on my GPU. One of the things I want to try next. Maybe a bit more complex than a fish :)

"real" is the most important and the most precious thing

If chess is a solved problem, think about skating to where the puck is going to be, an interest area a bit further away from relatively easier verifiability such as coding, math, and hard sciences.

Do you have any interest in digital humanities? Knowledge work where verification is still important but not as black-and-white as does the math check out, does the code run.

Do you have any interest in family history or genealogy?

https://vibegenealogy.ai/p/the-genealogical-research-assista...


Why is that hard for you to believe? Do you think having wealth or success eliminates you from sexually assaulting others??? I'm sorry have you read the Epstein files?? Our current president is an adjudicated Rapist found guilty of raping a woman by a jury. 25 women have come forward and publicly accused him of sexual assault. Let me guess you think he is innocent as well? 3 of his cabinet members have sexual assault/pedophilia claims against them. Do you know what the percentage of rape claims being false is? Less than 6%...and thats the high ceiling,the actual percentage is 3%. Do you have any idea the damage being sexually abused causes? Especially by a sibling? It's a lifetime sentence of trauma. Annie Altman was a pre-med student at Tufts. You can read text messages where she asks for $45 to pay a therapy bill and Sam tells her no,you'll just have to work harder or figure it out. When her father passed in 2018 her mother and brothers hid the will from her bc her father left his 401k to her. A year later a lawyer contacted her to tell her about the inheritance. But then her brother Sam hired his mother a lawyer to file a challenge to the 401k. Even though Mr.Altman had been separated for over a decade from his wife,she never signed the divorce papers do thru a law for surviving spouses she has filed to claim the 401k. Sam Altman has been described by numerous associates as a sociopath who is morally bankrupt. He is a habitual liar who constantly pitted employees against eachother. He forced workers to sign ndas stating they were not allowed to say anything critical of OpenAi or him. He was a huge donor to the pedophile in the White House and continues to give him money and support him.People who abuse and assault keep company with those who do as well. Your comment is EXACTLY the type of victim blaming that got us where we are.

nice to see become landing in nightly. does this work well with async or is it purely sync tail calls for now?

Interesting, haven't tired it. I'm afraid Anthropic will be cracking the whip here and there. On the other hand, it makes sense that other providers would move fast on this. Curious, have you tried it with Kimi2.5?

How much training data did you end up needing for the fish personality to feel coherent? Curious what the minimum viable dataset looks like for something like this.

I agree, YouTube search is completely useless when we really need it. Especially with fizzy search!

how are they broadcasting in what seems like near real-time? i don't have a whole lot of understanding on the topic, but if they're seeing the dark side of the moon then i assume they don't have line of sight to Earth. it makes me feel pretty rotten about some SQL queries i have knocking about.

Definitely not for all companies let alone all positions.

I’d be pissed if my code was used for training an AI too but that seems legal thus far…

Only evidence? What evidence do you think normally is present from a sexual assault of a child years before? I would suggest that his support of an adjudicated rapist and accused pedophile who is covering up for a child sex trafficking ring operated by another pedophile that Altman had a known documented relationship with...is a pretty good indicator of his guilt. Altmans lawyers argued this week that because the alleged abuse happened so long ago the case should be thrown out. They didn't say because Altman is innocent,they said bc it occurred years ago. Annie Altman is asking for $75,000...that's it. You can read text messages from Sam Altman refusing to give her $45 for therapy bills. Altman is described as former associates as a sociopath who is devoid of any morals and lies to attain his desires. He is suspected of having a former business partner killed. I don't know how anyone can think his sister would make this up. She was a pre-med student at Tufts. Anybody who suffers sexual abuse from a sibling is going to have ptsd and other emotional & mental health issues.

Yeah you should, ignore these fags. They just don’t want you to compete

its not a competitor to FIX, it is an agent MCP based "sister" protocol. they serve different purposes.

Here’s a tool you guys might find useful. A local search engine for your private knowledge bases, wikis, logs, documentation, and complex codebases.

Instead of stuffing raw documents into every call, you index your data once and query it with simple prompts like “how does X work?” to get grounded, cited answers from your own data. Your main agent can also delegate low-level RAG questions to a smaller local model for token efficiency, while a stronger frontier model handles higher-level reasoning.

That makes it a good fit for setups that pair a local model such as Gemma 4 with a more capable orchestration model. Tokens go down, latency improves, and the whole system becomes more efficient. qi can also run fully offline, so you keep full control over your data, models, and infrastructure.

You can plug in whatever model stack you prefer, whether that is Ollama, LM Studio, llama.cpp, MLX, or cloud APIs, which makes it easy to balance cost, speed, and quality. It also integrates cleanly into agent workflows, including as a Claude Code plugin, so SOTA models can delegate retrieval and lightweight knowledge queries instead of wasting context.


Exactly. And at that point Alice is debugging code she didn't write, using patterns she doesn't recognize, with no institutional knowledge of why things were built that way. That's the real cost of not investing in Alices.

Where is the strongest community of Gramps developers?

I love Gramp's extendibility and would like to get involved. I'd love to talk to some of the folks involved, to learn more about their plans. I see the community at the official site, but I'm wonder if there's more activity elsewhere.


Microsoft has a very coherent strategy, described aptly in "fire and motion" by Joel spolsky 24 years ago: keep changing the API/GUI every 2 years:

https://www.joelonsoftware.com/2002/01/06/fire-and-motion/


Is it so bad if different countries can have different values?

This is a really good article but one of the paragraphs at the end rubs me the wrong way.

> In theory, you can try to preserve this context by keeping specs and docs up to date. But there’s a reason we didn’t do this before AI: capturing implicit design decisions exhaustively is incredibly expensive and time-consuming to write down. AI can help draft these docs, but because there’s no way to automatically verify that it accurately captured what matters, a human still has to manually audit the result. And that’s still time-consuming.

I agree that it's time consuming and we don't have a good solution yet, but my guess is that a huge part of the next 3 years of iteration in the craft of Software Engineering is going to be creating tools and practices to make this possible. Especially as AIs get better at the actual writing of the code, the key failure mode for agentic coding is going to be the intent gap between what you asked for and what you wanted.


The post was written by Claude. The semicolon in the Dune quote proves it: the only possible reason for that to be there is if the 'author' regex replaced all of the em dashes to obfuscate the source. https://boxobarks.leaflet.pub/3misaejnoqs2k

Still didnt release training recipe, data, methodology etc unlike deepseek. Mostly released to get developer ecosystem across their android built in ai. Still good and interesting, but not exactly philanthropic to the open source progress.

Not a good week/weekend for Microsoft on HN. There are currently 2-3 front page threads tearing them apart. I’ve even made my own thread regarding discontinuation of not only support, but runnability of Publisher (what the absolute fuck!). What are their product managers doing?!

Booking a place to stay shouldn’t feel like a gamble.

And yet, we’ve all been there. A stay looks perfect. The photos are polished. The reviews sound great. Then you arrive… and something feels off.

That’s exactly why we built BookYolo.

Today, we’re excited to announce the public beta launch of BookYolo, an AI-powered travel inspection engine that helps travelers understand what a stay is really like before they book.


This may be weird.. but I have been listening to a bunch of extended "save room" ambient tracks based on music in Resident Evil.. Someone under the name of Survival Spheres has a crapload of these on YT-music.. They are all about 10-12 mins long.. and they stay of the way mentally..

Thanks! Yeah, maybe I should cut it off at some point haha! But I think part of the fun is playing it multiple times and trying to improve the number of moves it takes

Same... Source Direct - Approach and Identify

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: