HNHacker News
TopNewBestAskShowJobs

smallerize

1,049 karma · joined August 14, 2013

submissionscomments
smallerize··on Gemini 4 Carbon
But Gemini (the app/website) lets you set the Thinking level for Flash or even Flash-Lite. And Flash with Low thinking is the default.
smallerize··on Gemini 4 Carbon
It was only in beta for 5 years. 17 years ago.
smallerize··on Tell HN: I know Firefox has update when it starts to glitch
What version are you on now?

Edit: I ask because releases are scheduled in advance, you can see the calendar here: https://whattrainisitnow.com/calendar/

smallerize··on Genie 4
I don't think so, and in fact an em dash would be much less confusing.

audio, e.g. music

audio — like music

smallerize··on Solar power is becoming bigger than nuclear
You only vote once every couple years. Political movements mean less waiting.
smallerize··on New iPad mini
If your phone is choppy when scrolling a list on a 60 Hz display, it's not going to scroll smoothly on a 120Hz display.
smallerize··on Nikon photography contest winner loses title after scientists spot signs of AI
That one is dead, and not flagged dead. Maybe email the Contact link at the bottom of the page and ask what's up.
smallerize··on Nikon photography contest winner loses title after scientists spot signs of AI
I still don't think the disqualified photographer was intentionally cheating. He was pretty open about his process, and Nikon took some time to determine whether it was within the rules or not. https://arstechnica.com/science/2026/10/winning-nikon-small-...

Edit: an earlier statement from Nikon said that he was complying with the investigation and "At this time, we don't see that any rules were violated" https://www.bbc.com/news/articles/ck4gjn1yzprno

smallerize··on Dream-1
It only takes a few upvotes to get any story to the front page briefly. And on Saturdays there is just less competition.
smallerize··on Sam Altman: 'I No Longer Think We're Gonna Be Able to Just Patch All the Bugs'
Notably, those didn't need a ton of patching over time. I mean relatively speaking. We've known how to build and deploy defensible software for a long time. We might just have to stop using the projects that need constant patching to keep up, and migrate to ones that are designed better.
smallerize··on The alien space of science: Coherent but cognitively unavailable directions
Damn, this was the killer app for semantic spaces I'd been waiting for since Word2vec.
smallerize··on Every AI model has the same "controversial" take on humans
AI thinks atheism is "controversial"
smallerize··on I no longer care to work on any Ruby gems so I've archived all my repositories
They mean it's technically possible, but you wouldn't want to.
smallerize··on Subscribe and Save items have a price increase
Is this the first time you have ever seen a price increase on a subscription item? It's pretty routine.

This probably isn't about Covid (although Covid is still killing and disabling people). The prices that wholesalers pay for goods (Producer Price Index) went up a whole percent just last month, bringing their prices to 6% above where they were a year ago. So not all of it is price-gouging.

We already had one disastrous Trump administration. Why did you expect a second one to go any better? You elected this government with your eyes wide open.

smallerize··on Chrome and Edge are removing XSLT: what breaks and what does not
The HTML spec actually includes how to parse broken HTML. https://html.spec.whatwg.org/multipage/parsing.html#parse-er...
smallerize··on Chrome and Edge are removing XSLT: what breaks and what does not
The main one is that even a slightly broken XHTML document will fail to render. HTML is much more forgiving.
smallerize··on Bleached corals thrive within days when touching healthy neighbour
I think the article says they just haven't tested other cases. I wouldn't expect the symbiotic microalgae to be that picky.

Edit - The article makes Okkef's point, I just missed it.

smallerize··on Solar surpasses nuclear power globally for first time
China's power generation is still >50% coal, so I don't know if I would call renewables "overbuilt" yet.
smallerize··on Solar surpasses nuclear power globally for first time
Denmark's power generation from wind is around 20TWh/year, Norway hydro 140 TWh/year, and Swedish nuclear 50 TWh/year.

China's wind power is at 1,000TWh/year, hydro 1,500 TWh/year, nuclear 326 TWh/year.

(Stats mostly from statbase.org which is a couple years out of date.)

So on the one hand, it's less likely that all of the wind power in China will lull at the same time. Then on the other hand, their hydro and nuclear can't absorb nearly as much % fluctuation. Maybe they could make up the difference with batteries though.

smallerize··on US federal budget deficit jumps 12% in one year to nearly $2 Trillion
$1 billion/day way the initial estimate, but it was never actually that high. CBO estimates $38B through August 1, or ~ $7B/month. https://www.cbo.gov/publication/62756
smallerize··on Plex Media Server Unauth RCE Exploit
https://zmain.info/blog/plex2shell

Some of those individual bugs are egregious. A guard that doesn't work at all? A /player/timeline endpoint that doesn't check the token value?

smallerize··on Woman arrested after blocking delivery robot in Center City Philadelphia
I would be surprised if these robots were taking many jobs. Aren't they mostly piloted by people?
smallerize··on Wanted to download an ISO, ended up writing a kernel exploit
What I mean is, using LLMs for search and coding, they return the wrong answer all the time, unpredictably. Often if I redirect them, they will agree that they were wrong. So even though the knowledge that I need is in there somewhere, it doesn't get applied reliably. I don't want that experience for my infrastructure, where the model usually gets it right, but it just goes wrong a noticeable fraction of the time.

Do you know what I mean?

smallerize··on Wanted to download an ISO, ended up writing a kernel exploit
The randomness is biased, so it's usually fine. But sometimes the output is gibberish. Ask your LLM what sampling temperature is.
smallerize··on "We're not going to make rules that say you can't use tools for X"
Most projects don't have "lieutenants" that triage incoming patch sets. I totally get why projects smaller than the Linux kernel ban AI code out of self-defense against the tidal wave of crap.
smallerize··on Wanted to download an ISO, ended up writing a kernel exploit
I'm being very literal about the randomness. The language model outputs a distribution over all possible output tokens, and the chosen output is sampled from that distribution. The process intentionally includes output that is not the top-ranked token according to the model.
smallerize··on We are entering a period of overlapping crises
The term "polycrisis" started getting popular in 2023.
smallerize··on Wanted to download an ISO, ended up writing a kernel exploit
I understand that it works most of the time. I just want to be clear about what the stakes are.

"Because I have config backups/know how to unbrick?" This is the important part for me. I'm not going to gamble with more than I can afford to lose.

smallerize··on Wanted to download an ISO, ended up writing a kernel exploit
I'm at a loss why you would put a random number generator in front of such a sensitive operation. It's literally a matter of time before the output sampler chooses a somewhat-less-probable string like the one xatt posted. Or just deletes your stuff.
smallerize··on Thunderbird for iOS
I don't understand how scrolling is so bad. There's some kind of frame-in-a-frame effect where both frames try to scroll at the same time, and it just rapidly blinks one of the scroll bars on the edge.
Page 1 of 18Next →