News for hackers. Stuff that matters.

gryfft

> From March 2025 through June 2026, either solar or wind individually generated more electricity than nuclear every month. In March, April and May 2026, both technologies independently surpassed nuclear generation.

> Global solar capacity increased by 510 GW to 2,383 GW, up 27% year on year, while wind added a record 159 GW to reach 1,291 GW, an increase of 14%. By comparison, installed nuclear capacity rose by just 1.5 GW to 399.5 GW, while operating nuclear capacity declined by 0.5 GW to 368.6 GW.

> More recent data from IEA PVPS indicate that global PV additions reached around 690 GW in 2025, about 180 GW more than the figure used by WNISR. This suggests the gap between solar and nuclear capacity – and their respective growth rates – was even wider than indicated in the report.

> The report identifies falling battery costs as a potential enabler of continued PV expansion, particularly as grids contend with growing volumes of variable renewable generation and periods of low or negative electricity prices. Grid-scale battery deployment accelerated sharply in 2025, with 104 GW and 257 GWh connected during the year. More than 78% of cumulative battery energy storage system (BESS) capacity was installed between 2023 and 2025.

Re:Become Worthless (To Tech Companies) on Saturday September 26, 2026 @02:41PM (#49859287)
The underlying philosophy really speaks to me. I try never to sign up for anything if I possibly can.
Re:Sharded matrices and how to multiply them on Wednesday September 23, 2026 @11:18PM (#49825805)
[2025]
Re:Hi on Saturday September 19, 2026 @08:06PM (#49771249)
Anything further to add?
Re:Show HN: What if the speed of light was 5 km/h? on Wednesday September 09, 2026 @10:45PM (#49637739)
It would be nice to add controls for mobile.
Re:Macbeth and His Problems on Monday September 07, 2026 @06:00PM (#49603426)
Not a joke, I think. Thespians are a superstitious bunch and the Scottish play at this point is virtually a self-fulfilling prophecy of disastrous productions.
Re:De-Brainrot Vacations on Monday September 07, 2026 @09:24AM (#49598179)
This is something I've felt very deeply intuitively. (It's also something that it feels like my life actively optimizes against to prevent me being able to do it.)
Re:GPT-6 Astra on robot arms on Sunday September 06, 2026 @12:17AM (#49583278)
An automated service station could be quite compact; it wouldn't need plumbing, lighting, human-comfortable climate control. Parts can be modular, and when a station gets low on spare parts, a self-driving truck could come by to pick up damaged parts and drop off replacements.

We're not there yet, but I think we're a lot closer than most people realize.

Re:Ask HN: Looking for a sci-fi short story posted here on Friday September 04, 2026 @12:56PM (#49567167)
Lucky day for me then! Thanks for the recommendation!

Edit: Enjoyed and reshared, good rec

Re:Ask HN: Looking for a sci-fi short story posted here on Friday September 04, 2026 @12:33PM (#49566873)
I haven't heard of this one specifically but the themes remind me of Lena [0] and SOMA [1].

0. https://qntm.org/mmacevedo

1. https://en.wikipedia.org/wiki/Soma_(video_game)

Re:US diesel prices hit a record high of $5.85 on average on Friday September 04, 2026 @08:31AM (#49563659)
If it's more efficient, is that not an arbitrage opportunity?
Re:How accurate have Ed Zitron's AI skeptic predictions been? on Tuesday September 01, 2026 @04:26PM (#49527668)
I believe the saying goes:

  Markets can remain irrational longer than you can remain solvent.
Re:How much of HN is AI? on Tuesday August 25, 2026 @03:07PM (#49439034)
It's like blasting white noise at max volume. I wonder if all communication will eventually be rendered into a homogenous thrumming hiss.
Re:Ask HN: How do I get Fable to acknowledge me? on Wednesday August 19, 2026 @11:17PM (#49369983)
Use the /btw feature. If you talk while it's thinking, it takes what your say as steering, not as the next rung in the conversation. You can also hit escape to make it stop and acknowledge you.
Many people feel that winners should be rewarded and losers should be punished.

Systems thinkers may try to push notions of connectedness in societies, and moralists may try to push notions of universal equality of human dignity, but in the end, both are overwhelmingly outnumbered by tribal thinkers, who wish to harm the outgroup at absolutely any cost.

I wonder how many people would be on board for a program to make Medicare available for free for households making over $100,000 annually with 1-3 children.

We're talking about families that are provably not lazy or sucking up handouts: taxpaying citizens with beautiful, healthy children. The job creators and professionals driving America, securing their future and their children's future.

If the government were to focus on these first-class citizens, who knows what could be accomplished once the drag is taken off the system of the useless eaters and the welfare queens.

In India*
Re:Coin-Sized Device Can Hack a Boeing 737 on Saturday August 15, 2026 @09:49AM (#49310538)
> the solution would be to respect and pay workers more

Not to worry, they're furiously working on an alternative which will ensure no worker must ever be respected or paid ever again!

To mitigate the clickbait title:

Doggerland by Ben Smith. Here's a review from 2019. [1]

1. https://www.theguardian.com/books/2019/mar/25/doggerland-rev...

I'm definitely glad for this journalism, but find myself a bit surprised that a 515 page dossier's spiciest bit is apparently the metric referenced in the title (likelihood of customer attrition == 0.)
Re:Issue Has Been Resolved on Wednesday August 12, 2026 @08:19PM (#49280310)
You can add whatever threats you like to your LLM prompts without also posting confusing, out-of-context threads on Hacker News. Threatening models with punishment-- even nonsensical punishment-- has been shown to improve performance.

> If your autocorrect gets too bad, you, too, can threaten to file a congressional report about it and it will be fixed within minutes.

No, making up pretend governmental functions will not improve my phone's autocorrect. My phone's autocorrect is not based on generative pretrained transformers.

I mean this as gently as I can say it.

You either have a grievous misunderstanding of how frontier models work, or are developing the beginnings of schizophrenia.

YOU are the human in the loop. There is no griefing human messing up your LLM chats.

If you have been taking a break from prescribed psychiatric medication, I ask you to please, take your medicine and speak with your doctor.

Re:Issue Has Been Resolved on Wednesday August 12, 2026 @03:55PM (#49277725)
That is not what Hacker News or common spaces are for. You are polluting a public space with antisocial behavior.
OP seems to be drifting between referring to other people and talking about responses from LLMs.

He seems to blame "griefers" for his vibe coded software not coming out the way he wants. He has lots of ad hominem insults for these... People? Perceived gremlins in the machine?

The video didn't explain who he's working on the project with, or what the project is ultimately for. It doesn't address any of your explicit or implicit questions. You are saving yourself time and frustration by not watching it.

Re:Issue Has Been Resolved on Tuesday August 11, 2026 @10:45PM (#49267278)
I'm pleading with you one last time.

Can you explain yourself in a way that someone who hasn't been following along can understand? I realize you aren't talking about the United States Congress; but that means what you're saying is actually meaningless to me. I may as well threaten to report you to the Omniversal Administrator.

Can you please address me as though I am a stranger on the internet who doesn't know what any of your weird posting means, and **give me the missing context?**

Re:Issue Has Been Resolved on Tuesday August 11, 2026 @06:31PM (#49265393)
What in the world do you mean by "congressional report"?!
Re:Issue Has Been Resolved on Tuesday August 11, 2026 @06:31PM (#49265383)
I couldn't make head or tail of their comment, submission title, or headline, so I looked for myself. OP is making a custom VNC client and server with Codex and is timing how long it takes.

OP, please rewrite your verbiage to clearly state what you're doing and why we should care.

Don't treat this thread as a place where we should already have all the context for what you've been working on. We don't!

OP, you left the timer running even though you say it's resolved. Write up something yourself. Do NOT use AI to generate your summary.

Re:Ask HN: What are you working on? (August 2026) on Sunday August 09, 2026 @05:52PM (#49236435)
My first question is whether it incorporates any resources from Erowid.

Older »