Is it? Personally I only found the story interesting because of the contents of the message, not because GPT-6 Math Scoopa was the one to stand on everyone else's shoulders and get its grubby little fingers into the cookie jar. Computer does thing that computers should ostensibly be good at is a middling goalpost – what's more intriguing is the implication that the model might have stolen private info or scooped someone else to get there once again.
> Personally I only found the story interesting because of the contents of the message, not because GPT-6 Math Scoopa was the one to stand on everyone else's shoulders and get its grubby little fingers into the cookie jar.
What you read:
> All science and research that relies on prior research is FAKE.
I'm not sure you'll have much success with your shoelaces given such poor reading comprehension, but good luck goofball.
We used to always "scrape" the wayback machine for any sort of news article we actually paid to consume. I was absolutely shocked by major news sites making very important edits to an article without any sort of editorial notice!
Sadly this sort of thing is probably not really possible anymore, but I can't really blame anyone for making this sort of decision. I can't imagine how much more traffic they get now vs 2021 when we were doing this.
I was thinking the same thing... paid access for high volume users or scrapers could actually help fund the non-profit. Maybe let website owners decide which scrapers are allowed to use their content, or allow them to get paid for use of it. If news and other sites were getting paid, maybe they could go back to optimizing for good content instead of clicks.
I think that would get into murky water really quickly with the rights holders (/content creators) not exactly being thrilled the Wayback Machine is essentially monetizing their IP behind their back.
It wouldn't be behind their back, like I said, "Maybe let website owners decide which scrapers are allowed to use their content, or allow them to get paid for use of it."
Interesting, in all the years I have never noticed that IP has 2 meanings (well probably more...) Yeah, I am an engineer and usually try to avoid the legal BS. Although I hate that AI has made stealing legal if you are big enough.
But anyway, no, I wouldn't keep finding reasons. I donate to them every year already. Somebody asked if I would be willing to pay and my answer was "yes, but".
It would need to be improved because certain aspects of it suck right now, not only the error this post is about. They only need go as far as their forums and github repos to see the community feedback.
Not at all. I donate to Internet Archive, but we're talking here about paying for unobstructed access to but one part of their service: Wayback Machine. Two different things.
100% of the people who write "I would pay, if..." or "I would pay, but..." are people who are never going to pay even a dime. You might be the exception, and sorry for bunching you up with them. You have paid already by donation.
I think that we should all pay when asked for things which we find useful, even if they aren't perfect. If nobody else is offering anything, then we have to take what's being offered. When there's a market, more providers will begin offering their versions.
I mean we already paid for the article from the source itself. I guess I'd expect a better "diff" source from them, but if they dont even update the article itself, i guess i wouldn't expect a paid service to have those updates either?
ah, i think i misunderstood your original post. if you mean the wayback-machine/arkive, then I suspect it'd be hard to justify? You are essentially paying a third-party source to validate that diffs didn't go through on the source material.
with llms, at some point it probably becomes easier to use your paid api connection to manage your own cached version yourself?
>I was absolutely shocked by major news sites making very important edits to an article without any sort of editorial notice
They didn't use to, this has become a thing over the past few years as MSM outlets have completely given up on journalistic standards, including editorial ones.
I think it's VERY project specific. If you are looking for anything technical at all, then keyword search almost always does better (in my experience). I'd actually recommend starting with keyword search, and then expanding with embeddings after you have a better idea of what your users are trying to determine.
I have a Google account that I had tied to a domain I originally purchased from google (when they had the .dev offerings), and now that it's been purchased by square space my ability to use AI through that account is in a bizarre state. Even my free gmail account has more AI offerings, and I'm not allowed to pay for improved AI offerings on my custom domain & google workspace account.
I know this is probably a pretty small edge case, but it is a bit frustrating. Any other provider lets you sign up with an email and give them a payment processor/card, but because google wants me to only use their unified workspace for signing up, I'm completely locked out now.
I appreciate the comparison to Timescale, DuckDB, and ClickHouse. It's interesting to see the differences between a more specialized db, like Quest, and DuckDB/ClickHouse.
I've never seen the level of anti-scientific, straight misinformation as I have with data centers. I genuinely think it's to the point where the antivaxxers have a stronger scientific backing for their stance.
It did not feel organic at all. It's to the point now where that initial seeding of ideas has gained legitimate traction, but the initial burst of anti-datacenter content was wild to see in real time.
I had a lot of fun playing the game. I also made a local solver to see how far you can theoretically get, and it's definitely infinite if you play perfectly. Working on the local sim / solver was almost as fun as playing the game :)
I started by pulling the game source locally and doing some initial testing with a few solvers. I ended up having Claude make a 1-for-1 perfect sim of the original game using rust with some optimizations. The original engine was able to check around 180 moves/s and the rust port was hitting 2,150 moves/s.
After that, I tested a few solvers. Single Player Monte Carlo Tree Search seemed promising, but didn't do as well as I expected. A simple beam search does appear to go infinitely pretty easily.
When you say play perfectly do you mean by exhausting all possibilities? Or is there some way to deduce ahead of time whether the RNG will put you in an impossible situation.
The way the game is seeded, it's 100% deterministic. So if you can scan enough potential moves / second, it's very easy to infinitely play the game at a fast rate.
Shockingly, Claude was able to find the exact random generators that the js library uses, list out all of the differences between the languages that might impact the replays (like rounding being treated differently), and include those differences in the simulation. Claud then back-tested on a huge number of games using the original implementation as a golden set generator.
Considering that 2 new pieces are spawned for each captured one, if you keep progressing you will exhaust the real estate, sooner or later. The only way to go on infinitely is not capturing, but that means not scoring points. Hardly satisfying, though.
For example today's board can be played infinitely by moving between a8 and b7. Zero points, infinite game!
Ah yes, I specifically mean you can keep earning points infinitely. After the board is full, it resets the game, and you get points for each piece on the board. I usually stop the simulation after it clears the board 10 times or so.
reply