@Quantumfreakonomics's banner p

Quantumfreakonomics


				

				

				
1 follower   follows 0 users  
joined 2022 September 05 00:54:12 UTC

				

User ID: 324

Quantumfreakonomics


				
				
				

				
1 follower   follows 0 users   joined 2022 September 05 00:54:12 UTC

					

No bio...


					

User ID: 324

Lessons in attempting to value things in the absence of an active market. What is the exchange rate between units of emotional distress and dollars? What even are the units of emotional distress?

Here’s the pickle: it seems obvious that the best way for Alex Jones to make money is to continue producing infowars content. I’m not sure that this is the closure that the plaintiffs had in mind.

The fatal premise of falsificationism is the denial that inductive reasoning can be a basis of true knowledge. A theory can never be "confirmed", it can only fail to be falsified. Falsificationism preserves the veneer that science can be based on deductive reasoning alone. A theory that has been falsified is logically impossible to be true*. So one can attempt to define science in a crude way as the set of tested falsifiable theories that have not been falsified.

The cost of this is the denial of objective scientific truth. If general relativity were falsified tomorrow, would you feel comfortable walking out of a fifth-story window? Everybody knows gravity is real. It's obvious. Inductive reasoning works. We now have stronger theoretical justifications for induction than Popper did, but the damage is done.

*Kuhn does a good job of poking holes in this assumption. It's a shame he goes even further off the deep end of denying objective truth.

Popper and After: Four Modern Irrationalists by David Stove. It's pretty crazy how everyone just sort of nodded along as Popper's falsificationism became the dominant idea in philosophy of science, but hardly any scientists act as though they actually believe in it, because it's absurd.

Women are the dark matter of history. We know that they were always there. We can assume that they must have had opinions and motivations, and that they took actions in accordance with these motivations. But they didn't write much, so it is easy to simply whitewash them out of history, or alternatively to adopt a feminist reading of history that projects modern sensibilities centuries or millennia into the past in order to fill the gaps.

Still, I can't discount the possibility that revolutionary France was unique here. Charlotte Corday was 24 when she stabbed Jean-Paul Marat to death in the bathtub. You don't see pretty young women doing that kind of thing in other times and places.

There’s no crime here. There are no victims. There’s no CSAM, because the images are not of children (notably the AI models are trained on nude adults), nor did any sexual abuse occur in its production.

Let’s check 18 U.S. Code § 2256(8)

(8) “child pornography” means any visual depiction, including any photograph, film, video, picture, or computer or computer-generated image or picture, whether made or produced by electronic, mechanical, or other means, of sexually explicit conduct, where— (A) the production of such visual depiction involves the use of a minor engaging in sexually explicit conduct; (B) such visual depiction is a digital image, computer image, or computer-generated image that is, or is indistinguishable from, that of a minor engaging in sexually explicit conduct; or (C) such visual depiction has been created, adapted, or modified to appear that an identifiable minor is engaging in sexually explicit conduct.

I think that definition is way too expansive, but that’s the definition that our elected representatives came up with. As written, it is definitely a federal crime to create deepfakes of Stacy from English class getting railed and texting them to your bros. Some of these provisions are oddly specific. I don’t have time do dig into the legislative history right now, but I suspect they were added recently in order to cover this exact thing.

We know that it is physically possible to have mosquito-sized flying machines capable of short-range independent operations. We have an existence proof. They’re called mosquitoes.

It's really tiring seeing people ask AI questions that are worded in a tricky way that most humans would screw up, then using this as proof that AI is way dumber than humans. I'll be honest, I had to read that question four times to understand it. It's the logic puzzle version of the Stroop Effect.

Fired Superalignment Researcher Drops Four-and-a-Half Hour Podcast With Dwarkesh Patel, 165-Page Manifesto.

[Podcast] [Manifesto]

Leopold Aschenbrenner graduated valedictorian at Columbia in 2021 at the age of 19. He then worked for the FTX Future Fund before the fiasco, then wound up at OpenAI on the Superalignment team. In April of this year, he was fired, ostensibly for "leaking". In Leopold's telling, he was fired for voicing security concerns (not to be confused with safety concerns) directly to the board. At post-coup OpenAI, being the kind of guy who would write a manifesto is a massive liability. Private interpretation of the Charter is forbidden.

Leopold's thesis is that AGI is coming soon, but that national security concerns, not alignment, are the main threat. A major theme is how easy it would be for the CCP to gain access to critical AI-related capabilities secrets via espionage given the current state of security at frontier AI labs. I was a bit confused at the time of the firing as to what Eliezer meant by calling Leopold a "political opponent", but it is very clear in retrospect. Leopold wants to accelerate AI progress in the name of Western dominance, making America the "compute cluster of democracy". He is very concerned that lax security or a failure to keep our eyes on the prize could cost us our lead in the AI arms race.

What comes through in the podcast in a way that doesn't from the manifesto is how intellectually formidable Leopold seems. He is thoughtful and sharp at all times and for all questions. Admittedly I may be biased. Leopold is thoroughly Gray Tribe ingroup. He has been on Richard Hanania's podcast, and mentions Tyler Cowen as one of his influences. It is tempting to simply nod along as the broad outline of the next 5 years is sketched out, as if the implications of approaching AGI are straightforward and incontrovertible.

The one thing that is notably missing is are-we-the-baddies? style self-reflection. The phrase, "millions or billions of mosquito-sized drones", is uttered at one point. It makes sense in the military context of the conversation, but I really think more time should have been spent on the political, social, and ethical implications. He seems to think that we will still be using something like the US Constitution as the operating system of the post-AGI global order, which seems... unlikely. Maybe disillusionment with the political system is one of those things that can't be learned from a book, and can only come with age and experience.

If we're going to just be giving money away, give it to the workers, not to excess elites.

It's not really going to elites. It's going to middle-class young women. They are both the beneficiaries of the loans (being the ones getting worthless humanities degrees) and the ones who's salaries are being paid with the loans (being the ones working in education and education administration).

The industries of choice for middle-class young women are 1. Education, and 2. Health Care. A few rhetorical discussion questions:

  • Where does government money seem to be flowing these days?

  • Which industries are being ravaged by cost disease?

  • What demographic forms the base of the Democratic Party?

I am not sure at the moment which way causality flows, but I do not believe these things are unrelated.

Not impossible. The price of shares is the expected value to the shareholder of holding the shares. The assumption is that the CCP won't let that cash be paid out to Western shareholders, not necessarily that they don't have the cash.

It is downright inhuman to listen at anything other than 1x. Audio is an inherently time-sensitive medium. We evolved to interpret and understand the subtle audio cues in human speech. This includes timing.

How far down the rabbit hole should they have to go for it to be the "same" crime? What if there are two separate alleged incidents that both fall under the same legal citation? Is that the same crime? Should that count? If half of the jury thinks that crime X happened on Monday, and the other half thinks that crime X happened on Thursday, should that count?

You could convict any important person of a felony if you pierce attorney-client privilege and prosecute hard enough. "Three felonies a day" isn't literally true, but my guess is most billionaires commit at least three felonies a year.

…what? I admit I don’t have a great model of the federal government in 1860, but this seems like an insane claim to me.

To preserve the illusion of federalism? Let the state legislatures feel in charge. Everybody knows the one thing Grill-Americans won’t tolerate is not having the D or R on the presidential ballot.

I went on a big Titanic binge after that happened. Part of the intrigue is that there is just so much content out there. You can read the transcript of the official investigation. Apparently they had stenographers back then?

I would have a much easier time believing his, “Aw shucks, I had no idea we were signing that. Must have been those silly lawyers,” routine if there wasn’t a long history detailing Altman’s penchant for plausibly-deniable power grabs.

No one thing Sam Altman has done sticks out as evil. You have to have followed events for some time in order to get a feel for the pattern. I get SBF vibes from him. He does seem cleverer and more well-adjusted than SBF, but fundamentally he is making the same kind of gamble. Sam Altman thinks that there is a non-negligible probability that AGI will destroy the world, but he is building it anyway.

Are Teslas any good? I need a new car soon and I’m thinking about getting a Tesla. What is autopilot like? Is listed range legit? Is maintenance a nightmare? How do they hold up to wear and tear? Will Elon throttle me if he finds out I’m short Tesla stock?

In the two minutes of Paw Patrol footage I watched in response to this comment, it feels more like propaganda for Apple or Tesla than anything. Every single problem no matter how minor was solved by deploying a neat technological gadget.

Is it instability if Israel just wins? Do any Iranian officials really want to spearhead a new anti-Israel campaign? Sounds like a death wish.

It's eerie to see a twitter thread like that one these days that isn't overflowing with noticeposting. It's easy to forget that the Austrian School focus on individual human motivation and action is by no means universally held. To see concepts like "white flight" or "the car lobby" tossed out as explanations as if they are fundamental forces of society, and not simply the aggregate preferences of individual actors, is jarring. The real questions are, why are the white people fleeing? and why was the car lobby so popular?

I was trying to come up with some candidates, and for a split-second thought, "Scott Aaronson seems pretty emotional and prone to poor decision-making on political topics. Maybe him?"

Lol. Well, I was half right.

Still on Human Action.

Given how much I like Austrian Economics, I figured I should actually read(listen to the audiobook of) this.

The entirety of the first 90 minutes was spent belaboring the same two or three points about human nature. Yeah, these points are foundational to the entire discipline, and some core insights are genius, but I stand my my take that SBF was right about books.