Will not one of you try to steelman this decision? Or do you truly, fully believe the entire US government and intelligence complex, supported by roughly 50% of your compatriots, are warmongering baboons?
If assume the goal of the US administration is to really limit the weapons of mass destruction
then why the previous peacefull attempts were thrown out? I remember Iran agreed to limit the production or ivite the people to watch their actions, the US sanctions were even lifted in 2015. But then later reiplemented by the US for some reason.
unrelated to all of this, I don't understand what's the problem is with a country developing nuclear weapons. US can nuke Iran with ICBMs anyway if they try to use them?
Israël is threatened by Iran. Iran has been working on nukes. There have been negotiations but they haven't been definitive yet, and Iran has never been a trustworthy negotiating partner anyway. That is why this President ditched the last round of agreements.
That's the steelman. Reality is that half my compatriots are warmongering baboons.
I am saddened by your gullibility. Your first instinct is to trust this administration? Who has repeatedly showed utter contempt for the very idea of truth, the constitution, the rule of law, and science, merely because half of American voters are brainwashed?
This administration's arguments do not deserve to be steelmanned.
I think it's possible it's a good decision (As with most wars in the middle east, I think hubris is playing a heavy role and we're underestimating the risks involved) but I think I'm in the minority and this is _not_ supported by 50% of my compatriots.
I don't know if by 50% you're talking about left vs right, but I'm center left, and a decent number of center-left thought leaders support this action. I think the people who support this are a relatively narrow 25-30% of remaining neo(lib|con)s in the center, and the more left and the America First right crowd hate this. My guess would be loosely speaking Trump's base hates this more than the typical HN poster. Tucker Carlson for instance will be way more against this than anyone here.
Claude is the best of these models/services out there in my experience so far, so no surprise there. It's a company with leadership that hold principals they stand up for -- huge disappointment to see the government going out of it's way to hurt it. So bizarre, and deeply anti-American.
Too bad government devs will have a much harder time using some of the best tools out there.
Despite the complete Archie Bunker energy in that post here is something interesting from the post:
"Anthropic better get their act together, and be helpful during this phase out period, or I will use the Full Power of the Presidency to make them comply, with major civil and criminal consequences to follow."
with major civil and criminal consequences to follow.
I’d love to get the reaction if all the VC investors who “reluctantly” had to support Trump because Lina Kahn was just too meddlesome, and they had to support a candidate who would be hands off with big tech.
That was in the cards since Hegseth's initial announcement, when he brandished the option for Anthropic to be put on the same level as China and Russia, i.e. the same level as State enemies of the US. Got to give it tho Anthropic's CEO, I had first though that he'd give in faster, looks like he's still clinging on.
Also see what happened to Joseph Nacchio in the early 2000s [1]:
> He was convicted of 19 counts of insider trading in Qwest stock on April 19, 2007[2] – charges his defense team claimed were U.S. government retaliation for his refusal to give customer data to the National Security Agency in February, 2001.
Unfortunately, I think the same thing is in the cards now for Anthropic's CEO, that is if he doesn't choose to play ball.
The part of Dario's response I found funniest was pointing out the inherent contradiction in the DoW threats. Somehow, they could both be a national security threat as well as a national security necessity. Not by doing something differently - at the same time, to the same people, in the same context.
They won’t be brandished as a supply chain risk because the supply chain members are good at lobbying and they like Claude.
As an example, Amazon is a defense contractor and uses Claude heavily internally for development. They are also major investors in Anthropic. Amazon would not want Claude to be banned from use on developing AWS services that may be cross sold to the government. Multiply this by every defense company that uses Claude (eg anduril and Palantir).
They could totally try and punish Anthropic executives of course. That seems likely.
This shouldn't surprise anybody at this point. This is how Trump behaves on a daily basis. He thinks that he can direct the federal government to do literally anything he wants and operates on pure retribution.
So many people wrote think pieces about how Trump couldn't possibly be a fascist because fascism involves state takeover of corporate power. Hm...
I am honestly surprised he is being this nice given he could use the war powers act and eminent domain to just seize the company and conscript their employees. I am sure someone will say that is not legal but when has that stopped anyone.
What I don't get is why they really want it. I can't think of a worse platform in general to do anything with combat that is, anything in a data-center. I would take the Quake 3 Arena engine on a new ultra insane mode combined with a tiny self hosted model to detect humans, uniforms, vehicle make and model plus a simple Friend or Foe over all the big AI platforms any day. Add an optional feed from an encrypted meshtastic like network to sync nodes using pre-defined Ostiary like commands. Ultra-fast, light-weight on all resources, decentralized.
The enemy could neutralize all the big platforms in one day by simply activating a few dozen of the sleeper agents in the US or a couple High-Altitude Electromagnetic Pulse (HEMP) deployed by a few stratospheric balloons as China flew over every major military installation in the US. Adding to this having a network dependency to the big platforms from autonomous weapons is an extreme vulnerability. Any design that depends on a central command is a single point of success.
> I am sure someone will say that is not legal but when has that stopped anyone.
it has routinely stopped them as the courts have already struck down countless of nonsense by this administration, and they rely exactly on this bluster every time they try something else.
The issue is that even though courts work slower than a president with a smartphone eventually it will all get sorted out and they know this, which is why some people falling for this shock and awe behavior is so silly.
The issue is that even though courts work slower than a president
So ... not stopped the president. Make a move, eventually ruled naughty, shift to another move, ruled a no-no, take an alternate path, rinse repeat. How does one fix the courts or is it working as intended?
> when agents started to compile the Linux kernel, they got stuck. [...] Every agent would hit the same bug, fix that bug, and then overwrite each other's changes.
> [...] The fix was to use GCC as an online known-good compiler oracle to compare against. I wrote a new test harness that randomly compiled most of the kernel using GCC, and only the remaining files with Claude's C Compiler. If the kernel worked, then the problem wasn’t in Claude’s subset of the files. If it broke, then it could further refine by re-compiling some of these files with GCC. This let each agent work in parallel
This is a remarkably creative solution! Nicely done.
There's parts of LLVM architecture that are long in the tooth (IMO) (as is the language it's implemented in, IMO).
I had hoped one day to re-implement parts of LLVM itself in Rust; in particular, I've been curious if we can concurrently compile C (and parse C in parallel, or lazily) that haven't been explored in LLVM, and I think might be safer to do in Rust. I don't know enough about grammers to know if it's technically impossible, but a healthy dose of ignorance can sometimes lead to breakthroughs.
LLVM is pretty well designed for test. I was able to implement a lexer for C in Rust that could lex the Linux kernel, and use clang to cross check my implementation (I would compare my interpretation of the token stream against clang's). Just having a standard module system makes having reusable pieces seems like perhaps a better way to compose a toolchain, but maybe folks with more experience with rustc have scars to disagree?
> I had hoped one day to re-implement parts of LLVM itself in Rust
Heh, earlier this day, I was just thinking how crazy a proposal would it actually be to have a Rust dependency (specifically, the egg crate, since one of the things I'm banging my head against right now might be better solved with egraphs).
One thing LLMs are really good at is translation. I haven’t tried porting projects from one language to another, but it wouldn’t surprise me if they were particularly good at that too.
as someone who has done that in a professional setting, it really does work well, at least for straightforward things like data classes/initializers and average biz logic with if else statements etc... things like code annotations and other more opaque stuff like that can get more unreliable though because there are less 1-1 representations... it would be interesting to train an llm for each encountered new pattern and slowly build up a reliable conversion workflow
This is the proper deep critique / skepticism (or sophisticated goal-post moving, if you prefer) here. Yes, obviously this isn't just reproducing C compiler code in the training set, since this is Rust, but it is much less clear how much of the generated Rust code can (or can not) be accurately seen as being translated from C code in the training set.
I will say one thing Claude does is it doesn't run a command until you approve it, and you can choose between a one-time approval and always allowing a command's pattern. I usually approve the simple commands like `zig build test`, since I'm not particularly worried about the test harness. I believe it also scopes file reading by default to the current directory.
Will not one of you try to steelman this decision? Or do you truly, fully believe the entire US government and intelligence complex, supported by roughly 50% of your compatriots, are warmongering baboons?
reply