The comrades in the US want the existing system removed. So, for now, while it remains, they want no policing, no oversight, no enforcement because weak governance advances their goals. Later, when the system is replaced with their preferred system, they will want maximum policing, maximum enforcement, maximum purges of undesirable elements because strong governance advances their goals.
It sounds like that until you realize that the baseline behavior is for countries that are not in NATO to be subservient to NATO. So, as you can see, it is the governments and populations that are coercing and intimidating themselves and NATO is helpfully returning them to normalcy.
I am biased here, as somebody who grew up next to an American military base in Iceland and had a pro-communist upbringing and learned to hate NATO (as well as the American imperialist agenda as a whole) from an early age.
I don‘t see a difference. I was told this when I was a child, and as I’ve grown up and seen more of the world (including by immigrating to the USA) I am more convinced then ever. In my eyes, there is no meaningful difference between USA and NATO.
There clearly is. Because NATO isn’t getting involved in America’s idiotic war of aggression and Trump is very upset about this. They’ve spent years being very upset about every other NATO nation’s spending, too.
So even if you don’t believe it, the U.S. clearly cares about NATO member behaviour and therefore recognizes a distinction.
NATO Chief, Rutte: "Europe is a platform of power projection for the United States."
Of course, the local political cover story in Italy is that these bases are 'only used for logistics, never direct attacks', but that's just a cover story to deflect any retribution Italy may receive from Iran and its allies, for Italy's participation in the US' heinous acts, and to keep the internal messaging politically correct within Italys' current circus ...
This is just posturing to avoid retribution. Everyone in Europe knows that NATO bases have been used almost every day since April 2003 by the US to commit its crimes.
Lets also admit that NATO-aligned bases in Poland and Germany are also essential to the US' crimes.
See also, Camp Darby near Pisa, actively supporting NATO and U.S. European and African Command operations.
They have been involved in pretty much all previous USA "idiotic wars of aggression", so while it may become useful in the future to make a distinction, it's also fair to appreciate that this distinction doesn't make a lot sense to a whole bunch of people.
The one time NATO was called on to defend a member was the U.S. asking for help after 9/11. And nobody but the UK followed them into Iraq, shortly after the Afghanistan war began. So there’s certainly strong history of NATO not just doing whatever the U.S. is up to.
I recall this time as a very strong refusal of the above miscategorization of NATO. Bush and his cronies had a whole “you’re with us or against us” tantrum when NATO didn’t follow. I think Americans even did some proto-Trump trolling with “freedom fries” and other nonsense.
This is simply false history. Every NATO member went to Afghanistan with the USA. Even Iceland despite not having a traditional army participated in the invasion.
As for Iraq: UK, Poland, Spain, Portugal, and Denmark provided soldiers to partake in the illegal invasion. Even Iceland participated in the propaganda war by putting them self on the farcical “Coalition the willing” along plenty of other NATO countries.
It is very hard to see NATO as anything other then an extension of USA‘s imperial ambitions. It is as if in the year 1986 you would look at the Warsaw pact and claim this is a pact of independent nations and has nothing to do with the interest of the Soviet Union.
Afghanistan: Every NATO member went because of article 5. The Taliban explicitly refused to comply when requested to hand over Bin Laden. This is as clear cut casus belli as it gets. And then Australia, South Korea, Malaysia, Azerbaijan and even Mongolia participated. How are those NATO? How is Mongolia even aligned with the US or under their influence in any way?
Iraq: Japan, Australia, Nicaragua, Philippines, Ukraine, Georgia... And I'm leaving like a dozen non-NATO countries, aligned AND unaligned with the US, that took part in the invasion or post-invasion missions. I don't agree with it, and while there were numerous lies to sell the invasion, equally numerous countries were unhappy about UNSCR 1441 being treated like toilet paper by the UN itself. Moreover, actual major NATO countries like Germany or France were diametrically opposed to the invasion and told the US to take a hike.
When you hold a hammer everything looks like a nail.
You are basically saying that supporting American imperialism is not exclusive to NATO. And I don‘t dispute that.
I also think that history has been far too kind to France’s supposed opposition to the illegal invasion. France stood firmly inside a military alliance who’s members broke international law and disregarded human rights. If France and Germany were serious about their supposed opposition to that war they would have at least threatened to withdraw. But such actions were never even considered. France and Germany were perfectly happy to allow American Imperialist agenda to materialize, even though most other NATO countries took a more active role in the crimes.
> You are basically saying that supporting American imperialism is not exclusive to NATO. And I don‘t dispute that.
No, I'm saying that states are relatively free to agree or disagree with the US, at least until president 47. France faced no consequences, and countries like Nicaragua are on the opposite end of the geopolitcal spectrum with no particular interest in a war on the other side of the world.
Because whether Iraq or Afghanistan bear the historical hallmarks of imperialism is up to debate, especially with Afghanistan with clear casus belli, stated war objectives, and no intention to annex or directly control/govern the territory. The actual number of countries supporting either is a good metric on that; I think Afghanistan war had a much wider support for a reason. Putting both conflicts together is simplistic, if not disingenuous.
>France stood firmly inside a military alliance who’s members broke international law and disregarded human rights.
The USSR stood inside an economic and arguably military alliance with literal Hitler until 41. The USSR fed the Wehrmacht, fueled their tanks, and provided metals for their war machine. Mentioning this since you already mentioned you're on the left side of the spectrum. Can we call the USSR as subservient to Third Reich imperialism?
Breaking an alliance simply because an ally does something you don't agree with, while morally commendable, it is also simplistic. France didn't pull out, like nobody pulled out when France was committing atrocities in Algeria, or when Turkey invaded Cyprus; no Warsaw Pact countries left during Hungary 56, Albania did leave because of Prague 68 but only because of it being the last straw on their already strained relations, like CSTO countries aren't pulling out because of Russian imperialism in Ukraine, Georgia or Moldova. Armenia is trying to pull out only because the CSTO wasn't there when it was needed, that is, when the treaty doesn't address its very own points.
Alliances are born out of self-interest, not out of moral purity or consistency.
> Can we call the USSR as subservient to Third Reich imperialism?
Yes we can, we should, and we do. People on the left (aside from actual Stalinist) do admit that this alliance was bad, kind of like how liberals admit how the British government was wrong to pursue the strategy of appeasement.
What makes this different though is that both Stalin and the British government (under new leadership albeit) reversed course and fought and defeated fascism. We have not seen France nor Germany do this. To the contrary.
> Alliances are born out of self-interest.
This is the world we live in, but we don‘t have to accept it, and we can and should pressure our leaders into making the moral choice. Perhaps a better world is possible, and perhaps blaming our leaders for moral failures is our way to a better world.
I think your personal beliefs fit reality if you ignore the parts of reality that strongly refute it. That kind of mentality is getting popular these days. I'll leave you to it.
Which parts of my “personal believes” are wrong? And maybe this mentality is getting more popular because it is true. People are tired of being fed propaganda for Western imperialism and capitalism more broadly.
My last attempt to convince you: Ask your self, why did Iceland not vote in favor of a ban on nuclear weapons at the UN general assembly 2016? Why has Iceland still not signed and ratified the treaty? Why has no NATO member done so? Why are Austria and Ireland one of very few European nations to do so?
Aside from France and the UK, it is hard to find an answer here which does not include American imperialist agenda.
I think it's true that NATO is for most practical purposes an instrument of the US. But it's also true that there's a distinction between NATO and the US, and some other NATO members do have influence on NATO policy - most notably Turkey, which really complicated things for the US in relation to the Syrian civil war and the Azerbaijan - Armenia conflict.
> Calling the second system a “next-move predictor” would be strange. It is not trying to predict what move appeared next in a dataset. It is trying to choose a move that wins.
i dont understand the distinction here. does working backwards from a set of win states instead of working forwards from the current state somehow change whether it’s a prediction or not?
I think the most useful word in both cases is "extrapolating".
An LLM extrapolates from its context window to the immediate next token. This word applies whether you view what's happening as "reasoning", "prediction", or as a math function.
If you're assigning steering 70 of your 100 output points because it's what you think we should go with most of the time in this situation, I'm going to call that a prediction of how to steer.
The point was, if your internal model of the world makes a prediction of a negative outcome at some point in the future, and you optimise your individual actions to avoid that negative outcome, then wouldn’t it make sense to focus on the fact you’re building and optimizing towards an internal world model rather than the fact you’re executing your actions one at a time in series?
Yes, but I think the same construction could also be used to characterize the first system; it determines the next move based on a prediction of its reward signal, where its reward signal is a measure of how likely it is that a grand master would make that move.
Like stanleykm, I found this analogy somewhat puzzling. On reflection, I think the author's point is this: the statistics of actual usage do not seem sufficient to produce a fluent LLM; it also takes reinforcement learning.
A classically pretrained LLM does not have a concept of having determined its previous tokens — it has only ever observed inputs that it had no causal influence over. This is why it's valid to say its actions are predictive and not determinative.
In the article I made 3 claims, and I agree it was a bit clumsy.
1st I say that "working forwards" in the sense of outputting one token at a time could be some form of prediction, I don't argue against that. This is what LLMs do at inference time.
2nd I say that to me what really constitutes a prediction is the pre-training. Here it's the classic setting for the word prediction in ML. The model outputs a prediction of the ground truth label: the next token.
3rd I argue that in RL there is no ground truth next token, so prediction doesn't apply here anymore.
Back to your question then: you're asking points 3 and 1 are different. Working backwards from a set of win states is basically what RL does in training. Working forward from the current state is what inference does. To me there is a distinction worth thinking about. First between the mechanism at inference time and at train time. Then between what happens in pre-training vs. RL post training.
Neither is there a ground truth in rock paper scissors (the pattern varies by the player). But one would still be trying to detect patterns and predict the next "token" the opponent is going to output.
In any case this is all very pedantic. In the process of selecting a move to make there is a prediction. Whether that prediction is the opponent’s next move or what your next move should be based on the game’s existing state, there is a prediction that the next move you make will improve your chance to win. Maybe the probability in that selection is 100%. You have no other possible move. It doesn’t matter. All we are doing here as far as I can tell is arguing over where the prediction happens and whether that counts as predicting something.
There is no truth for RLHF or RLVR. You can't predict against something if you can't check against the truth.
It's not pedantry. The objective function changes. The optimization changes. THese are real things when training a model, not hand wavy philosophical ideas.
The LLM does not determine the next token. It generate odds for all of the tokens it knows as to their likelihood of being 'next'. It's up to the harness running the LLM (and in most cases the a temperature setting) to actually decide on a particular next token. I think it's more accurate to call the thing the LLM actually generates (an ensemble of probabilities) a 'prediction'. It might be accurate to say the harness decides on the next token based on the prediction from the LLM. The role of the LLM is much more akin to predicting your opponents move than deciding your own.
Respectfully, go build one, including doing RLHF and RLVR. Those phases generate lots of tokens, then get scored on the entirety of the output, then optimize based on a scoring of that output. It doesn't check a "prediction" against what was actually "next" in data, because there isn't any "next token" data it's training on.
It does in pre training, but not in RL post training. And not at inference time. Reading over all these comments I get the feeling my mistake was not clearly delineating inference time and train time.
Your mistake was assuming people would be bothered to understand the details of how things work. Most people are lazy and don't know the details of how anything works.
My point is that “it doesn’t check the accuracy of the prediction against the data” is a non-response, because no one calling it a “next-token predictor” is making the claim that it does do that or that they’re calling it a next-token predictor because it does that.
Many people are in fact claiming the thing you are saying they are not - even if you are not. The reason they are claiming it is that it was true at one point, and most intro courses/blog posts/videos still describe them that way and then hand wave some "other stuff at the end". You can even see a comment here that refers to the gpt-2 paper. LLMs were trained to predict the next token, produced a distribution to do so, were scored against their prediction v the truth, and the weights updated so that the probability distribution made it more likely to predict the truth from that sample next time. They were, in every sense of the word, a next token predictor.
They are no longer that thing due to post training. They simply aren't making a prediction, and they aren't even optimized for the next token. If I give a distribution of the heights of the population, I'm not giving a prediction either. Distributions don't imply predictions.
Why the desperation to hang onto the word "prediction"?
> Many people are in fact claiming the thing you are saying they are not - even if you are not.
My original comment said “no one here.” Please show me where someone in the comments here is claiming that.
> Why the desperation to hang onto the word "prediction"?
No desperation here. It’s just a word that conveniently describes (especially to laypeople) what’s going on, even if it may not be the most mathematically correct or rigorous word to describe what’s going on. I think you’re being needlessly pedantic.
Our problems are caused by like 5 guys owning everything, not because stupid people have more kids than smart people. So no, Idiocracy did not predict all of this.
It's possible that stupid people in government have helped lead to the "5 guys owning everything" though. The world we live in is a result of policy choices, not a natural inevitability. And the people we've put in office over the past 50 years have played a large role. If they were being put in office by a population that was getting dumber and dumber every election cycle, that could certainly be a big contributor to the problems in wealth disparity we see today. I'm not saying that's the case, only that the two theories are not mutually exclusive.
I was just following the parent's wording. I don't really think they were stupid, or even necessarily greedy and shameless though some no doubt have been. I think mostly it has been ignorance. Likewise the general voting population: I don't think they've gotten dumber, but I do think they've gotten more misinformed. As the saying goes, "It ain't what you don't know that gets you into trouble; it's what you know for sure that just ain't so." Never has that been more true in American politics.
Our problem is society never was stable without a permanent surplus. Our problem is adaptions to this endless loop. Look at the world where things have run out, those without shoes murdering those with, the decomlexification alure of stone-age ideologies. We are a damaged, danger to itself species, whose highest hope is to aquire one more stick of high tech dynamite for a endless banton run against the odds.
Now is where we are going atm perfect? No. But at least its a plan. A self repairing asbestos orphanage for retarded firestarters which allows for progress at a realistic pacr is the way forward. Everything else is retardation whorship, in booms and busts, where every "unforseeable" regression eats a little something from civilization. If you couldnt predict trump in the 2000s, your whole model of humanity is wrong and flawed, and shouldn't be anywhere near where futures are planned.
Personally I think the USA's problems go much deeper than some vague sense that rich people are the source of all problems. Compared the whole US economy, they're a drop in the bucket, and they're not the reason that American institutions are dysfunctional; those dysfunctions were present long before Trump took power and Google got rich.
My opinion is that progressive populists don't know how to fix our problems, let alone have they even come close to identifying them, so they just focus on everything as "us vs the elites." You can see this simply by observing that the actual places that are progressive controlled - major blue cities - are the most dysfunctional places in the country. The solution? Wealth taxes that will cover a tiny percent of expenditures for a year or two tops, and no mention of why things are so expensive and slow in the first place, which has almost nothing to do with those rich guys, and have a lot more to do with dumb bureaucratic requirements; irrational procurement processes; capture from groups that sometimes are tied to wealthy interests but also come from the people like unions, neighborhood and environmental organizations; watering down every project by lacking focus on what they're actually meant to solve; procedural issues that allow cynical political actors to selectively interpret and enforce laws; and so on.
I'm sympathetic to the feeling of unfairness, but I want actual solutions, most of which are too boring for the populists on either side to stay focused on.
"You can see this simply by observing that the actual places that are progressive controlled - major blue cities - are the most dysfunctional places in the country."
Serious question - what is 'dysfunctional' about 'blue cities' that works better in 'red cities'?
The classic obvious example is cost of living. All of the blue cities have awful zoning and development restrictions that cause them to fail consistently to actually produce new housing that comes even close to demand. The oft repeated success stories are how cities like Houston, despite being not the kind of city I want to live in, are much much more affordable since they actually allow the market to function. Progressives go on and on about how supply and demand doesn't actually apply to them, and how they need to demand 100% affordable housing with absolutely no explanation of how that will pencil out for developers, digging themselves even deeper into the hole.
More examples include how crime has gained San Francisco an image something like the land of the walking dead, where progressive governance treated homelessness and mental addiction as a problem of equity of opportunity - which in the super long run is perhaps true, but doesn't lead to any good short term outcomes. It's only now that some blue city mayors realize that, actually, order might be kinda important.
As well as the inability for these cities to maintain and improve their transit systems, usually because they can’t focus on cost and service quality and instead try to make their transit projects be about irrelevant things like income inequality, protecting obsolete jobs, saving random endangered species or whatever, so you end up with the NYC MTA being unreliable, filthy, and hopeless to expect anything positive to come out of it. Red cities just sidestepped it by organizing around the car, I don't love it but at least you can get around places instead of relying on the worst transit systems on earth.
American progressives seem to have a hopeless zero-sum mentality, where the only way the common man can experience a better life is at the expense of some rich person, and perhaps that is related to the fact that many live in American cities, for which daily life continually has gotten worse because of how incompetent their governance has been.
I can't really talk though. I just gave up on the USA and moved abroad since it was too hopeless for me.
Another thing that comes to mind – education, this has been covered a lot in the media recently. While lefty progressive cities got rid of standardized testing and basic stuff like phonics education in the name of equity (you get more equality if you just don’t teach your kids anything!), other states like Mississippi are seeing great strides in educational attainment by doing the opposite. I just see a consistent pattern of progressive leaders doing incompetent things.
Healthcare - rather than just assuming that the most maximalist possible universal healthcare strategy is the best solution, actually consider how different healthcare strategies balance the cost of long-term solvency, cost control, speed and quality of service, and so on. For example, I am partial to the Japanese system, which unlike the Canadian system (which has crazy wait times), people pay 30% of costs so they still have skin in the game while the national healthcare system mandates fee schedules for common procedures to control costs. No blocking access to specialists behind primary care doctors, no banning private insurers. You get fast healthcare with a more sustainable fiscal basis, and less frivolous use of limited healthcare capacity than e.g. a system where there are no copays. The problem there though is that they have not enough supply of doctors (read: US should think about significantly increasing the supply of medical education and reducing the cost of medical training) and too much incentive for churning people through the system extremely quickly (read: US should have pricing not just be flat fee so that you aren't incentivized purely for speed).
The progressive wing seem irrationally tied to whatever Bernie Sanders claims is the best healthcare arrangement, and think that insurers are the source of all problems, when they actually are terrible from a business standpoint, they barely make money, and the cost of healthcare isn't from them but rather from other fundamental issues that have little to do with them. The result of this is that they land on a policy that half of America doesn’t want, and are too stubborn to budge on it, which means the United States will never get a rational healthcare system.
Housing & transportation - get rid of restrictions that reduce the ability of developers to actually build housing, from zoning restrictions to environmental legislation that is weaponized by NIMBYs to block development, to union labor requirements that send costs of building transit and housing through the roof, to affordable housing requirements that make projects not pencil out. Commercialize and densify areas in and around train stations as much as possible, specifically with for-profit developments, not trying to turn prime real estate into low-income enclaves because they hate gentrification more than they like progress (failing to think critically about what would likely happen if a poor and dangerous place became less poor and dangerous, or the people within such a place made some money to have the option of living elsewhere...).
I am partial again to the Japanese model where the available zones are set nationally, the rules for whether or not you get a permit is basically a checklist, if you check all the boxes you can build what you want, it doesn't discriminate on who's doing it, and for transit in particular, the train companies are literally private real estate companies that develop stations to the maximal extent possible because they make money from it. If you've visited Japan, did you think Shibuya felt awful because it was largely developed by Tokyu Corporation? Or that neighborhoods lack character and energy because the architecture is designed by a small handful of mass-manufactured modular home developer companies?
Progressives seem to think that supply and demand doesn't apply to their beloved cities, corporate landlords are the source of all evil when it comes to housing problems, and that private transit companies are not even a possible option because capitalists are greedy (when in reality they're more motivated than any of these activists are to actually do things that affect change on communities). My lived experience tells me that what I consider to be the greatest housing and transit successes in the world often come from large private developers who are incentivized well, because I've been places and observed how they function.
Also, we should be switching to land value taxes to incentivize efficient use of land, by making vacant lots expensive to hold, while also raising huge amounts of money, because the logic is simple enough for a 3 year old to understand - we have a lot of land and it is expensive. I haven't heard a single progressive even mention the idea despite being a great way to tax the wealthy while providing a strong incentive for better outcomes in cities, because they haven't thought about the root causes of why their cities feel like dumpsters long enough to care, and instead want to protect the historic laundromat and parking lot. Instead they harp on about wealth taxes that will mostly be evaded, and in the end net out tiny amounts of money that don't affect the economy as a whole and have no effect on the quality of their communities besides driving out economically ambitious and successful people.
I can keep going. I'm not a Republican, I'd much rather have the moderate Dems who actually think about the causes of issues to win primaries rather than populists who just try to get everyone pissed off (and probably are lying to their constituents anyway once they moderate as soon as the primaries end because they realize how unpopular and irrational their ideas are), but that isn't where the country is going these days. People are more interested in ideological purity tests than they are about substantive argument, in this sense the leftists aren’t any different than the MAGA people. Instead people like you already decided that people like me have no substantive opinions - very in line with the political zeitgeist these days where everyone's angry all the time and can't have rational conversations or learn from anyone else's experiences in the world. So, idk, enjoy your life.
Whether that is true or not, they have shown they do not care what happens to your child AFTER it is born. Without better morals and education we are doomed to repeat history ad nauseum/ad infinitum.
The article seems to regard some of these places as failures without examining _why_ things might be the way they are. Would life for Cubans have been better under American rule? Perhaps, but its not like life under Cuban rule was given much of a chance…
Like how is there a section on Cuba that does not mention the American embargo on Cuba for the past 50 years? How can we take this analysis seriously?
Accessibility aside, I’ve never understood the keyboard maximalism that some people insist on. Are you really so locked in that you can’t move your hand off the keyboard? Really? Even in this, the era of turn based programming?
