> We show that this metric has been consistently exponentially increasing over the past 6 years, with a doubling time of around 7 months.
If true, how much of this is a result of:
1. Genuine technical advancement
or:
2. Shoveling trillions of dollars into compute resources in order to service incoming LLM requests in a way that is completely unrealistic over the long term?
In other words… are we talking about genuine, sustainable innovation that we get to take with us moving forward and benefit from? Or are we talking about an “improvement” that is more akin to a mirage that will eventually disappear when the Ponzi scheme eventually collapses?
Much of this is due to vastly better posttraining RL, not models that are much bigger. The idea that most of these gains comes from training really big models, or throwing immensely larger amounts of compute at it, is not really true.
I wonder how much of this stuff is attributable to true model advancement, or if it’s an improvement in the genetic harness? It’s impossible to separate strict model improvement from improvement in the associated tools.
Gotta have the digital laugh track analogue to show people how to react.
(I'm not kidding, a ton of people will watch e.g. movie trailers or live streamed events through one of many reaction streamers so they get prompted on how to feel. I hated laugh tracks back then, I hate reaction streamers now, let me have my own feelings!)
Definitely not. The article does go on to acknowledge this:
"The result of this volume bias in the system is an onslaught of low-quality legislation. Compliance is often impossible. A BusinessEurope analysis cited by the Draghi report looked at just 13 pieces of EU legislation and found 169 cases where different laws impose requirements on the same issue. In almost a third of these overlaps, the detailed requirements were different, and in about one in ten they were outright contradictory."
Whenever I hear a politician patting himself on the back for how many pieces of legislation he got passed, I cringe at the thought of all the junk in it.
It’s amazing how similar legislation is to software engineering in that regard: It all comes down to managing complexity. A good law is achieving its effect with as little special case handling in as few lines as possible while covering most of the problem space.
That was meant ironically. The article explains in great length that this quantitative "productivity" does not result in qualitative "productivity".
The fundamental problem, in my view, is that any significant reform of EU procedures would mean strengthening the European Parliament. In other words, EU governments must be persuaded to relinquish some of their sovereignty. Since the signing of the Lisbon Treaty in 2007, there has been no significant progress in this regard. This is also related to the fact that, unlike 20 years ago, many center-right governments are now in power in many EU countries, and strengthening the EU is not on the agenda of most of them—often quite the opposite. France is an exception, but Emmanuel Macron's initiative was met with little response.
I truly hate how this buzzword is misused with regards to the EU. Voluntarily delegating authority is not the same as losing sovereignty. If you can un-delegate the authority at your own prerogative, you have not lost sovereignty. If the UK, for example, had genuinely lost its sovereignty, it would not have been able to voluntarily withdraw from its participation in the EU.
I would rather say that the term “sovereignty” is multifaceted. We have the concept of popular sovereignty, which means that political power emanates from the people and all other sovereignty is delegated.
However, there is also a use of the term “sovereignty” in the sense of self-determination over one's own state structure and the ability to ward off external interference. When a state transfers certain sovereign rights to the EU, this is more than just delegation. In German constitutional law, for example, this means that the transfer of such rights to the EU has constitutional status.
If there is a lawsuit before the German Federal Constitutional Court (Bundesverfassungsgericht) that challanges an EU law or regulation, the court first examines whether the EU law in question regulates something that actually falls within the EU's area of responsibility or whether it is something over which Germany has reserved its sovereignty.
The most prominent example of such a ruling is the PSPP (Public Sector Purchase Programme) case from 2020, where the German Federal Constitutional Court ruled that another ruling from the Court of Justice of the European Union (CJEU) regarding the European Central Bank (ECB) program of purchasing government bonds is not binding in Germany because the CJEU exceed its judicial mandate and violated the sovereignty of the German Bundestag. The case was "solved" when the European Central Bank provided the Bundestag with additional documentation regarding the program and the Bundestag concluded that everything is in order.
In this decision the term "sovereignty" is explicitly used to outline the case: "In particular, these [complaints] concerned the prohibition of monetary financing of Member State budgets, the monetary policy mandate of the ECB, and a potential encroachment upon the Members States’ competences and sovereignty in budget matters."
The decision later concludes:
"This standard of review [in the ruling of the CJEU] is by no means conducive to restricting the scope of the competences conferred upon the ECB, which are limited to monetary policy. Rather, it allows the ECB to gradually expand its competences on its own authority; at the very least, it largely or completely exempts such action on the part of the ECB from judicial review. Yet for safeguarding the principle of democracy und upholding the legal bases of the European Union, it is imperative that the division of competences be respected."
Cutting down scope doesn't necessarily reduce cooperation.
Until 2004-07, the EU was an economic and political union for Western, Northern, and parts of Southern Europe - all of whom are largely aligned from a developmental, economic, and social perspective. It was after the rapid Eastward expansion of the EU without updated checks and balances that dysfunction arose.
The EU will remain dysfunctional as long as CEE countries that are not aligned with the core mission of the original EU remain politically relevant. The only way to reduce this dysfunction is to either decouple the policy component from the economic component, or reduce the amount of nations that should have a say in policy to those that are aligned with the EU.
The fact that the government of an EU member state like Hungary still has political privileges yet is clearly preparing for some form of economic [0][1] warfare and potentially actual [2][3] warfare highlights how tenuous the project is as it stands today.
First it's Hungary, then it's Slovakia, then ...
Clearly the EU status quo is unsustainable and needs to be reformed ASAP.
Because a disunited Europe was the cause of both World Wars and multiple genocides.
The continent should be tightly linked together so that war is unthinkable, and so that the pluralistic and very factional nations of Europe can negotiate as equals with other great powers.
Compare the UK pre-Brexit and post-Brexit for some evidence.
I agree. But also, are the words and actions of Russia, China, North Korea, Israel and the US (and the rest of the nuclear club) the sort you think ideal?
That the US decides who can have them is darkly hilariously.
> It is the mission of the Islamic Republic of Iran to erase Israel from the map of the region.
> Israel is a hideous entity in the middle east which will undoubtedly be annihilated.
> Iran's stance has always been clear on this ugly phenomenon (Israel). We have repeatedly said that this cancerous tumor of a state should be removed from the region.
> Western countries allow no freedom of expression, which they claim to advocate, with regard to the myth of the massacre of Jews known as the holocaust, and nobody in the West enjoys the freedom of expression to deny it or raise doubts about it.
- Ali Khamenei, Supreme Leader of Iran
Those who defend the Iranian regime or suggest that the Israeli government is the greater threat do so to their disgrace. SMH.
reply