qlte

That's the usual failure mode of any grid, aggressive load shedding will be used as a last resort to keep the frequency within tolerances.

The renewable aspect is because historically they were installed with "grid following" inverter systems instead of "grid forming" which can maintain frequency even if temporarily islanded in a partition without spinning mass generators to set frequency.

There's nothing intractable about a 100% renewable grid, the technology exists and has already been deployed in new installations.

I love how increased healthcare spending always makes it into the top line of the strongest drivers of growth and example of consumer "willingness" to keep spending.

  Less affluent Americans in particular have been squeezed by higher prices. But Wednesday’s data showed that on the whole, American consumers continued to splash out in the third quarter, with spending on healthcare, food and accommodation rising.

Those same models have already been subjected to the equivalent of probably millions of hours of RL on simulated user messages containing every possible combination of self-harm, torture, murder plots, genocide, child endangerment, CSAM, etc to test refusals.

"Model welfare" seems like an objectively absurd concept given the inherent training process needed to produce a usable instructions tuned model.

...and yet despite similar anecdotes it's still not showing up in the actual data.

It's entirely reasonable for a financial analyst to point out the gap between vibes/anecdata and financial metrics for F500 companies.

The woman in the article was in an ongoing, multi-month chat discussing her medication with ChatGPT.

I don't use it for medical stuff very often, let alone on models within the last year or so. But I have certainly experienced the "drift" that can occur (even on subjects with sensitive guardrails like medical advice) where it gradually bends towards the user's perspective and away from the inflexible canned responses from the system prompt over time.

The language in many of the quotes from ChatGPT included in the article seem like they could be miscategorized as following directives to be supportive/friendly/helpful even though it implicitly was providing (reckless) medical advice at the same time.

Or, the lower barrier to entry crowds out the committed indie devs willing to invest years of their life honing their unique artistic vision for a game that might end up being a hit on Steam (among many other failures). The perceived odds of getting some recognition or even a moderately profitable hit go from "slim but possible" to "hopeless".

Because now their competition outnumbers them 1000 to 1, the competition isn't devs or artists but anyone with a phone, who can be from any part of the world who may not even speak the same language. Many to most of these solo "devs" aren't interested in games at all, and just hoping to rack up sales, ads, hype a crypto shitcoin launch or whatever. These slop games all have flashy attention grabbing visuals, slick product pages, five minute promo videos, with the games boasting 500+ hours of gameplay... of which 99% are recycled borderline unplayable trash.

The indie market for lemons stifles indie devs who struggle to find any audience, with modest dividends going to the most deceptive, scammiest solo sellers churning out 10+ titles per day (mostly 1:1 copycats of GTA/AAA games, psuedo-porn games, gambling games, meme/news cycle one-off games, etc). Middlemen like Steam print more money than ever as purchase volume increases while the indie gaming market becomes closer to the mobile games industry today.

The vast majority of the force multiplier value from AI is captured by the handful of AAA conglomerates who follow the tech monopoly strategy of maintaining market share by constant acquisitions and collectively control the most valuable IP while constantly reducing headcount thanks to models like Genie.

But who knows. It just seems like wishful thinking whenever I hear talk about how AI will empower the little guy to take on Goliaths considering the huge amounts of capital driving AI research and products are coming from those very same giants.

Have you considered the alternative hypothesis that he was in fact hired into upper management specifically to lead a planned pivot to offshoring, so the company leadership can pretend their hands were clean without acknowledging the plan to rank and file.

Good luck, you'll just get sinister vagueposting like "If you know you know" because it's much easier to spread these claims without the burden of actual evidence that could undermine the unfalsifiable "common knowledge".

Why would that matter? They don't disclose info like parameter sizes to begin with, that's just going off rumors/"leaks".

The logical response from companies seeing a hit to their bottom line will be to remove the path of least resistance: e.g. L1 support agents no longer empowered to provide a courtesy discount/refund without additional approvals, gym memberships going back to a certified letter requirement, etc.

Making the overall experience worse for everyone in the future.

The alternative approach leads to those "costs have dropped by 7 orders of magnitude in the last 50 years" style of deceptive graphs popular among people with an agenda to insist AI memory price inflation isn't a real problem.

Yeah I wanted to love the form factor of the Mini and thought the 60 he wouldn't be an issue but was reminded any time I used the device.

Actually surprised to learn Apple did the OLED swap and didn't use the opportunity to bump it up at least to 90 hz like the OLED Steam Deck.

The profit comes from the enterprise customers who have to pay for usage at API rates. Not the subsidized subscriptions, which both OpenAI have designed to be unavailable for large enterprises by withholding IAM and auditing features they need except on the usage based entperise plans.

As an individual I'm still paying for a ChatGPT Plus plan since $20/mo gives me a lot of value and I use many multiples of $20 at API rates per month. But in terms of capabilities, I am entirely satisfied when using my Opencode Go plan with open models like Deepseek Flash/Pro, Kimi and MiMo.

At this point keeping my ChatGPT subscription isn't because of a significant capability gap compared to open models. It's 90% about the value of the subscription priced well below API rates (10% the fun of trying latest models).

If OpenAI/Anthropic dropped their cheap subscriptions or drastically undermined the value proposition I'd be happy using open models exclusively and currently barely notice a difference switching back and forth for the vast majority of tasks.

I'm expecting the subs to increasingly provide less value as the US labs IPO and follow the SV pattern of moving from marketshare capture mode to profit extraction mode. Thankfully I'm already comfortable using OpenCode with open models and avoid lock-in to any specific harness so would be a very soft landing.

...uh, by using the abundant evidence from radiometric dating, geology and archeology that the world is much older than 6000 years old?

Plus astronomical observations used in cosmology to model the state of the universe milliseconds after the big bang which eventually evolved into our locally observable region of space around 13.8 billion (not 6000) years later?

That doesn't prove the dinosaurs were in fact living beings but this isn't necessary to disprove the other assumptions of the alternative hypothesis. By parsimony that leaves the hypothesis that dinosaur fossils emerged by natural processes like everything else we've observed on Earth to date.

Am I missing a joke? Because this is definitely not a good example of a hypothesis impossible to disprove with available evidence.

(If the response is "how do you test whether all that evidence was faked 6000 of years ago to deceive scientists" then we're not actually doing science anymore but games of make believe)

Every other comment on LLM news in 2023:

"this is the worst it will ever be"

I've seen this "yeah well they'll just connect to open WiFI" for 10+ years at this point and have never seen any vaguely credible reproduction.

Best you get is "my friend's LG TV connected to the internet even after changing his WiFI password" vaguely recalled hearsay.

Or the second order retreat-to-motte argument of "well they'll add that to the firmware overnight when ad revenues drop if enough people stop connecting to the internet".

Which is a great example of what makes him dangerous: with the exception of Sam Altman he's arguably the single person most responsible for the current capabilities arms race.

Anthropic has pushed the frontier harder and faster than OpenAI or anyone else in the last few years. Beginning with Anthropic's founding to quickly raise huge amounts of cash with a pitch to investors to rapidly leapfrog ChatGPT in scale, and culminating with Mythos/Fable which carved out a whole new tier of super models.

He's always talked big about Anthropic prioritising safety and their "Responsible Scaling" but his words just simply don't align with reality while they keep going bigger, faster, until only OpenAI could briefly and half-heartedly claim near parity at the top (while cutting corners in safety because the existential threat to their market position is far more pressing to any corporation than some abstract risk for humanity).

Whether it's wilful cognitive dissonance, cynical opportunism, financial greed or something else, that kind of self-assured moral superiority paired with his real world record of self-interested decision making for Anthropic's market position is a big red flag.

I think it's generally considered good etiquette on HN to disclose your employer when jumping into a thread to argue on their behalf.

Sure, but in evolutionary timescales a hundred years of selective pressure from newly abundant plastic waste might produce an organism optimized to exploit the niche from human landfills and ocean trash in a breezy 20 million or so years.

Adding a USB 3 Gigabit adapter to my Onn 4k Pro box was a massive improvement. I wanted to get the Onn model with built-in Ethernet but later discovered it was 10/100 and more consistent than the built-in Wifi, but much slower top speeds.

Plugged in a 15 dollar gigabit adapter and now I can stream anything instantly from my LAN with super low latency, even with huge uncompressed media.

Highly recommended upgrade, for sticks without extra USB-3.0 port, best option is a cheap USB-C dock for tablet/laptop with Gigabit ethernet. Amazon sells various versions for <$20 these days.

If you're not as concerned about the tracking/privacy aspect of the unused Google launcher (or use DNS blocklists), you can keep the default launcher enabled and just intercept the home button to go the custom launcher.

That's what I always do and except on one cheap Hisense with a shitty built-in Google TV, there's no lag or stuttering and I see Projectivy immediately without ever triggering the default launcher (unless I want to open it for some reason).

> This is entirely too much effort

It really isn't. I've done this on like 4 different Android TV devices and it's like 30 minutes of effort at most and you only need to do it once.

If the alternative was a totally open Linux box you'd almost certainly spend way more time than that tweaking config and stuff (even if you'd mentally categorize than as "voluntary" time for a hobby and therefore not actually wasted).

I'll happily take that insignificant one time setup for a 100% ad-free experience (including Youtube) with a customizable launcher I can adjust to my liking on a 4K Android TV stick that cost me $39.99. And automate basically any function via ADB, including remapping the branded Netflix/HBO/etc buttons on the remote (I have a "Free TV" button that now goes to .... Stremio which I still laugh at occasionally).

I couldn't do any of that on a $199.99 Apple TV. An N100 Mini PC would cost 6x as much as I spent on an Onn stick, plus I'd need to buy a remote and would have to work around a keyboard/mouse centric UI and couldn't navigate with HDMI-CEC either.

I can't even think of anything I'm unable to do with my Custom Launcher/Sideload/Remote ADB Android TV setup like OP. It's just about the perfect couch-based interface. It does what I tell it without distractions and fades into the background completely like any good UI should, so I can focus on what I'm actually watching instead.

There is zero evidence of this happening. There are tens of thousands of people who have posted about devices doing sketchy things who monitor their network and yet this has never been demonstrated.

Anything you own could also have a secret 5G radio built in spying on you, but that's reasoning via Russell's Teapot and pointless except to baselessly spread paranoia.

I just saw a similar post from Anthropic which also landed like a thud. The deep need of OpenAI/Anthropic to churn out flashy announcements today to with upcoming IPO is in serious conflict with the realistic pace of biological or medical research, let alone successful clinical applications. They would produce many cool leads like this which ultimately turn out to be useless before landing on the one that makes it to human clinical trials (and may fail there too).

A previous blog post from Anthropic used the mention of "CRISPR-like" to activate a mini hype cycle by association with a well-known groundbreaking discovery. Without that kind of manipulative hook these announcements are pretty thin gruel and below the bar of thousands of Pubmed papers that never made it into public view.

Yeah I have one for Frigate on a Raspberry Pi. Has worked well but definitely showing its age performance wise. I got a cheap mini PC (right before prices like quadrupled) which has about the same performance on an iGPU). I wouldn't be surprised if Frigate was the funnel for like 90% of Coral purchases or more.

...then the next predictable stage in the starve the beast cycle: "Why does CISA even need to exist? When was the last time you heard them do anything useful at all? Might as well shut them down and use a contractor in the private sector who knows how to get shit done and save taxpayer money on pensions."

($400 million no-bid contract goes to Oracle paying 5x as much per person, misses deadline for every key milestone, after one more $200 million extension claims government terms were impossible to meet and never delivers anything useful)

Well it's from a prediction market so could just as easily flip to like 5% a few minutes after a suitable story in the news breaks.

Wish this was basically any other source than the Polymarket X account that loves to post misleading sensational content for engagement.

EDIT: Found a better source than a gambling site's social media.

Kinda funny remembering how many HN comments I've read in the last few months insisting "Anthropic is profitable now" because of those cherry picked numbers for one quarter.

https://www.reuters.com/business/anthropic-recorded-net-loss...

At least Google lets me install Projectivy Launcher on the device I paid for so I never have to see a single ad on my TV. And use ADB to customize those branded OEM remote buttons for HBO/Netflix/etc to do what I want instead.

Plus sideload apps like Smarttube to watch ad-free YouTube that would never be allowed on Play Store/App Store for either Android/iOS.

User hostile stuff out of the box is much less aggravating to me if I'm also provided the means to bypass it. Even if it's mildly inconvenient to setup, it's still vastly better than the alternative.

In practice that would create even more ineffective veto points that already result in stasis and dysfunction in the US Congress.

For more than a decade the US has been stuck operating by renewing the same budget more or less with minor modifications at random intervals of time depending on the political mood. It always gets approved eventually after occasional pointless, wasteful shutdowns. But nothing actually changes from those veto points, just more routine dysfunction and grandstanding.

Same with debt ceiling. Another veto point that gives leverage but has zero effect on actual spending. It probably sounded like a brilliant idea to control spending in the 70s or whenever it originated. But is entirely ineffective at the purpose of addressing the deficit.

Self expiring legislation would very likely end up the same way. Just another pro forma procedural roadblock that wastes time and makes the legislature even less effective at real problem solving instead of posturing.