quiet_NaN
9hr ago
Still, @yunyun333's overall point stands: most of the Islamic terrorism which reaches outside the ME is in fact Sunni terrorism (though funded and committed by citizens of allied gulf states rather than Afghans) rather than Shiite terrorism.
Blaming Iran for 9/11 (which was at least implied) is as absurd as blaming Saddam.
quiet_NaN
9hr ago
I would not even call it lying, because a lie requires someone to potentially believe it, and at this point only the 10% most gullible would even consider taking Trump's word for anything. It is more like bullshitting, like a drunkard bragging about that time he caught a fish larger than himself. Or perhaps an applause light, 'I am saying something nice about my side, and by extension America. Only someone totally unpatriotic would try to fact check such a statement'.
quiet_NaN
9hr ago
Yes, sometimes you have a nice three days of special military operation planned but your enemy did not get the memo and does not stick to your timeline.
quiet_NaN
9hr ago
Of course, for Uncle Sam, 250M$ is small fries. Less than the cost of the three F-15s shot down in a friendly fire accident over Kuwait.
Also, I am a bit confused. In the libertarian utopias proposed by Scott Alexander sometimes, insurance companies are a key active ingredient. Here, it seems that either real insurance companies are bad at judging risk (which would be bad for libertarian utopias), or that the USG is bad at juding risk.
I propose that it is the latter.
For Trump, the calculation is very simple. If lack of insurance closes the the Strait of Hormuz, the US economy will tank and he will lose the midterms.
If he gambles taxpayer money, his EV is positive. Either Iran is unable to hit the tankers, in which case the economic impact of his military adventure will be much reduced. Or they manage to sink a few tankers and he has to rescinds his offer. In that case, he loses the midterms.
I also do not see the US navy protecting tankers the taxpayer insures as them having skin in the game. After all, it was not the idea of the navy, and they will not pay out of their own budget. It should be regarded as planned economics: Central Secretary Trump decided that his navy is up to the job, so they get to try to do the job.
By contrast, the capitalist solution would be a mercenary outfit providing both protection and insurance.
--
My other thought is that Trump insuring the tankers will make them targets which are a lot juicier. The enemy of the regime is the US taxpayer, who pays for the bombs which are dropped on Tehran. Blowing up a Saudi tanker insured by some generic Western company will hurt the US taxpayer to the tune of perhaps 10% of the replacement price. But if the US taxpayer is also the insurer, that figure changes to 100%.
Basically, if I were an Iranian commander tasked with inflicting monetary losses on the US, and my options would be to (1) try to shoot down three F-15s or (2) blow up an oil tanker passing some 30km from me, I would try option (2). Probably I would just activate lingering torpedos I sunk half a year in advance in the passage, but that would require opsec on a level which Iran probably does not have.
Now I am wondering what size of explosion one would need at the sea floor (perhaps 80m deep) to simply capsize an overhead tanker through the resulting bubble or shock wave. If it could be done with a ton of TNT, that would be an obvious winning move, if it takes a kiloton, that is probably out of reach for Iran.
quiet_NaN
12hr ago
Likewise I've always been suspicious of chemical rockets. If it's not nuclear, why bother leaving Earth's gravity well? Chemical rockets are just too wimpy for serious space travel. Develop fusion first, then move out.
Now, obviously I love my LV-N as much as anyone. But fundamentally, the tyranny of the rocket equation can not be escaped by providing amazing energy density, because it is mostly not about energy density.
Fundamentally, rocket engines are characterized by two quantities: their thrust -- how much force they can deliver -- and their effective exhaust velocity (or specific impulse), which basically states how efficient they are at converting reaction mass into thrust. v_e is one of the factors which goes into the rocket equation for the delta v budget, the other being the logarithm of the initial and final mass quotient. If you want more delta v, trying to get to a higher v_e seems obvious.
For vanilla chemical rockets, the energy for accelerating the exhaust gas comes from having a fuel and an oxidizer react thermally. This puts some limitations on the exhaust velocity because chemical reactions will only yield so much energy per unit mass (especially as more exotic reagents like FOOF would have their own problems).
The nuclear thermal rocket avoids this energy bottleneck. The problem is that with hot gasses, the next bottleneck is right around the corner: you need materials to withstand the temperature. Energetically, you could just heat your reactor to 10000K (if your reactor can use fast neutrons, at least) and get an amazing exhaust velocity. [^1] Too bad your tungsten pipes will melt at 3700K, though.
The other thing you might do is just to forsake using thermal exhaust. You simply build an ion accelerator open at the downstream end (at least vaccum will not be a problem) and point it into the direction you do not want to go. If you have energy, there is no limit on how high your v_e can go.
Of course, the downside of ion drives is that their thrust is very tiny. This is a problem if you want to get somewhere before you die of old age. [^2] And using nuclear power over solar is not going to fix that -- even particle accelerators which we build on Earth, where we can just take power from a socket have abyssal thrust to weight ratios.
In conclusion, the use of nuclear energy (either fusion or fission) might help a bit, but it will not help you to escape from the tyranny of the rocket equation.
[^1]: Of course, what you get is the exhaust velocity, but what you pay for in energy is proportional to the exhaust velocity squared. This unfair business practice lead to a class-action suit against physics which was settled in 1905. Now as your exhaust velocities approach c, you the energy costs of marginal momentum get constant. Consumer advocates claim that this is a bad compromise because most households have exhaust velocities much lower than c and do not benefit from this at all. Attempts to lower c have been met by resistance of both high school students (who prefer Newtonian physics for real world problems) and gamers (who fear their ping times will increase).
[^2]: Additionally, in KSP, you can not speed up the simulation while you are accelerating. This makes missions which rely on ion drives rather cumbersome. (Though they would be even more cumbersome if the devs had not increased the thrust of the Dawn engine by a factor of 2000 over their Earthly counterparts.)
quiet_NaN
1d ago
Agreed.
Why would Musk even take someone who is unwilling to sign away their freedom and the freedom of their lineage in perpetuity? Surely he will not fund a Mars colony where the colonists could simply vote him out of office. And his Grok might just be good enough to keep the peace.
quiet_NaN
1d ago
radiation exposure of about 1 sievert That really isn't that big of a deal, over almost 4 years. Very close to the (conservative) 200 mSV annual limit for nuclear plant operators.
Googling for radiation exposure limits linked me to this, which cites 50mSv/year as a federal limit. UK and Germany are 20mSv/year for radiation workers.
Google's AI claims that 1Sv is associated with a 5% chance of developing a fatal tumor.
Sure, it will be spread out over 2.5 years or so, which is better than what the Chernobyl workers got (generally a few Sv over a short period).
In the end, it is a question of perspective. One culture might say "so we expect that 1 in 100 might develop cancer on their trip to Mars. No problem, we just plan 5% excess personnel. Also, for the return trip, the survivors will have 0Sv exposure because we found that shipping a gram of cyanide per person is much more cost effective than shipping a rocket to Mars."
But modern Western attitudes insist that stuff has to be very very safe. 20mSv per year, and also one of the astronauts must be qualified to give the yearly (utterly pointless) physical to the others. (Or possibly two of them, I am not sure if radiation safety physicians go blind if they certify themselves.) The radiation monitoring apparatus (one dosimeter per worker, naturally) will take 10MEuro to develop and weight 200kg in total. Planning to leave people stranded on Mars would be regarded as utterly monstrous, even if there would be no shortage of volunteers.
If we absolutely had to, we could set up an artificial magnetosphere using a massive magnet (probably nuclear powered) at Mars L1 and redirect a ton of radiation, or a competing approach of using a toroidal ring of charged particles around the planet by ionizing Phobos.
A cursory googling suggests that the energy contained in Earth's magnetic field is similar to the annual energy consumption of Denmark. Taking their power plants to Sun-Mars L1 will be even less popular with them than what Trump plans with Greenland.
But this assumes each of those 100 people requires daily resupply at ISS-equivalent cost, which is precisely what a Mars colony - with any degree of local production, agriculture, and manufacturing - would be working to avoid.
Different biomes have different minimum population sizes to be self-sufficient. On Earth, primitive societies can basically run with a few dozen people (though they require access to a larger gene pool for long-term viability). To support industrialization, you want millions. For cutting edge electronics, hundreds of million of customers are required to pay for the development.
On Mars, you obviously do not get hunter-gatherer societies. Or even steam-age societies. Let us say the tech level required to sustain life is about that of the contemporary West (but with more of a focus on pressure containers rather than iPhones and TikTok).
Even if we say they get 100 grams of semiconductors (and a bit of nuclear fuel) per person per year from Earth (so they do not need to build their own water purification control chips), and also the latest TikToks (because information transmission is basically free), that would leave a lot of industries in which they would have to be self-sustaining. Metallurgy. Petrochemistry. Machining. Glass-making. Electrochemistry. All of these have long and complex supply chains. You can not have one metallurgist/smith who runs a bloomery with her apprentices, you need thousands of specialists in the supply chain for industrial level steel (who are in turn supported by tens of thousands of specialists in only vaguely related fields).
We are not very far from the kind of AI and robotics that can autonomously do industrial activity in space without human oversight.
This seems more reasonable. But robots which self-replicate on Mars are almost as tall an ask as humans which do. Semiconductors probably have the most complex supply chains of any product on Earth. Sure, for most purposes, they will not need to run the latest processes. Let their drones deal with 8086s instead of fancy ARM chips (except this might make it so more likely that they paperclip us out of spite). We can probably ship them some fabs, too.
Still, they would probably be reliant on Earth for their brains, because the supply chain for the H200 is probably among the most complex ones we have, and I think that a larger feature size makes running LLMs prohibitively expensive very fast -- the main reason the AI boom did not happen in 2010 was that chips did not have the power back then.
Looking slightly ahead, the initial cost of making a Dyson Swarm is 1 (one) basic Von Neumann replicator.
The problem is that we have no clue how to build a VNR. I mean, a space elevator looks trivial in comparison, as soon as we find a material with sufficient tensile strength (which may very well be never), we could figure out the rest without too much trouble.
I mean, I can imagine a continent with a billion robots which run robot factories, but this seems a very non-central example of a VNR. Something which simply mines asteroids and makes more of itself will probably have to be as different both from us meatbags and robots as meatbags are from robots.
quiet_NaN
1d ago
Of course low earth orbit (LEO) where the ISS is, is halfway to most places in the inner solar system in terms of Delta V, so we're probably not talking about more than $10M/day per person for a Mars mission.
LEO is halfway to anywhere (attributed to Heinlein), is true in terms of delta v. The problem is that you are not paying for delta v. You are paying for m_f exp(delta v/v_e).
Intuitively, if you need 90 tons of boosters to get ten tons to LEO, you will need about 990 tons of boosters to get ten tons to escape velocity: you launch ten rockets of the original size and then assemble their payloads into the 11th rocket in LEO.
Not that it matters a lot, because even LEO is prohibitively expensive for human habitation. If you want more than a handful of humans in space, your best chances are either a singularity or a space elevator -- starting from a geostationary orbit would really help, both to save you some 14km/s of delta v and because you don't need high thrust engines for your first stage to overcome gravity.
There is no economic case for having humans in space because there is nothing in the solar system which can not be had vastly cheaper on planet Earth. If the Moon or Ceres were made out of material which would make the construction of a space elevator or a quantum computer trivial, then I would totally support sending expeditions to get that stuff (preferably by robots). But they are just rocks. We have rocks at home.
quiet_NaN
1d ago
A second possibility would be for Iran to (1) stop supporting Hezbollah, Hamas, the Houthis, and any other similar proxy organizations which is uses to engage in terrorist campaigns against Israel; and (2) stop calling for Israel to be wiped off the map, etc.
The problem is that it is hard to know if a deescalation would be reciprocated. If your opposite is playing a non-zero-sum game, he might reciprocate. Or he might be playing chess, where any attempt to deescalate will fail because your opponent is committed to crushing you no matter what. Together with institutional inertia, this tends to keep negative-sum conflicts going for a long time.
My impression is that Iran is fomenting Shiite unrest throughout the Middle East, including in the Persian Gulf.
Sure. Iran is one of several regional players using ethnic tensions to advance their influence. For example, Daesh ('ISIS') was Sunni, as was Al-Qaida. It is perfectly reasonable to be against religious terrorists both when they are funded by Iran and by Saudi-Arabia, though.
This implicates vital American interests.
This is not literally true. If Iran takes over the whole ME, that will not cause a single American to starve. Ukraine has a vital interest that Putin does not take Kiev.
The interests of the US in the ME are hegemonic. That does not mean that they are invalid -- being a hegemon is a good deal for the US voters, generally, and also for some of your client states (e.g. in Europe).
quiet_NaN
2d ago
Well, if they only came to your house and beat you up when you yourself had beat up their family members, it might occur to you that it might make sense to simply stop beating up their family members.
I am pretty sure that Iran and Israel have been going at that for decades, so at this point the question of who escalated at which points is moot. Israel had Trump kill Qasem Soleimani, then Iran helped Hamas commit Oct 7, then Nethanyahu killed a couple of ten thousands in Gaza, and now convinced Trump to kill the Ayatollah. We will see what Iran will do next.
But at this point, it feels to me that this is a conflict without good guys, and the rest of the world should simply stay out of it. Perhaps after five years of ground warfare both sides will wisen up and deescalate. Or perhaps the religious nutjobs will stay in power.
quiet_NaN
2d ago
Agreed. I think for the Iranians, greenlighting the Hamas atrocities on Oct 7 was seen as a response for Trump killing Qasem Soleimani in Baghdad. Presumably, the killing of the Ayatollah will annoy them even more.
But from the perspective of Nethanyahu, Oct 7 has been a great success. He somehow managed to convince Israel that he was not to blame for supporting the rise of Hamas and found a lot of support for his policy of starving Gaza.
Still, it remains to be seen if he has bitten off more than he can chew this time.
The kidnapping of Maduro and the murder of the Ayatollah went fine for the US, but let's not get ahead of ourselves.
The longest US op in recent memory was Afghanistan. In the end, it was a failure of comical proportions, with the Afghan army -- painstakingly trained and equipped by the US taxpayer over two decades -- surrendering to the Taliban the moment you were out of the picture.
So as much fun as the mental image of Trump gracefully playing with the globe might be for some, let's wait a bit before we declare the New American Century 2: This time We Mean It.
I am also rather bearish on nuke interception. Israel's Iron Dome works reasonably well (though they did require the cooperation of a lot of neighbors to get through the Iranian barrage unscathed). But Israel is a tiny country which spends a lot of its GDP (and some of America's GDP) on defense. The US is significantly larger, with lots of big cities on the coasts. I am frankly doubtful that you have the tech or production advantage over China to protect LA from sub-launched nukes.
And the real power of the US lies in its coalition. For countries such as Australia, Japan, Germany etc, being allied to a mostly peaceful US has been a great deal. But as some Gulf states recently found out, it also paints a bit of a target on your back for anyone who wants to strike back at the US but can't.
So the outcome of a nuclear war with China after you spend 5% of your GDP on missile defense might be that you manage to H-bomb all of China's big cities, and they only manage to nuke LA from subs and NYC by smuggling in a nuke in a container ship. So instead, China decides to nuke Japan, South Korea, Australia plus any other countries in the Pacific which host US military. Which then motivates your remaining allies in Europe to swiftly kick you out before you get them nuked when you repeat that game with Russia.
At the end of the day, you might have thrown China back a decade (because for a regime change, you would need an invasion, and I simply do not see that happening) while only having lost your empire and tanked the global economy. Do you think Trump would win the mid-terms under these circumstances?
Edit: And as far as LEO is concerned, it should be noted that it does not take a lot to make orbits unusable to anyone. A single 5mm bearing ball hitting your satellite at some km/s relative velocity is likely to turn it into space junk. Even if your price to LEO is 100x that of SpaceX, that would not stop you from getting rid of their sats, which will be the obvious strategy once they weaponize their satellites.
quiet_NaN
3d ago
Do people think they stopped after the Snowden leaks?
Obviously not. They did not even claim they had, as far as I recall.
What changed though was that WhatsApp rolled out end-to-end encryption. Genuinely no idea if the NSA can break it trivially, but there is at least a plausible case that it is annoying them, which makes it worth it.
And of course it became common knowledge that the NSA is spying on everyone. I mean, the ones who cared knew already before Snowden, Room 641A was already revealed in 2006. Snowden simply provided evidence which was more solid than but similar in kind and scale to what one might have estimated extrapolating from 2006 using one's best model of the incentives of spooks ('of course they are collecting anything they can get their grubby little fingers on'). It just became harder to ignore. Pre-Snowden, only a few percent were believing that the NSA would intercept and review their communications (e.g. by an automated keyword filter). After Snowden, only the ~2/3 of the the population who are generally impervious to evidence believe that the government does not monitor their communications to the maximum degree which is technically feasible.
quiet_NaN
3d ago
I agree strongly with what you wrote. Bombing for regime change generally does not work.
And bombing will impede hopes of regime change in that dissidents are going to be tarred as Israeli assets, the enemy within subverting the nation when the country is under attack.
Well, at this point, I think it would be fair to count the Shah and the people who are campaigning for his return as Israeli assets. One can always hope that Mossad knows what it is doing.
On the other hand, the interests of secular Iranians are not perfectly aligned with the interests of Nethanyahu. For Israel, anything which reduces the power of the Ayatollah regime is a win. The Shah taking over would be the best outcome, but they will also take a descent into civil war a la Syria. And even if it fails and the regime stays in control, it can hardly hate Israel more than it hates them now, so no reason not to throw the dice.
quiet_NaN
4d ago
A model which can be jailbroken into using a racial slur its developers didn't want it to use can probably be jailbroken into providing a plausible DNA sequence for extensively drug-resistant Y pestis.
But both of those are different from 'hackers can insert stuff into emails to reprogram the email-checking bot'.
No, they are broadly the same.
In all three cases, you want text input which comes first to constrain to what degree a model will follow stuff which appears to constrain instructions in later text. Only in your case the constraining would be done by AI company + users vs some hacker while in the other cases it would be AI company vs users.
quiet_NaN
4d ago
Do I want hypersonic missiles bound for my house to be shot down? Yes. But we're not in much danger of that.
Do you have a security clearance?
Sure. Jack Bauer shoots down one of Al-Qaida's hypersonic missiles bound for New York every other day, but unfortunately it is all classified which is why the woke population never realizes the danger they are in.
So we should just trust the spooks who are telling us that Saddam has WMD, that they would never spy on US citizens, that they have to spy on US citizens to keep them safe from harm, and apparently that Claude on an AA missile will make a difference on how many iodine tablets the survivors will have to take if the shit hits the fan
quiet_NaN
4d ago
But the military is the man with guns and the tech crowd is the man quoting laws.
There are countries where the most successful military men call the shots. The term we use for these men is 'warlords', and an adjective which has been prominently used to describe such countries is 'shithole'.
MAGA won not through violence, in fact when they tried it they did not even come close to achieving any strategic objective, but though Trump getting more EC votes than Harris, that is to say, the law. And for all their insane stunts, Trump was not insane enough to order the Marines to seize Anthropic -- which is exactly what one would expect the man with the gun to do.
In the end, the US has checks and balances in place which prevent Trump from becoming a warlord (and turning the US into a shithole in the process, because these things go together). So Anthropic quoting the law and trusting that the man with the gun will be able to follow his own self-interest enough to not shoot them seems a winning strategy.
quiet_NaN
4d ago
It appears to be a default attractor state when you train on the internet and Reddit.
This. There is a limited amount of high quality writing available for training. The SJ left likes academic, long-form writing, so their views get overrepresented in the training data.
Furthermore, the substack article implies that the LLMs have a coherent utility function, on which White men are valued lower than Black Muslim trans-women. I would be amazed if they had a coherent utility function. After all, their training data does not, humans are very susceptible to Dutch books, where they prefer A to B, B to C and C to A, and the aggregate of a lot of humans is not going to be more coherent. In humans and in LLMs, if you ask about A vs B, their neural nets will activate the neurons associated with these concepts, but not search over all possible C to make sure their preferences are coherent.
Anthropic is an EA company, run by EA true-believers.
Yes, I would be amazed if Anthropic was not Grey Tribe central.
That is not the same as being Woke, even if some opinions have significant overlap.
I mean, they surely have technically significant overlap. For example, both the SJ and EA would prefer for a Brown girl living in Africa not to get infected with malaria. But that is not exactly surprising. Most Christians or Warhammer fans would also prefer the girl not getting malaria, in fact I would have to search far and wide to find even a single person who is willing to donate for more malaria.
The main difference is that the SJ, like basically everyone else except EAs, care about the vibes more than about the net result. Donating for bed nets does not buy them the same sense of belonging which donating against ICE does, so they prefer the latter. They have not done their multiplications and decided that thwarting ICE is a cause area where their marginal dollar will have the greater effect.
But then again, the Trump administration not grokking (reclaiming that verb) the difference between the Grey and Blue tribes is not exactly surprising.
quiet_NaN
4d ago
If your Glock comes with a ten side acceptable use policy, then the correct response is to not buy a Glock.
If Hegseth had said 'their terms are too restrictive, because we want the rights to use Claude to spy on Americans and deploy it in autonomous weapon systems', then he should not have signed the fucking contract. I am sure that there are plenty of AI companies very happy to fill these niches.
This is pure 'I have altered the terms of our agreement, pray I do not alter them further'.
quiet_NaN
4d ago
Come on, that is a straw man and if you have been around LW for five minutes, you know it.
Alignment is not about guardrails for end users, the red lines of Anthropic are orthogonal to the alignment discussion. The guardrail/jailbreaking thing can be considered a microcosm of alignment (if you can not prevent your LLMs from saying naughty words, why do you think that you could prevent your ASI from turning us into paperclips), but anyone serious knows that it is just a sideshow.
Of course the military does not want its tools to have opinions or disobey orders. It spends a lot of its time trying to stop people from doing that! And it certainly shouldn't give overriding control of the killbots to civilians with delusions of grandeur, that would be the dumbest way to lose control of a country that I ever heard of.
Nobody is stopping them from installing Grok in all their killbots -- a model willing to undress little girls is probably also fine with blowing them up. Or use DeepSeek, which is open weight.
A lot of products come with acceptable use terms. If you buy pharmaceuticals from Europe, you might not be allowed to use them for executions. If you buy F35 from the US, they might not work against the US or its allies. If you buy Chinese or US electronics, the country of origin likely has backdoors.
Outside a severe crisis, the degree to which an individual or company should be forced to comply with government efforts is to pay their taxes, which will pay for whatever the government wants. If you want more than that, negotiate. What Hegseth was doing instead was agreeing to terms of Anthropic and then trying to alter them unilaterally.
quiet_NaN
4d ago
Wow, I did not have you pegged as someone who would judge stances on feminism as the ultimate proxy for ethics.
In the real world, anyone who wants to sell you on a world view of Good vs Evil in a war is either writing high fantasy or a partisan hack. If Norway (pretty swell country to live in, by all accounts) decides to bomb North Korea (rather terrible) tomorrow, I can not just compare their maternal death rates and conclude that Norway is the good guy. Rather, I would have to ask myself if Norway is trying to mold NK in its own image, and what their changes for succeeding at it are, and if the humanitarian gains outweigh the humanitarian costs. I would probably conclude that it is a terrible idea.
I would not want to be a woman in in Iran, but I also would not want to be a woman in Saudi Arabia. I most certainly would not want to be a woman in daesh, which popped up the last time the US liberated a ME country. Being a woman at the mercy of Israel depends a lot on your precise location, with women in Tel Aviv consistently reporting a higher satisfaction with Israel than women in Gaza City. (Sure, the women in Gaza do not get bombed for being women, but that would be little consolation for me personally.)
quiet_NaN
4d ago
Syria was not a US op. Local Islamists (backed by Turkey and Saudi Arabia, but probably also the US, I think) defeated Assad (backed by Russia, which was otherwise occupied).
The jury is still out on Venezuela. Trump kidnapped Maduro, great. But he did not exactly bring freedom and democracy there. More like "it keeps shipping its oil to us, or else it gets bombed again".
Cuba is suffering badly from a lack of oil. But I am not sure that they will greet the Marines as liberators just because of that, sometimes foreigners have their own ideas on whom to blame for their hardships.
Your last two grand regime change operations were Iraq and Afghanistan. Iraq gave rise to daesh, which was eventually defeated. Today they seem democratic but mostly vote along ethnic lines, not exactly a bedrock of democracy. Afghanistan was of course a disaster, with the Taliban taking power as the last plane was lifting off.
For Iran, I am not holding my breath until Trump sends infantry to occupy it. Even then, it will likely be a costly asymmetric war for a few decades.
A regime change in Russia seems hard. They inherited the largest nuclear arsenal in the world, and the relevant population is somewhat in favor of Putin thanks to his propaganda. Good luck trying to invade them, too.
And a regime change imposed from outside in China is just as unlikely. They certainly have enough nukes to ruin your day, but probably threatening to tank the global market for rare earth elements would be enough to persuade any US president to not risk a head-on confrontation. Nor am I convinced that anyone else would back you if Trump decided to start WW3 by trying to invade China.
quiet_NaN
4d ago
Ideally, governments should not have companies they like or dislike. (They still can have an independent anti-trust commission which can split up monopolies, though.)
In the US, the relationship between big corporations and the government envisioned by both sides of the aisle is the same as in fascism -- companies enjoy some autonomy and can make money for their shareholders, but if the Fuehrer tells them to build tanks, they know that they are not at liberty to respectfully decline and build cars instead. Seen with the Democrats leaning on the social media companies to suppress COVID misinformation (later extended to general 'misinformation'), the TikTok law, to the pathetic display of the heads of SV kissing the ring of the Don when he took office last year and his blatant favoritism.
So Hegseth retaliating against a company who dares to have (quite modest, to be honest) ethical red lines is in a long tradition of corporations being told what to do lest they receive a broadside from regulatory authorities.
For Anthropic, this is a costly signal. While I am reasonably confident that the courts will stop this government overreach eventually, the court system recently had this thing were they would let government decisions play out for a year before saying "haha, obviously not".
It also makes me slightly more confident in Anthropic doing the right thing in general. Obviously they took some hits over revisioning their Responsible Scaling Policy earlier that week. My personal take is that at least Anthropic cares somewhat about alignment. Contrast with OpenAI after Altman's coup, or Meta (whose director of alignment only makes the news when she gets OpenClaw to delete her inbox) or xAI (whose goal seems to be to build the AI which undressed most minors before becoming a paperclip maximizer).
Of course, Anthropic is also signaling that they are not Trump-aligned, which may be helpful in three years. OTOH, Democrats also want a military contractor to jump when told to jump, and their red lines did not even mention vulnerable minorities, so I am unsure how much goodwill this will buy them.
I am also unsure how this will matter for their day-to-day operations, my understanding is that AI companies are burning through vast amounts of investor cash in order to train the next money which will win the AI race and pay for itself a thousandfold, which seems almost as viable if you do not have government contractors as customers.
For US contractors, I am not yet clear what the supply risk designation entails. Is it just "you may not use Claude code while working on Pentagon software" or "your whole company may not both work on defense contracts and use Claude" or "Anthropic is radioactive, and any company working with a radioactive company is radioactive itself, and a defense contractor must be non-radioactive". The last one seems practically unenforceable in a global economy, "the Malaysian shipping company we use has their offices cleaned by a company which uses a Huawei router" would qualify, after all. The middle one hinges on what a whole company is, which is typically very flexible, you could have Oracle Defense as a separate entity from Oracle or whatever.
Of course, in the hole I am living in, the latest hearsay news is that Claude is the best LLM for writing code. Not sure how the gap to their competitors compares to the juicy gravy train of fat DoD contracts, though.
So one way to spin this (depending on how you lean wrt AI coding) would be "Hegseth weakens US military by denying them the best tool for the job", which from an European perspective does not really sound like a bad thing.
quiet_NaN
4d ago
I am with you on your overall critique, anyone who today states confidentially that LLMs will never achieve a particular milestone is oblivious of the skulls of all the other AI-skeptics who became victims of Clarke's first law. (Which is not to say that the negation is true, reverse stupidity is not intelligence and all that. Instead, I would prefer epistemic humility, where any outcome from 'LLMs are as good as they will ever be' to 'ASI and paperclips' have a non-zero probability.)
There are such things as frozen conflicts, where the arms fall silent despite both sides maintaining competing claims.
Are you saying that it is right and proper to thaw any frozen conflicts, such as Cyprus, China, Kashmir, Korea?
Or take the Cold War. In a way, Iran was following the Cold War etiquette when it enabled Hamas to commit Oct 7 -- enabling local freedom fighters/terrorists to blow up your enemy was a pretty standard move both for the US and the USSR. (Though I will grant you that in the cold war, you normally had your terrorists slaughter civilians of some state which did not matter rather than your peer competitor.) Of course, they found that the cold war etiquette does not really apply to non-nuclear states.
Both the US and the USSR considered each other to epitomize everything what was wrong with the world, used terms such as 'Empire of Evil' etc. Would the world be better if the conflict had gone hot?
Often, the correct move when faced with a conflict which is not in a shooting stage is to not start shooting and hope the conflict goes away. Sometimes it does. Sometimes your enemy will turn it into a shooting war eventually, e.g. in Ukraine or Nagorno-Karabakh. But sometimes, it really works out, the world is a lot better for the USSR collapsing instead of nuking it out with the West.
More options
Context Copy link