> there are better models that are state of the art now like Nemotron and Parakeet both by Nvidia
Is parakeet state of the art? It always transcribes speech fragments for me, like if I stutter and say "m-m-m-map" parakeet will dutifully transcribe "m m m map". Which I guess could be a good thing or a bad thing depending on what you want. Whisper does not do that however.
> As for offline, yeah, my Internet web page doesn't work offline.
That's really too bad honestly. What about your users who want to use your site when they're offline? Sympathy with those users is why I made sure that my site works well when offline.
Improving service is much better, but the relative scales here are crazy. She options are "completely eliminate cost" or "slightly improve service in one two-mile stretch". Granted, 10 million trips a year go through that two-mile stretch that was improved, but that's just a small fraction of all bus trips in SF.
If we had to stick to transportation-specific improvements, I would use the money to eliminate bus fare for a year. (muni generates about $100M in fare, so really it would be enough money to eliminate fare for ~3 years.)
Unlike subway fare, bus fare doesn't prevent n'er-do-wells from getting on the bus (they just ignore the fare and get on anyway), and it probably increases ridership and gets some people out of their cars, which reduces congestion (public benefit and therefore deserving of government subsidy).
We definitely need more metro, trams, bus routes, and bike paths. I especially want the latter because I bike to and from work every day. However, I have to wonder about the cost of these projects. In SF, we recently spent $346 million on the Van Ness Improvement Project, which added 2 miles of bus lanes. I just can't see how that's a good use of funds. If someone gave me $346 million and told me to improve society, I could think of many better options than adding two miles of bus lane. So a prerequisite to the whole idea seems to involve solving whatever is making these projects so expensive.
As a user I don't really appreciate sites that do this as they tend to have worse latency for UI interactions and less pleasant page transitions. And of course, they are unusable offline
Do you pick your databases based on how quickly they compile and how many dependencies they have? I normally chose based on factors like performance and reputation for reliability
There's an odd idea that increasing the number of educated people in the country makes it harder to find jobs. Where is it easiest to find software jobs? Probably san francisco. You'd think it'd be easiest in some small town in idaho where you'd be the only software developer, right? You'd have no competition! But instead it's the easiest where all the other software developers are. The causation is two-way. Software developers go to san francisco because that's where jobs are, and companies start in san francisco because that's where software developers are.
That's agglomeration effects, and it disproves the idea on its own. If america is the place where all the talented engineers and scientists go, that is a very very good thing for engineers and scientists in america. Or you are free to go try to be an engineer or scientist in estonia, and see if you make more money due to the lack of competition. (Most top engineering and science talent in estonia leaves for other countries, which by your logic should be amazing for the remaining domestic talent.)
It's easy to say that now, but it used to be that all mainstream typed languages had absolutely terrible type systems that got in your way as much as they helped
It just doesn't make that much sense to me. "And also the entire movie staff that chose floors with seven pointed stars for the bank" - so the entire movie staff is antisemitic? And what is the connection between a seven pointed star and Judaism / antisemitism anyway?
> and cunning, greedy creatures with long, hooked noses that run the wizarding bank
I don't think they're portrayed as greedy? Although I'm willing to be proven wrong if anyone has textual evidence otherwise. They are also mistreated by wizards in the books, which I think portrayed them sympathetically. I remember scenes where Harry notices their mistreatment and thinks it's wrong
I can believe that the origins of this trope are antisemetic, but I don't see very much reason to think that any antisemitic mindset was in JK Rowling's head when she wrote it.
I don’t. although I tried to get it to tell me how to take crack cocaine, and it wouldn’t. But when I was taking grey market peptides, claude walked me through reconstituting, dosing and administering them. I assumed it would behave similarly for other drugs, but it doesn’t seem to
I'd guess opus refusals are not an issue for 95%+ of people. Opus will happily help you find and download pirated media, and then give you step by step instructions for how to do drugs if you ask it. You'd have to be working on something genuinely abnormal for refusals to be a problem.
I would never want to compel someone else to buy services from me if they had a better option available. My goal is to make money by contributing to society, not force others to pay me for services they don't want from me. If AI makes it so programmers aren't needed anymore, well, I had a good run and made good money while it lasted. I'll find something else productive to do.
But the WASM is integrated into the file. So what scenario would happen where the decode fails? They wrote a file, wrote a wasm decoder, and didn't test the decoder against the file they just wrote? I think it would be a natural step as part of the publish pipeline to run the decoder against the file and compare the output against the source to ensure correctness.
The basic premise of "have a factory that produces reactors small enough to ship" always made sense to my uneducated mind. Is there a flaw in the idea? It seems like it should be much cheaper than if every reactor were bespoke
Can you point to an example of this happening in the past? Where a supply shortage leads to price increases and "record profits", and the price never goes back down?
Downloaded it and giving it a try. I love the aesthetics and how colorful it is! It is also pleasant to use and fast, although my personal preference is TUIs that don't break scrollback (claude code actually does a good job at this).
> Anthropic claims Mythos is in a class of its own, the evidence corroborates this and the government believes it.
They didn't release Mythos, they released Fable, which was Mythos + a classifier that detected potentally-dangerous prompts and blocked them. Everyone who used it noticed how aggressive the classifier was. It would trigger constantly over totally innocent stuff.
Apple messed up one thing about the iPad, which made me never use mine and eventually give it away. Basically, my iPad would die in a couple days if left unplugged. Because I only want to use it about once a week, that means I have to leave it plugged in all the time. Of course I or someone else inevitably wants to plug something else into that charger, so the iPad gets unplugged and forgotten about. Then, in a week when I actually want to use it, it's dead, and I use something else. The result was, I literally never used it.
Some cantons in Switzerland use a system called biproportional apportionment. They say "okay, party A got 25% of the votes, so they must get 25% of the seats. And district 1 accounts for 30% of the population, so they must get 30% of the seats." And so on. Then they do some matrix math to determine who gets what seats in which districts. The method ensures each party gets the right total number of seats while each district gets its predetermined seat count. You could in principle extend the method to handle three or more dimensions.
> Then why did curl only find one new vulnerability thanks to Mythos
Maybe there weren't that many serious vulnerabilities in curl? It's like asking why it didn't find any vulnerabilities in fn main() {println!("hello, world");}.
> When we ran other frontier models through the same harness, they found a fair number of the same underlying bugs, and in some cases they got further than we expected on the reasoning side too. Where they fell short was at the point of stitching the pieces together. A model would identify an interesting bug, write a thoughtful description of why it mattered, and then stop, leaving the actual chain unfinished and the question of exploitability open. What changed with Mythos Preview is that a model can now take those low-severity bugs (which would traditionally sit invisible in a backlog) and chain them into a single, more severe exploit.