Government is a huge ship where everything happens slowly. If you fire all sailors at once, you won't notice the difference immediately. The phantom ship continues to sail.
Then engines start to heat up and fail. And if there's an obstacle along the way, it's game over.
This is a dangerous game played by morons full of hubris.
On June 16 YC organizes an "AI Startup School"; the first speaker is Elon Musk:
> The good thing about these massive ships is that they take a loong time to sink.
Faster than people realise. The CIA was surprised by suddenness and swiftness of the collapse of the USSR — yes, there were other predictions that it would collapse imminently, but such are made of every country and organisation all the time, of those who should have known best (the Russian government on the inside, the CIA on the outside), it was a surprise.
> If my guesses are correct, it’ll work just long enough for another administration to take over to deal with the fallout.
Assuming things have not deteriorated too much by then that it can still be fixed.
Further, if it completely breaks down by then they may get the blame, even though it was not their fault.
This also assumes that an administration with a different philosophy gets in: if things are still "fine", another bunch of GOP deconstructionists could get in and continue things. In addition, DOGE is not about cost-cutting but about ideological purging, and so what's left of the civil service could act to sabotage any 'recovery efforts'.
> The massive ship named USSR sank in a matter of months
There was a decade of dysfunction and ossification before the USSR collapsed within a couple months
A decade of low oil prices (affecting exports), high defense spending (increasing deficit), societal unrest due to the resurgence of nationalism, and slowing productivity all happened quietly over the 1980s before catalyzing into collapse.
US productivity is going into things like un-avoidable advertisements from your new Jeep when you come to a stop, embedding dark patterns for everything, and spending all R&D on a tool to replace all human workers. A nation can totally be held together heathily under today's 'productivity'.
> decade of low oil prices (affecting exports), high defense spending (increasing deficit), societal unrest due to the resurgence of nationalism, and slowing productivity
Good thing the US hasn't .. oh.
(the oil price is actually OK as is the trade balance, so far, but social cohesion is fraying and people are always complaining about productivity)
Many of these things have been happening in the US for the better part of a decade (or longer).
Did the USSR have an entire administration rapidly and deliberately tearing down their government, or was it more of the cracks in the foundation finally giving out?
Just because the former took a decade doesn't mean that this will require the same amount of time.
I guess it could be argued that it had been in the process of sinking for a longer time, but I don't know enough about the history of the USSR to assert that being the case
Such resets are sometimes followed by civil wars, economy crashes style "there is no food for tomorrow" and a generational trauma. As the one who has been through it, be careful what you wish for.
> I honestly prefer that outcome. At least that’d be a reset instead of this infinite downward cycle.
If the USA gets a repeat of the USSR collapse, you're looking at an independent Texas and California (and perhaps Hawaii?) within a few years, 50% GDP loss, proper hyperinflation, infighting over the nuclear arsenal being under federal authority or the authority of whichever state it happened to be physically in at the time, and a 6-7 year reduction in life expectancy.
> Government is a huge ship where everything happens slowly. If you fire all sailors at once, you won't notice the difference immediately. The phantom ship continues to sail.
Happens to business too. People don’t realise how long these entities can circle the drain.
I feel like businesses aren't actively trying to dismantle themselves, but rather the decay, due to neglect/hubris/greed/..., builds to a critical mass over time. This administration is actively trying to dismantle as much of the government as fast as possible. I think the fall could happen much faster than people expect.
> Nuclear security is highly specialized, high-pressure work, but it's not particularly well paid, one employee told NPR. Given what's unfolded over the past 24 hours, "why would anybody want to take these jobs?" they asked.
The people who will be hired in those jobs in the future are the ones who can't find any other job, and don't care one way or another. It will be a catastrophe of epic proportions, but hard to notice at first because it happens so slowly. Which makes it all the more dangerous.
The EU doesn't have control of defense policy - that is the mandate of individual European states.
The only EU member with nuclear weapons is France, which has it's own domestic nuclear ecosystem.
More critically, public sector hiring is not the same as private sector. Working for a foreign government's NatSec apparatus would be a red flag for any sort of hiring - especially in the domestic nuclear industry in France.
> These are not normal times and I believe USA also hired foreign nationals to top secret programs during and after WWII
The 1950s is ancient history now.
All countries have drastically ramped up background checks in NatSec and NatSec adjacent industries, and hiring foreigners (even from aligned states) can be a potential threat, as they will continue to retain family ties with their country of origin.
This same incident literally happened last week in the French nuclear industry, where a senior exec was canned because they weren't able to pass background checks due to their familial ties in Russia and past work in the Russian defense space before naturalizing as a French national.
If the US is to be viewed as a threat by EU member states (like a lot of Redditors-turned-HNers argue or imply), then it suffices to say that these states need to view Americans working in NatSec industries as potentially compromised.
So the US is a threat because it refuses to pay disproportionate funds for the defense of Europe? I think everyone whining about it is losing their minds over nothing. We have been massively subsidizing these "friends" so that many of them can afford free healthcare. Fuck that noise.
The ROI to USA that it has been the dominant military power for the last decades, combined with the cooling effect of NATO across the globe – i think you can imagine the instability and increase risk of nuclear war without NATO (Japan, South Korea, Europe, etc would acquire nukes without USA guarantees)
It has enabled a very stable and predictable world. This has benefited USA immensely. It it not clear to me, that rewinding and dismantling this system will have net positive effects in the next 50 years for USA.
But in the 1990s, militaries like West Germany, Canada, etc remained competitive and helped balance the load.
There's no reason Germany (and in reality it's only Germany that's the laggard) can't rebuild their conventional fighting capacities to help load balance again.
It's just German instraginence because of their fanatical opposition to deficits that is hampering their ability to do so.
> increase risk of nuclear war without NATO (Japan, South Korea, Europe, etc would acquire nukes without USA guarantees)
As I wrote below, that is highly unlikely in much of Europe, as most European states (except the UK and France) lack the capabilities to develop credible nuclear delivery systems like ballistic missiles or nuclear submarines.
>There's no reason Germany (and in reality it's only Germany that's the laggard) can't rebuild their conventional fighting capacities to help load balance again.
They could, yes. But would they need to if Russia is weak? The current strategy from USA seems to be to appease Russia, give them what they want and weaken security guarantees. The point im trying to make is that there is another way here which i beleive nets the West (both USA and Europe – USA's natural ally) more benefits than gearing up for war (aka the Peace Dividend).
>lack the capabilities to develop credible nuclear delivery systems
Is this the case? I think Sweden were months away from testing their nuclear bombs when they were conviced to dismantle the system in return for protection from USA's nuclear umbrella and were in process of producing supersonic nuclear bomber in the 1950s.
They still produce today world class submarines, develops and build their own fighter jets.
I would think that Europe has knowledge and skillset. I mean, North Korea managed.
> The point im trying to make is that there is another way here which i beleive nets the West (both USA and Europe – USA's natural ally)
And this is the crux of the issue. It's hubris to assume Europe is our natural ally and should always be our top priority.
In the US, our Pacific allies (Taiwan, Japan, South Korea, Australia, New Zealand) are a higher priority than our European ones.
As can be seen in US Military Deployments (the majority of US Armed Forces personnel is deployed in the Pacific) along with economic relations (APAC trade is larger than EU trade)
For the US, China is the primary adversary to worry about, not Russia. Why should European states assume the US has an obligation to always support Europe? At least France and UK have historically tried to maintain some strategic autonomy, and Eastern NATO states like Poland, Romania, and Turkiye have continued to build domestic defense capacity.
And unlike most European countries, our Asian allies (SK, JP, TW) have continued to build fairly competitive domestic defense industries. Japan and South Korea can manufacture their own ballistic missiles, tanks, submarines, airframes, heavy artillery, etc. Only France has a similar diversity of domestic defense R&D and manufacturing capacity in Europe.
> more benefits than gearing up for war (aka the Peace Dividend).
It's Europe that gets the peace dividend. Not the US. We still need to the capacity to fight a two continent war. That's a bum deal.
> I would think that Europe has knowledge and skillset.
Europe as a continent, sure. But in reality, it's a number of individuals states working on their own domestic production, procurement, and supporting their domestic champions.
France will continue to protect Thales Group, Arianne Group, Dassault Group, etc, just like how Germany continues to back Rheinmetall, ThysennKrupp, Eurofighter, etc.
There is no ability to unify production and procurement without also undermining domestic industries and jobs.
France's Ariane Group will never transfer their Medium Range Ballistic Missiles technology to a German company - they don't want to help a potential competitor.
This same thing happened with the Eurofighter project, with France deciding to back Dassault instead.
>Why should European states assume the US has an obligation to always support Europe?
Well, it agreed to:
>In 1994, Ukraine agreed to transfer these weapons to Russia and became a party to the Treaty on the Non-Proliferation of Nuclear Weapons, in exchange for assurances from Russia, the United States and United Kingdom to respect the Ukrainian independence and sovereignty in the existing borders.
I don't necessarily disgree with the central thesis of your comment and your perspectices, but i think there is a more fruitful balance to be had than what i see as completee capitulation to Russia and the abandonment of Europe (after almost a century of collaboration and investment).
At some point one need to ask oneself: what am i defending?
I agree that we in the US need to continue defending the Budapest Memorandum and helping Ukraine where possible.
That said, individual European nations have had over a decade to re-arm and further help Ukraine (even before the 2022 invasion), but it ended up primarily being US, UK, Canada, and Turkiye providing support and training for the Ukrainian Armed Forces.
European states are starting to step up, but using Trump as a punching bag to distract from the very real issue of Central Europe's pigheaded lack of preparedness is foolish.
We are starting to see these changes now with Starmer and Macron's announcements, but plenty of individual European states are not viewing this crisis seriously enough, as Poland's Donald Tusk pointed out today [0]
We can't afford to police the whole world. NATO expansionism is one of the causes of the latest conflict in Ukraine.
It's not clear to me that unwinding our interventionism around the world is going to make the world more stable. But it is clear to me that we can't afford to keep doing it. China is eating our lunch and they have about 3x as many people as we do. Neither Europe nor the US are producing much, and the entire West is in massive debt. Do you seriously think we can win the inevitable war with China? We can't even defeat Russia in Ukraine, and China would fight much dirtier than Russia. It's time to get real, restructure our debts, and rebuild our own country.
> NATO expansionism is one of the causes of the latest conflict in Ukraine.
There's no such thing as NATO expansionism. Eastern Europe became dead set on joining NATO after seeing the destruction of Russian democracy by the remnants of the Soviet security-military complex. They correctly predicted that Russia would degenerate into an authoritarian dictatorship that would turn outwardly expansionist after KGB hardliners consolidated power and wiped out all internal dissent.
Eastern Europe's entry into NATO is like townsfolk signing up for neighborhood watch after seeing the social order break down in the next town and fearing that criminality will spill over into their own community. The criminals, of course, are disturbed that people are setting up security cameras and looking out for each other. Breaking into houses would be much easier in a town where everyone keeps to themselves.
Eastern Europe managed to break free of Russian military occupation only in 1994. Criminal gangs had taken over our town by force at the end of WWII and prevented us from living normal lives for half a century. Now, the same criminals demand that we dismantle security cameras, fire security guards, and stop participating in neighborhood watch because it "violates their interests." Damn right it does.
>Do you seriously think we can win the inevitable war with China?
I believe showing Russia, China, and the world, that USA stands behinds it commitments and allies, defends the rule based world order and is true leader of the democratic society would deter leaders such as Putin and Xi from trying anything.
NATO is (imo) close to collapsing. All it would take is a minor just-under-war incursion, e.g. in northern Finland by Russia. Would USA (Trump) defend Finland in this scenario? If not, NATO is dead and the next 100 years belong to China and Russia.
Note that "US as a threat" was a strawman argument.
(Although USA has threatened to occupy EU territory in the last weeks.)
It does not have to be a threat, but if Europe should take more responsibility for their own defense, it would make sense build up their nuclear capacity.
(And I am well aware of the differences between EU, Europe and individual countries. But it seems to me tha France is the country to build upon.)
Ukraine had nuclear weapons that they gave away. Maybe they shouldn't have?
Also - universal healthcare is not about 1-2 percent lower military spendings (much of which goes back to the country itself). USA is a rich country, you could also afford it.
> I think everyone whining about it is losing their minds over nothing
I agree. It is doable for individual European states to rebuild their conventional fighting capabilities.
France has been a proponent of this kind of "strategic autonomy" for decades, and so has the UK to a certain extent.
Most of the angst is coming from Germany, who let the Bundeswehr degrade from being one of the most capable Armed Forces in Europe in 1990 to what it is today.
> Maybe that will change. Germany certianly has capability to manufacture nuclear armements. The UK may rejoin if the situation gets dire enough.
Given the politics here, I don't see that happening. No nuclear reactors, even.
I won't be too surprised if Finland gets some, but I don't really know the full politics of the area, only that they have reactors and are concerned about Russia.
Poland might, but no earlier than 2030.
In Europe but not in the EU, Ukraine… entirely depends on if they think they have nothing left to lose. If it looks like Trump will sell them out and they don't think anyone else will pick up enough slack to keep them independent of Moscow, they may rush a development program.
I can easily believe that Ukraine would try to hire any of these workers. No reason to think any of these workers would be interested, but I can see Ukraine trying.
A nuclear shield is useless without second or third strike capabilities - which requires a nuclear triad - as both Russia and China have developed those capabilities due to their rivalry with the US.
Poland hypothetically building 10 nuclear devices does nothing if Russia can launch multiple strike after getting striked.
Furthermore, developing nuclear weapons is causus belli enough to justify a hot war for China or Russia, nor does it actually prevent war as can be seen with the Kargil War.
More critically, nuclear programs are expensive and that money is better used in further building out conventional military capabilities, as can be seen with the Russia-Ukraine War, where a country with an ossified MIC (Ukraine) is able to cause significant pain to a nuclear power.
Tl;dr - Nuclear weapons alone are useless in a world where most Nuclear Powers (US, Russia, China, India, Israel) have a nuclear triad and multi-strike capabilities.
> Also South Korea and Japan may
The Indo-Pac theatre is different from the European theatre, as there is bipartisan support to prioritize Asian defense over European defense.
Furthermore, in a situation where American support is reduced, Asian countries can continue to retain conventional warfighting capabilities. This is because defense spending across Asia had always been high since the early-mid 2010 standoffs, so there hasn't been the same level of angst that much of Central Europe has.
Furthermore, Korean and Japanese military exports are extremely competitive, with Indonesia, Vietnam, and Philippines all closely aligning with SK and JP on defense posture.
Russia is not the USSR. In practice, Russia is Moscow and St Petersburg. Pose a credible threat to those cities and be set. Because Poland and Russia are so close geographically, plausible second strike capability can be achieved many ways without having ICBMs and nuclear strike subs.
I'd focus on your previous point about second-strike capabilities.
> Even if Moscow or St Petersburg are completely wiped off the map, Novosibirsk, Kazan, Omsk, etc will remain while much of Poland is irradiated.
Yes, but the goal isn't to win a first strike, it's MAD to prevent the other side doing that.
Russia has enough warheads to not just level Poland's cities, but every settlement and forest in the country.
Poland with 10 credible nuclear weapons is enough to break the economic back of any country who attacks, so they won't attack.
This needs what you said before, second-strike capability. Either that or a fast enough response time that they can launch while hostile missiles are still inbound. (Or does that still count as second-strike?)
> Poland with 10 credible nuclear weapons is enough to break the economic back of any country who attacks, so they won't attack.
If you are in a situation where you are even seriously considering a nuclear strike, that means you are viewing a threat as existential, which completely undermines the economic argument.
> This needs what you said before, second-strike capability. Either that or a fast enough response time that they can launch while hostile missiles are still inbound. (Or does that still count as second-strike?)
Absolutely, but the issue is that this takes A LOT of time to build and implement, and a country like Poland or Germany cannot build that kind of capability overnight. Yet a nuclear program can be viewed as an existential threat that can be used as a causus belli for war (conventional or nuclear).
This is a pretty bad RoI.
Nuclear programs are expensive, and instead of spending the amount you would need to build a nuclear program, it's much better for Poland and Germany to double down and concentrate on conventional war capabilities such as rocket systems, drones, artillery, and heavy weapons. The fact that a country with an ossified MIC like Ukraine is able to bog down a military like Russia's with conventional capabilities is proof enough that doubling down on building conventional war-fighting capabilities is enough to cause severe pain on an aggressor while not turning a conflict into an existential one which justifies nuclear warfare.
And this is why you never hear Polish or German military leadership talk about developing a nuclear program.
> If you are in a situation where you are even seriously considering a nuclear strike, that means you are viewing a threat as existential, which completely undermines the economic argument.
No, because it's not symmetric.
Party A may be an existential threat to party B, party B can prevent that existential threat just by being sufficiently painful. B doesn't even have to be close to an existential threat to A for it to be painful enough to reconsider.
That's how bees keep humans away from hives. Also how the Irish kicked my great-grandparents generation out of controlling Ireland, even at the height of the British empire.
> Nuclear programs are expensive, and instead of spending the amount you would need to build a nuclear program, it's much better for Poland and Germany to double down and concentrate on conventional war capabilities such as rocket systems, drones, artillery, and heavy weapons. The fact that a country with an ossified MIC like Ukraine is able to bog down a military like Russia's with conventional capabilities is proof enough that doubling down on building conventional war-fighting capabilities is enough to cause severe pain on an aggressor while not turning a conflict into an existential one which justifies nuclear warfare.
Yes, they are expensive. Also, I expect a multi-polar nuclear arms race to go hot much more easily, to normalise their use, to generally be bad for everyone.
So I hope you are correct (or, more importantly, that your opinion is shared by decision makers). On the other, there's clearly a constant undercurrent of "let's not give too much more aid to Ukraine just in case the Russian nukes actually work", so I don't think it's seen that way.
> German military leadership
Given the local attitudes towards even nuclear reactors, I think it's just a political non-starter around here. (I'll have to wait and see if @TeMPOraL sees this and responds regarding Poland's politics?)
> So I hope you are correct (or, more importantly, that your opinion is shared by decision makers)
Yep. I'm basing my stance on Poland's current defense strategy [0][1].
Furthermore, Poland's on track to outcompete Russia in rocket artillery and tanks, so it has day 1 capabilities that are comparable to a tactical nuclear strike minus the cost.
> No, because it's not symmetric
Yep. It isn't symmetric, but it doesn't matter, because crossing the nuclear launch threshold is enough to justify retaliatory strikes and counter-strikes - which is something a state which lacks a second strike or nuclear triad cannot deter against.
And the Kargil War in 1999 was proof enough that two states having nuclear weapons capabilities alone cannot deter a war.
It's not existenial if you don't have anyone on the other side of the "deal" to offer nuclear strikes in return. So far, it's been the US, but that doesn't seem assured for the future anymore.
It takes decades to build credible second strike or nuclear triad capabilities. A conventional war would be finished well before that.
For example, to launch into Russia, Poland and Germany would need Tactical, SR, MR, and LR Ballistic Missiles, but neither state has a ballistic missiles program so they would need to start from scratch or be entirely dependent on France (which the defense industry in both states have lobbied against).
If a Poland or Germany attempts to begin a nuclear weapons program, that is reason enough for a belligerent nation like Russia to start a war.
Nuclear weapons are not meant to be used. They are meant to prevent invasion attempts.
For instance, between China and its own nuke North Korea is guaranteed not to be invaded by the US.
Likewise, Israel is guaranteed that Arab countries or Iran won't try to invade it.
In fact, that's why France developed its own nukes, in addition to making sure they'll keep a seat at the "adults table".
Poland might be/have been another good example. They would only need to guarantee that St Petersburg, Moscow, and Minsk would be vaporised in case of invasion to be safe on their Eastern border.
> Poland might be/have been another good example. They would only need to guarantee that St Petersburg, Moscow, and Minsk would be vaporised in case of invasion to be safe on their Eastern border
As I mentioned above, the fear of vaporization goes away if you can continue to keep striking after being striked. If you've reached the point where you are launching nuclear strikes, you accept the massive toll that you will have to pay with a launch, and wish to enact as severe a toll as possible. To do that you need second strike and nuclear triad capabilities.
Every single one of the countries you mentioned has second strike capabilities and either has Nuclear Triad capabilities (China, US, Israel) or is working to implement them (France, North Korea).
A country like Poland or Germany is too late in the game to build second strike or nuclear triad capabilities, and even starting a nuclear weapons program would be causus belli enough in the short-to-medium term for a war while being unable to prevent a strike.
No,the fear does not go away. The point is that Russia isn't even going to try if they know their main cities are going to be wiped. Whether they can possibly ultimately "win" because irrelevant if it is a pyrrhic victory and/or cost is to high.
> The point is that Russia isn't even going to try if they know their main cities are going to be wiped
And my argument is if you as a state are seriously consider a nuclear strike, you have accepted that millions will die, but it's better to completely wipe out the other opponent. No amount of rationalization can resolve that level of existential threat. Poland saying "we can threaten to nuke Moscow" is enough of a justification for Russia to view Poland as an existential threat to wipe off the map.
Large countries, and especially large nuclear powers, have multiple cities where industries are distributed. This remains true in Russia, as most of their military manufacturing capabilities remain spread out across Omsk, Kazan, Novosibirsk, etc.
If you wish to scare a country from fighting a war with you, even conventional capabilities are enough to highlight the pain while also not reaching the threshold where a conflict becomes existential.
So no, it does not make sense for Poland or Germany to develop nuclear capabilities so late in modern world.
France can nuke Moscow and is still there, North Korea caan nuke Seoul and Tokyo and is still there.. once you have credible nukes you can't be wiped out, that's the point of having them.
There is no being 'late', either. It's about the threats.
I have never implied that Poland should develop nuclear weapons. I wrote that they are/were in the very situation in which nuclear weapons may be desirable (similar to France, Israel, North Korea). Relatively small countries have nukes simply to make any attempt at invasion not worth it even if everyone knows they can't "win" against the USSR/Russia/the US, anyway.
They have second strike capabilties and a nuclear triad
> North Korea caan nuke Seoul and Tokyo and is still there
They have second strike capabilities and are working on nuclear triad capabilities
> I wrote that they are/were in the very situation in which nuclear weapons may be desirable
But unlike the nations listed above, they do NOT have a domestic ballistic missiles program, and that would take decades to build.
SK and Japan both have had domestic ballistic missile development and submarine development capabilities for decades, and that's why if they wished to become a nuclear power, they could do so very quickly.
Poland and Germany does not.
> Relatively small countries have nukes simply to make any attempt at invasion not worth it even if everyone knows they can't "win" against the USSR/Russia/the US, anyway
You can't make nukes (and the associated delivery systems) overnight. It takes decades to build the entire ecosystem.
Just having a nuclear bomb isn't enough if you lack the ability to develop and manufacture ballistic missles or submarines domestically.
Furthermore, as was seen in the Kargil War in 1999, nuclear weapons can fail as a deterrence for war.
Sounds like the jobs probably already had people who can't find any other job and don't especially care, if you ask me. At least that makes as much sense as your argument about future hires.
The problem is what one means by "works". Is it just that it runs without triggering exceptions here and there?
One has to know, and understand, what the code is supposed to be doing, to evaluate it. Or use tests.
But LLMs love to lie so they can't be trusted to write the tests, or even to report how the code they wrote passed the tests.
In my experience the way to use LLMs for coding is exactly the opposite: the user should already have very good knowledge of the problem domain as well as the language used, and just needs to have a conversation with someone on how to approach a specific implementation detail (or help with an obscure syntax quirk). Then LLMs can be very useful.
But having them directly output code for things one doesn't know, in a language one doesn't know either, hoping they will magically solve the problem by iterating in "closed loops", will result in chaos.
It clearly does not result in chaos. This is an "I believe my lying eyes" situation, where I can just see that I can get an agent-y LLM codegen setup to generate a sane-looking working app in a language I'm not fluent in.
The thing everyone thinks about with LLM codegen is hallucination. The biggest problem for LLMs with hallucination is that there are no guardrails; it can just say whatever. But an execution environment provides a ground truth: code works or it doesn't, a handler path generates an exception or it doesn't, a lint rule either compiles and generates workable output or it doesn't.
About the cleaning up phase: the author cleaned up the bitmaps before the vectorizing stage, and that included separating lines.
In my experience it's a little bit easier to clean the vectorized results by removing unwanted points, because those are easier to see than line connections that may or may not pose a problem later. (But to each their own.)
Actually that's my usual approach, but you should see the vector output from these. I tried doing it and than went back to cleaning up the raster images.
I joined Pinboard at a time when there was no recurring costs, just a small fee to join, for life. I think I paid $10 total, 12 or 15 years ago, and not a penny since. For that amount it's a pretty good deal.
It was a single email, in 2022, and the wording was exceptionally friendly!
(For those who don't know, the site moved to a subscription model in 2015 and I asked old-timers to consider converting their account voluntarily. Everyone who chose not to still has a lifetime membership to the site.)
I loved that earlier pricing model, where the cost increased with each sign-up. It may not have been sustainable in the long-term, but it was an interesting attempt to do something different.
1 is plainly false. Enormous ressources have been poured into models since the end 2023 and the "intelligence" (for lack of a better term) has stayed roughly the same, around the level of GPT-4. Nothing has happened since then.
3 is a philosophical opinion, not based on any falsifiable evidence.
How is 1 false? Log improvement means for 10x the cost the model is 2x as good. For 100x the cost, the model is 3x as good.
Not a curve to be happy about TBH. You need to simultaneously find big efficiency wins and drive up costs substantially to get 4-5x improvements, and it is probably impossible to maintain good year on year improvements after the first 2-3 years when you get all the low hanging fruit.
1 is plainly false. Enormous ressources have been poured into models since the end 2023 and the "intelligence" (for lack of a better term) has stayed roughly the same, around the level of GPT-4. Nothing has happened since then.
You need to spend some quality time with o1-pro and/or Gemini Pro 2.0 Experimental. It is not the case that there has been no progress since GPT4. CoT reasoning is a BFD.
1 is plainly false. Enormous ressources have been poured into models since the end 2023 and the "intelligence" (for lack of a better term) has stayed roughly the same, around the level of GPT-4. Nothing has happened since then.
this would be true only for people
who have used the same model since 2023 :) Jesus!
Then engines start to heat up and fail. And if there's an obstacle along the way, it's game over.
This is a dangerous game played by morons full of hubris.
On June 16 YC organizes an "AI Startup School"; the first speaker is Elon Musk:
https://www.ycombinator.com/blog/ai-startupschool
What a shame.
reply