I think it's proper. When you release something like this, a raw data dump is the only way to cut out a BUNCH of the "this is modified and falsified" noise.
If planes have just been invented, you need to go somewhere, and you live next to an airport, the idea of flight might come to mind quickly. That's the era we're in with LLMs. It just comes to mind for the HN crowd sooner.
Much better to release the raw stuff; those and derived resources will likely be available in a much more accessible way on public mirrors within a few days.
And if they did put a lot of effort into it your comment would say "look at all the money that went into compute for setting this up". Can't let them win, right?
No it would not. This is an abdication on UI. There is no reason to release a file labeled with a number in 2026 when what the user wants is searchable, categorized content with thumbnails and the ability to export groups. That's web tech we've had forever. The fact that they didn't use their AI or compute to do that with it is just a big fail.
It's more likely to make it take longer to figure out that there is nothing there. If one had clearly indexed and searchable content, we would see it's just garbage.
or distact from the Iran war, or distract from Israel, or distract from corruption... distraction from distractions. We keep buying what they're selling, and then complain the milk is still sour.
Easy with the use of "we" there buddy. Just look at the polling. There are way more people not buying the bullshit, and the numbers keep getting worse as even the faithful are tiring of it as well. So just tossing "we" around becomes offensive as you've now included me into something I will not be a part of.
The numbers have sort of plateaued. There's a ~30% of the population that is all-in on Trump for emotional/psychological reasons, who have very different values from the rest of the population. Where others see malicious incompetence, they see him sticking it to their opponents and are even willing to suffer as long as they perceive their opponents to be suffering more. So although they don't like paying a lot of extra money for gas, they will put up with it for a long time because the payoff is seeing others suffer more. IT's not that Trump created this mindset, although he was able to capitalize on it due to being celebrity; about 1/3 of people are assholes and they're able to use the internet to network and coordinate like any other group. Unfortuantely, they are one of the largest social groups, while opponents have to deal with the friction of coalition politics.
That's fine, but at 30% "we" is doing a lot of heavy lifting. If it was 80% in favor, then maybe "we" could be accepted. Even at the less than a majority winning the election makes "we" difficult to accept.
A lot of people still look to the MSM to define reality for them.
But there's a huge and myserious disconnect between the MSM's reporting of Trump as a Serious Person, and the reality that he's a compulsive liar and fantasist and is seriously ill with advancing dementia.
Without honest reporting, "we" don't have a public voice.
I'm only aware of Avi Loeb, who AFAIK is generally considered a crackpot and a grifter within academia, and his claims about Oumuamua and aliens aren't taken seriously by the mainstream.
I'm not the poster you replied to but it's worth mentioning that there are, unfortunately, examples of more than a few highly-credentialed academics and scientists believing some pretty out there things. Due to such a large sample size, humans being human and tenure being for life, sometimes you're going to get outliers. Plus expertise in one discipline doesn't necessarily generalize to appropriate scientific rigor and skepticism in other domains.
While I don't understand it myself, I've seen a study showing how some scientists can compartmentalize and apply different standards of evidence between their professional life and personal beliefs. In other cases, scientists conducting rigorous lab controlled studies have been deceived by fake psychics doing simple magic tricks (and not nearly as well as a competent magician). For example, Harold Puthoff and Russell Targ at Stanford Research Institute being fooled by Uri Geller. While Puthoff and Targ were trained experimentalists having worked in laser physics, their parapsychology study designs had poor controls and lacked statistical rigor. https://en.wikipedia.org/wiki/Parapsychology_research_at_SRI
As a long-time skeptic, I've learned to avoid broad appeals to authority because relying on "a scientist said..." is ineffective when a true believer can cite a credentialed scientist spouting nutty stuff. In recent years the situation around military assessments of UFO sightings has also changed dramatically. In the mid-2010s, some UFO enthusiasts already in the military managed to work their way into positions as UAP investigators, largely because "UFO Investigator" was a role no serious military careerist wanted on their record. Suddenly, what were once hundred page dry, technical assessments boiling down to "inconclusive" (which no one cared about) became artfully crafted, overly-credulous reports highlighting sensational (but poorly supported) "possibilities." This coincided with a political recalculation from some members of both parties in congress and the White House during the past two administrations to stop fighting the tiny but highly vocal UFO community as it was a no-win battle and instead basically leverage UAPs as a sideshow either for attention or distraction. And it's working.
oh come on! where's that hacker spirit? you can download these and create a site that has them indexed as you'd like using the latest in LLM tech to parse the files and build the site for you. you can then turn around and give us a Show HN
How unhelpful, but how honest. They're admitting that they just randomly release stuff now with no explanation and the public is supposed to make up their own explanation. Terrific.
Hans Reichenbach predicted the LLM in his 1954 book, The Rise of Scientific Philosophy, which is a fantastic read for anyone interested in either science or philosophy. He would not, however, have been taken in by Dawkins's anthropomorphizing. We know this because the entire book is about avoiding assignment of human attributes to the scientific world, along with other false a priori beliefs about science.
Dawkins, whom I greatly admire and have read extensively, would do well to research how LLMs are made and how they produce the results they do, a much more effective empirical analysis than just his 2-hour chat.
This is ridiculous. They say they are not going to transmit electricity back to shore, but of course a data center is not useful unless it moves bits back to shore, using electricity. Were they planning to "sneaker net" the data out of there once the model finished training?
You cannot escape the physics of consuming data (which is stored as electrons) on land by producing said electrons at sea.
Dawkins's books are worthy of study and his insights into biology and social behavior are valuable. But like many scientists, he's unwilling to admit the concept of choice or free will, and this leads to delusions of built-in determinism.
Classical physics can argue about forces moving a ball and quantum physics addresses what goes on inside the atoms of which the ball is made, but neither system can tell you if Johnny will decide to catch the ball or not. You can read 30 books about the philosophy of physics without mention of human will!
Dawkins has fallen into this same trap, "science as religion" in which he believes that a fixed, god-like underlying system controls everything. This fallacy, inherited directly from Western religions, caused problems for all kinds of scientists and philosophers from Galileo to Kant to Einstein. So it's not surprising that Dawkins falls into the same illusion. If you believe there is no human consciousness or will anyway, then machines have "it" too.
An interesting and well-written piece that shows why broadcast TV cannot become mobile TV, and broadcast shopping like QVC isn't mobile shopping like TikTok.
This reminds me of the truism that no railroad company became an airline, even though they were already rich and in the transportation business.
> the truism that no railroad company became an airline
I don't know if that's a truism. A railroad company could have tried merging with an airline. That they didn't doesn't make it obvious that it wouldn't have worked. (Air-rail alliances started becoming a thing twenty years ago [1].)
No railroad company became an airline. The first airline was Pan Am. Railroads ignored air transportation until it killed all of their passenger business and they had to be bailed out and merged into a Federal conglomerate called the National Passenger Rail Corporation, branded Amtrak.
I bring it up as an example of how sunk costs and the fallacy that they'll always continue to produce the same income often prevent industries from adopting a new and fundamental way of doing business -- even when they have the clear lead in that business, as QVC did and as railroads did.
LLMs are being used in all those apps but it's inside the codebase where you're not going to see it. They either use AI-assisted development or they have app features that call an LLM from the inside.
None of those are "ChatGPT apps" which is just a poor and early UI layer over regular chat. No one wanted to use those or make them for other people.
> No one wanted to use those or make them for other people.
but why not?
personally i find it much easier to track my food, workouts, daily diaries etc. in ChatGPT/Claude as I already use them for everything else. no need for a standalone app when the app could simply plugin into ChatGPT.
wondering if it's a UX problem or a more fundamental one like platform risks
It not really even an app. It's just a thread. You'd be better off with something like Cursor where you are reading and writing files on your own drive.
He gave several real examples. The idea is that you have to audit your own agent file and other harness settings. He can't provide an example from your file and one from his doesn't help make the point any better.
Whose pricing is above API rates? Not Cursor. It's 100% at each model provider's published API rate. With a bigger sub, you get it cheaper than that.
Cursor makes a ton of money because the product is great. It's easily the most sophisticated harness out there, and it isn't an IDE anymore. It's an agent dashboard since version 3.
Suffice it to say it's not all idiot money being thrown at them by users.
That's ridiculous.
reply