Do you disagree? Come, let's inspect your thoughts. What news have you been consuming. Who do you trust? We can precisely evaluate your degree of susceptibility to propaganda. Aka "retardation"
You can start by listening to what any Venezuelan thinks of the situation. Those on the low end of the bell curve do tend to think highly of themselves.
> I’ll make this bet: any such model you come up with could be improved by including notions of international agreements and laws.
And you'd have lost the bet with such a naive understanding of geopolitics and power dynamics played by nation states. Are you reading the thread you're on?
To the commenter above: it seems like you are responding to something other than what I wrote. Perhaps my meaning didn't come across? I'll try again:
Start with model M which does not account for international law.* For any such model, that model can be modified by including information about international law. Call that M'. I claim M' will do better than M. Do you agree? Disagree? Why?
Onto my next point. Please take the context into account. I was responding to a comment that said:
>>> Having said that, international law is a myth.
This is why I said:
>> Arguments over definitions really bore me. To any reasonable person predicting the future, international law is an important factor. It cannot be simply waved off because it is flawed and unevenly enforced.
I am having a hard time understanding how you think I'm naive for saying the above. To me, it would be naive to ignore international law altogether, simply because it is nuts to ignore relevant information. Am I just redefining my claim to be "this information is relevant to predicting an outcome". Maybe, but even this seems to be getting lost in translation.
May I ask if you've done geopolitical analysis at the international level? I have no idea -- you very well might have. By the same token, I may have as well. This isn't a who-has-the-bigger GPU question. I'm just trying to understand if you understand the game we're talking about. If you're trying to predict price stability, election outcomes, how long a dictator stays in power, etc... what models do you build?
If you want to compare some models on this, let's do it. We'll compare and see if including international law/agreements has predictive value (relative to not including them). Are you game?
* It is possible a model could build up an internal representation of international law even if not provided it directly. If such an internal representation proves useful and predictive, this serves to prove my overarching point, albeit in a different way; namely international law (conceptually) matters. It doesn't matter if we call it 'real', 'fake', 'a myth' or whatever. Arguing such terminology is a waste of time. If we can measure it (somehow, to some degree) and use it to make better predictions, that is good enough for me.**
** It is also good enough for physicists! People may argue the _metaphysics_ of quantum physics tirelessly, but if the equations work, that is pretty darn impressive. Call it "spooky action at a distance" or "entanglement". In an an important sense, these are just words, metaphors, attempts to make sense of reality. Focus on how to turn the crank on the theory and don't get hung up on what is 'real'.
Yes I was incorrect to say such a model would be strictly worse off. But my read is that you over index on the notion of laws, hence your general befuddlement on the current outcome. Sovereign nations follow international law and order to the extent their goals align and perceived costs of contravening them exceeds some threshold. Might ultimately makes right, has always been the case. That's realpolitik for you, unfortunately.
I appreciate the discussion and thoughtful response.
> But my read is that you over index on the notion of laws
To be fair, nothing I said asserted the relative importance of international law in comparison to other influences (i.e. military power, strategic goals, economic interests, a vindictive leader).
> hence your general befuddlement on the current outcome
Where do you get the impression that I'm befuddled? I was disappointed in the lack of nuance of some comments, so I pushed back, but I'm not 'befuddled' by current events.
> Sovereign nations follow international law and order to the extent their goals align and perceived costs of contravening them exceeds some threshold.
This sounds like the 'rational actor' model from international relations. [1] But that model is not the only game in town, nor is it universally the best model to use!
> Might ultimately makes right, has always been the case.
I would happily see this phrase fall out of usage. It is what authoritarians want you to believe. What is right != who has power. Normative != positive. They are not the same. We would do well not to blur 'what is' with 'what should be', not even in an aphorism.
Yes I would like that phrase to fall out of use too. My intention was less of the idiom's original normative meaning but to emphasize that it is ultimately power that enables or constrains a nation's possible actions. Apologies for my confusing use of the phrase.
> but to emphasize that it is ultimately power that enables or constrains a nation's possible actions.
If one interprets this as "perceived power" I think we'd be getting closer. But even that is not enough.
I would also need us to recognize "the power that ideas have in shaping worldviews". Consider the Magna Carta, the Geneva Conventions, the UN Declaration of Human Rights, and so on. They can't simply be accounted for using a probabilistic calculus of consequences w.r.t. military force, economic sanctions, popular uprisings, and so on.
International law and notions to some degree also become normative. They are worldviews and aspirations that spread. (Memetics is a powerful analytic frame here!) These laws and norms take hold in people's minds and they shape how leaders and their people think about what is good and what should be done. In this sense, even though they are ultimately just neural patterns (if you are a materialist like me), they can be thought of as 'real' and impactful when it comes to making predictions about how leaders act.
I wonder if we would both agree on this: as people lose faith in the normative force of law, they care relatively more about the perceived consequences. Seems pretty straightforward?
Such a degradation, seems to me, cannot be good for civilization. A world where everything is purely contractual or consequentialist does not work in a world of agents with very limited computation.* It is just too costly to formalize everything in terms of individual incentives. Building systems where all the consequences are perceived by actors at the right levels of the system is really hard. Maybe it can work with certain kinds of information systems. But with humans, with our current biology and technology, I don't think it scales well at all. (At this point you might wonder if John Von Neumann is rolling is in his grave, but I suspect if he lived today, he would agree! His work spanned computation theory, game theory, and more.)
* Here is a guess that seems plausible (hypothetically): In a perfect world of unlimited computation, agents would be smart enough to think of interactions as long-run games and might be able to have a healthy society even if they don't 'believe' in norms.
Do you know the "standard technique of privatization"? Sanctions are the "Defund" aspect of that for American Imperialism. The woes of Venezuelans are largely economic and a product of the sanctions.
Venezuelans would disagree loudly with you, and in fact would take offense with the notion external interference caused their country's downfall given the extent Maduro's gov ran the country into the ground.
> I've been following Yann for years and in my opinion he's been consistently right
Lol. This is the complete opposite of reality. You realize lecun is memed for all his failed assertions of what LLMs cannot do? Look it up. You clearly have not been following closely, at all.
Sure and that is fair. Seldom are extreme viewpoints likely scenarios anyways, but my disagreement with him stems from his unwarranted confidence in his own abilities to predict the future when he's already wrong about LLMs.
He has zero epistemic humility.
We don't know the nature of intelligence. His difficulties in scaling up his research is a testament to this fact. This means we really have no theoretical basis upon which to rest the claim that superintelligence cannot in principle emerge from LLM adjacent architectures--how can we make such a statement, when we don't even know what such thing looks like?
We could be staring at an imperative definition of superintelligence and not know it, nevermind that approximations to such a function could in principle be learned by LLMs (universal approximation theorem). It sounds exceedingly unlikely, but would you rather be comforted by false confidence or be told the honest truth of what our current understanding of the sciences can tell us?
Lecun has already been proven wrong countless times over the years regarding his predictions of what LLMs can or cannot do. While LLMs continue to improve, he has yet to produce anything of practical value from his research. The salt is palpable, and for this he's memed for a reason.
What a shocking lack of imagination. Do you seriously think in a hundred years you'll still hold this opinion?
reply