News for hackers. Stuff that matters.

mmmmbbbhb

Re:Gemini 4 Argon is now available for everyone on Wednesday October 21, 2026 @03:50AM (#50177379)
The gemini app removed flash from free tier, and flash lite is absolutely horrible, like 2023 gpt4 level.
Re:OpenCode 2.0 on Saturday October 17, 2026 @09:29AM (#50131010)
Based just on the description, the desktop app is reinventing dsh. Then again reinventing frontends is a single prompt now, eh
Re:OpenCode 2.0 on Saturday October 17, 2026 @09:25AM (#50130982)
Agreed. Although, limiting it to opencode has been a fool's errand thus far, it's way too easy to "unlock" anyway.
Re:OpenCode 2.0 on Saturday October 17, 2026 @09:07AM (#50130836)
I use bubblewrap, but it doesn't play all too well with client server.
Re:DeepSeek: DeepSeek v4.1 Flash 1017 on Saturday October 17, 2026 @06:06AM (#50129614)
Tavily one is free, too. Easy to use cli with a skill.
DeepSeek harness is for you
Re:Artificial Analysis Intelligence Index v4.2 on Sunday September 06, 2026 @09:33AM (#49586403)
This website lost all trustworthiness with the astra blunder. The benchmarks are worthless.
Re:Zed DeltaDB on Monday August 10, 2026 @06:10AM (#49241680)
I don't see a need for a code editor anymore, though I understand that others feel differently. However, I think zed is going against the tide, and not in a productive way.
Did you ask an ai to critique? Gemini have some points that weren't addressed in the post (which I barely skimmed), here's the summary of summary of summary:

Omitted Variable Bias: Conflates development stage with state size. Developing nations have small governments and high catch-up growth; mature economies have large safety nets and slower frontier growth.

Bad Extrapolation: The power-law model predicts ~36% growth at 0% spending. In reality, 0% spending means a failed state with no property rights or rule of law.

Lacks Basic Controls: Ignores panel fixed effects, reverse causality (Wagner's Law), and bundles productive infrastructure with welfare transfers.

> It should be common knowledge at least among the people building these things that a true AGI isn’t possible with LLMs.

This is funny to me, where'd you get that idea? There's no evidence for that, and models keep getting smarter. I guess you heard some 'guru' say it out loud.

> Cookies can also be queried programmatically,

Misleading statement, essentially untrue.

> Most extensions have effectively full access to your cookies

False.

> There are ofc server set cookies which are not programmatically queryable, but that's not the norm

That is the norm for login cookies.

> and they're still visible via the developer tools,

This is irrelevant to the argument.

Tokenomics is an established cryptocurrency term.
Lmao, Mayor lives in dream land. Who's going to spend their days enforcing such a ban
Re:Transcribe.cpp on Sunday July 19, 2026 @01:57AM (#48965363)
You know English doesn't work like that. The word you're saying only becomes clear with the surrounding context. Eg, 'there' vs 'their'.
Re:Road to Elm 1.0 on Tuesday July 07, 2026 @02:13AM (#48814216)
I am also confused by this, and not just for web, but any dev technology. I personally couldn't care less what the llm is writing in.
Re:Migrating from GNU Stow to Chezmoi on Friday June 19, 2026 @02:19AM (#48595425)
Honestly, I think you'd benefit from vibe coding your own solution. That way it fits you like a glove.
Re:MiMo Code Is Now Released and Open-Source on Sunday June 14, 2026 @08:26AM (#48526566)
I like that they do their thing, but I'd be happier if there was a harness that could be tweaked, trivially, to do what their custom harness does. E.g. would this have been possible to do within Pi, as an extension? Then I could just "pi install mimo" and get the same results. I guess it would be more work on their part, to make sure it doesn't regress when something changes in pi itself.
> If you mean "AI Research," the bar is extremely high (Ph.D. level math).

What? Man, if you haven't a clue, don't simply tell things that sound right.

Firefox has built-in link previews with AI. Hold mouse1 on a link.
Re:Someone made a ton of money betting on Maduro's capture on Sunday January 04, 2026 @09:18AM (#46488173)
It's annoying that it's always Trump's cronies, not "someone."
Re:Exe.dev/ on Saturday December 27, 2025 @05:59AM (#46400898)
$20 a mo seems overpriced.
The actual issue with the sensors was that they'd report highs as lows, and lows as highs.
With chatgpt, your options are wide than ever. Quit your job, spend 3 to 6 months full time studying/contributing to a field of interest, and you've pivoted.

Chatgpt as a teacher is seriously a super power.

Re:AI just proved Erdos Problem #124 on Sunday November 30, 2025 @12:51PM (#46098804)
Meanwhile I have to use ai just to understand the problem statement.
Re:Show HN: Software Freelancers Contract Template on Saturday September 27, 2025 @10:21AM (#45395944)
Frankly my contacts are of form "you pay 50% up front, rest after," and never had an issue. Perhaps luck, but in the first place I'd never pursue legally so I don't care at all.
Re:Do things that don't scale, and then don't scale on Sunday August 17, 2025 @05:07AM (#44930074)
It's crazy that medium.com has an SSR hydration issue that just goes unfixed for years.
Re:Node.js can now execute TypeScript files on Sunday August 17, 2025 @05:05AM (#44930062)
Hadn't heard of that. Thanks
Re:Dokploy is the sweet spot between PaaS and EC2 on Sunday August 17, 2025 @03:08AM (#44929545)
Off-topic, but their monetization scheme is the oddest I've seen.
Re:Solving the Nostr web clients attack vector on Sunday August 17, 2025 @03:06AM (#44929538)
I'd say this is the least of nostr's problems right now.
Re:FFmpeg moves to Forgejo on Sunday August 17, 2025 @03:03AM (#44929520)
I have to say you're exaggerating, ignoring prs isn't such a chore.

Older »