LESSWRONG
LW

ChristianKl's Shortform — LessWrong

ChristianKl's Shortform

18th Aug 2019

1 min read

9

This is a special post for quick takes by ChristianKl. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.

ChristianKl's Shortform

8Alexander Gietelink Oldenziel

2David Scott Krueger (formerly: capybaralet)

1ghost-in-the-weights

207 comments, sorted by

top scoring

Click to highlight new comments since: Today at 6:05 AM

Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

[-]ChristianKl2mo25-1

Currently, it seems like NVIDIA is often able to manage to sell GPUs via Singaporean third-parties to China circumventing US trade restrictions.

It seems to me, that CIA and NSA should take up the task of analyzing when that happens and then relaying the information, so that the shipments can be stopped. While it's always hard to know that CIA and NSA are up-to from the outside it seems like they currently fail at the job.

It might be valuable to push for the CIA and NSA investing more resources for that task.
1) It's a way to push US AI policy in the right direction that's not in conflict with Trump administration policy and thus has more change of being implemented.
2) It could create a power-center inside the CIA/NSA who's self-interest it is to expand control of GPUs because they might get more resources if there's more that needs controlling.

8Eric Neyman2mo

I have about 50% credence on: Nvidia has successfully lobbied the Trump administration to turn a blind eye toward illegal chip sales, such that the Trump administration wouldn't stop the exports even at zero cost.

2ChristianKl2mo

I think that position models the Trump administration as a single entity in a way that's mistakes how complex political institutions work. Different people inside of the Trump administration have different agendas. It also suggests a level of competence at secret keeping that I don't think the Trump administration has shown. I think there are some people who are serious China hawks inside the Trump administration.

3jsd2mo

The AI Action Plan has a section on this, it recommends that Commerce and IC coordinate on enforcement, including monitoring.

3Erich_Grunewald2mo

Yes, this would be useful. This has previously been discussed some here (see the section "Increase the Intelligence Community’s Coordination with the BIS" in the PDF) and here, for example.

3anaguma2mo

Does Trump's policy currently prevent the sale of GPUs to China? I thought that Nvidia was able to sell GPUs to China with a 15% fee, but things may have changed.

4Erich_Grunewald2mo

The 15% fee thing would allow NVIDIA H20s, and maybe in the future B30As, to be sold to China, but while these are very useful chips, they are not the best chips. The H20 is especially good for inference but unless and until the B30A becomes available, there is no good legally exportable NVIDIA chip for training workloads. So there is still plenty of smuggling, as there was last year when the H20 was legally exportable without any fee.

[-]ChristianKl10mo23-2

I watched the hearings with RFK Jr. while a lot was about rehearsing existing talking points, it had a few interesting bits:

RFK Jr. is pro-DEI-requirements in clinical trials, so that drug companies have to have a certain number of various groups in their trial population. The example in the hearings was Native Americans. In case you fail to remember, DEI requirements did slow down the clinical trial to approve the initial COVID19 vaccine of Moderna.
When he was asked what he means with gold-standard science (one of his three priorities). On concrete goal he formulated was that RFK Jr. wants that 20% of NIH spending goes to replication studies.
He said that he's not against the childhood vaccine schedule and just wants more safety studies. Unfortunately, no senator asked him directly: "Given your concerns about vaccines, what changes in policy do you want with regards to vaccines?"
Elizabeth Warren made a point that RFK Jr. made a lot of money with suing drug makers. She asked him to pledge not to make money with way in the four years after he leaves office because of conflict of interests and RFK Jr. declined.
It sounded like RFK Jr. was not directly behind the executive orders in re

... (read more)

[-]Mateusz Bagiński10mo3227

pro-DEI-requirements in clinical trials, so that drug companies have to have a certain number of various groups in their trial population

If this is motivated by accounting for biological diversity that might translate into different responses to a drug, then this feels like a very non-central case of DEI at best and I would expect it's not what a majority of people think of when they hear/think "DEI".

4ChristianKl10mo

If you would want to account for biological diversity when doing drug trials you would require pre-existing conditions to be as prevalent in the population that you test in your drug trial. Saying that you have at least X percent obese people would likely give you more clinical knowledge than requiring at least X percent to be Native Americans. Big Pharma generally selects participants to have less preexisting conditions and take less other medication than the general population. In the end FDA, requirements for drug legalization have positive advantages but also make drug trials and new drugs more expensive. There are both arguments to be more for more FDA and for less FDA regulation. In this case, I think it's worth noting that we have a Trump appointee defending some DEI regulatory requirements.

4MondSemmel10mo

Yeah. Though as a counterpoint, something I picked up from IIRC Scott Alexander or Marginal Revolution is that the FDA is not great about accepting foreign clinical trials, or demands that they always be supplemented by trials of Americans, or similar.

4localdeity10mo

I expect that's about not trusting the foreigners who did the clinical trials (I have heard this)[1], and not so much about expecting that Americans are biologically different from foreigners. 1. ^ Specifically, someone with some knowledge told me that the FDA knows that there are some countries where the trials are completely untrustworthy. And that there was a political decision where they said, "If we disallow trials from some countries but not others, there will be much complaining and we'll probably be called racists", and solved the issue by disallowing trials from all foreign countries.

[-]ChristianKl6y150

Elon Musks Starship might bring us a new x-risk.

Dropping a tungsten rod that weights around 12,000 kg from orbit has a similar destruction potential as nuclear weapons.

At present lunch prices bringing a tungsten rod that's weighted 12,000 kg to orbit has a extreme cost for the defense industry that was labeled to be around $230 million a rod.

On the other hand, Starship is designed to be able to carry 100 tons with equals 8 rots to space in a single flight and given that Elon talked about being able to launch starship 3 times per day with a cost that would allow transporting humans from one place of the earth to another the launch cost might be less then a million.

I found tungsten prices to be around 25$/kilo for simple products, which suggest a million dollar might be a valid price for one of the rods.

When the rods are dropped they hit within 15 minutes which means that an attacked country has to react faster then towards nuclear weapons.

Having the weapons installed in a satellite creates the additional problem that there's no human in the loop who makes the decision to launch. Any person who succeeds in hacking a satellite with tungsten rods can deploy them.

4ESRogs6y

Interesting short thread on this here.

1no name2mo

Starships barely meet any of the metrics needed for a good ballistic missile The vehicle is too large to be easily intercepted, lacks stealth capability, intra-atmospheric re-orbiting capability may not have been designed to evade anti-missile systems, the launch platform is difficult to manoeuvre, and the fuel is not storage-resistant, resulting in long response times. The advantage of a recoverable rocket in wartime is also reduced, as it may take even longer to recover and refurbish a rocket than to build a new one. Since it can be used as a FOBS, no one would bet against it carrying a nuclear warhead, so its use would almost certainly mean at least a small-scale nuclear war. If you try to fix these problems, you tend to get one of the DF61 or the B21.Maybe it'll be 9K720. Using the same mass of nukes in a carrier is more likely to destroy cities and armies than using a carrier to deliver tungsten rods. If you want to use this weapon against geotechnical fortifications, then you should look at this. https://rockmech.whrsm.ac.cn/CN/abstract/abstract30303.shtml A logical improvement to recoverable rockets for war purposes would be to use small recoverable rockets propelled by liquid fuel that can be stored for long periods of time to launch downsized ICBMs or satellites, which would allow your nuclear capability to last longer if you want a nuclear war and can't expect to be able to produce missiles after the war but can refurbish them.

[-]ChristianKl1y*143

Annoying civilisational inadequacy:

USB-C cables differ a lot. Some only allow power delivery and no data, while others support different levels of data transfer. Power delivery capabilities also differ.

Most cables do have an E-Marker chips that contain the relevant information. However, Android does not provide that information to the user when they plug into an USB-C cable.

Edit: After looking more into it, it seems while some cables do have E-Marker chips, most don't :(

1CstineSublime1y

This was a shocking revelation to me, I only discovered it a few months ago when I was wondering why one USB-c cable was data transferring between my laptop and an external SSD so much slower than another. What is astounding is, at least in bricks and mortar retail, the price differential between different capabilities of cables. It's so high sometimes as to not even seem like a good deal "this cable costs three times that one, but only charges 30% faster with only one device I have which is capable of that speed of charging"

[-]ChristianKl2y130

Peter Thiel:

You know I think Eliezer Yudkowsky has gone kind of crazy but you know his arguments are not that bad and the people in Silicon Valley do not have great rebuttals to the existential risk of AI.

[-]ChristianKl4y130

An interesting thing about OpenAI's policies is that they ban DALL-E 2 from generating adult images.

It seems like their policy is to ban anything that anyone might object to. Porn that people on the right might object toward and than train their models to avoid being then 'toxic' which seems to be saying things that are politically incorrect for the left.

If that's the general spirit, we might end up with AI that's very restrictive toward what people can do.

6Vanessa Kosoy4y

A lot of people on the left are against porn as well (unfortunately).

3Viliam4y

The obvious extrapolation is that after Singularity humans will be made genderless and sexless. This would simultaneously solve the problems of porn, sexism, and overpopulation.

2Pattern4y

What were Eleuther's policies? Or did that never come up?

2ChristianKl4y

Eleuther's policy is to use the MIT license for their code which basically means you can do what you want with it.

2Dagon4y

It's a weird (and I suspect ineffective or counterproductive) limit to be sure, but the underlying idea of having somewhat arbitrary human-defined limits and being able to study how they work and don't work seems incredibly valuable to AI safety.

2Yitz4y

I’m slightly concerned how it would respond if you prompted it to display a totally innocent situation involving someone whose mere existence is “politically sensitive”. Maybe something like “trans girl reading a hardcover book,” etc.

[-]ChristianKl4y130

Unforunately, my internet failed yesterday during the part of Geoff's stream that had the conversation with Anna Salamon.

Is there a recording available?

2Viliam4y

I am also interested. If there is a public recording, could someone please post a link on LW?

[-]ChristianKl2y120

Metaculus suggests a 30% chance of China invading Taiwan by 2030 or earlier. While I have read some discussion about whether or not the event will happen I have seen very little discussion about how to prepare for the scenario happening.

It seems very neglected because it's uncomfortable to think about that world.

[-]ChristianKl7mo116

If you are thinking about making up new jargon, please ask an LLM beforehand whether there's existing jargon for the thing you are describing. You might have have valid reasons to deviate from the existing terms, but if you have them please be explicit about those reasons.

9faul_sname7mo

Also if the LLM tells you that an existing word points at the concept you asked about, open up a fresh chat without memory and ask what the term it just suggested means.

4Vladimir_Nesov7mo

But that would be slightly inconvenient! Therefore, won't be happening in practice.

[-]ChristianKl2y*110

The FDC just fined US phone carriers for sharing the location data of US customers to anyone willing to buy them. The fines don't seem to be high enough to deter this kind of behavior.

That likely includes either directly or indirectly the Chinese government.

What does the US Congress do to protect spying by China? Of course, banning tik tok instead of actually protecting the data of US citizens.

If you have thread models that the Chinese government might target you, assume that they know where your phone is and shut it of when going somewhere you don't want the Chinese government (or for that matter anyone with a decent amount of capital) to know.

4RamblinDash2y

I feel like this comparison of the enforcement here with the TikTok ban is not directed at the actual primary concern about TikTok, which is content curation by its opaque algorithm, not data privacy per se. By analogy, if a Soviet state-owned enterprise in 1980 wanted to purchase NBC, would/should we have allowed that? If your answer is "no," keeping in mind how many people get their news via TikTok, why would/should we allow what effectively seems to be a CCP-(owned or heavily influenced) company to control what content our people see?

2ChristianKl2y

Politico wrote, "Perhaps the most pressing concern is around the Chinese government’s potential access to troves of data from TikTok’s millions of users." The concern that TikTok supposedly is spyware is frequently made in discussions about why it should be banned. If the main issue is content moderation decisions, the best way to deal with it would be to legislate transparency around content moderation decisions and require TikTok to outsource the moderation decisions to some US contractor.

2Dagon2y

I don't have confidence in my models of how coherent and competent governments are at getting and using data like this. The primary buyers of location data are advertisers and business planners looking for statistical correlations for targeting and decisions. This is creepy, but not directly comparable to "targeted by the Chinese government". My competing theories of "targeted by the Chinese government" threats are: 1. they're hyper-competent and have employee/agents at most carriers who will exfiltrate needed data, so stopping the explicit sale just means it's less visible. 2. they're as bureaucratic and confused as everything else, so even if they know where you are, they're unable to really do much with it. I think the tension is what does it even mean to be targeted by a government.

4ChristianKl2y

The Office of the Director of National Intelligence wrote a report about this question that was declassified last year. They use the abbreviation CAI for "commercially accessible data". "2.5. (U) Counter-Intelligence Risks in CAI. There is also a growing recognition that CAI, as a generally available resource, offers intelligence benefits to our adversaries, some of which may create counter-intelligence risk for the IC. For example, the January 2021 CSIS report cited above also urges the IC to “test and demonstrate the utility of OSINT and AI in analysis on critical threats, such as the adversary use of AI-enabled capabilities in disinformation and influence operations." Last month there was a political fight about warrant requirements when the US intelligence agencies use commercially brought data, that was likely partly caused by the concerns from that report. Here, I mean that you are doing something that's of interest to Chinese intelligence services. People who want to lobby for Chinese AI policy probably fall under that class. I'm not sure to what extent people working at top AI labs might be blackmailed by the Chinese government to do things like give them their source code.

2Dagon2y

[note: I suspect we mostly agree on the impropriety of open selling and dissemination of this data. This is a narrow objection to the IMO hyperbolic focus on government assault risks. ] I'm unhappy with the phrasing of "targeted by the Chinese government", which IMO implies violence or other real-world interventions when the major threats are "adversary use of AI-enabled capabilities in disinformation and influence operations." Thanks for mentioning blackmail - that IS a risk I put in the first category, and presumably becomes more possible with phone location data. I don't know how much it matters, but there is probably a margin where it does. I don't disagree that this purchasable data makes advertising much more effective (in fact, I worked at a company based on this for some time). I only mean to say that "targeting" in the sense of disinformation campaigns is a very different level of threat from "targeting" of individuals for government ops.

2ChristianKl2y

Whether or not you face government assault risks depends on what you do. Most people don't face government assault risks. Some people engage in work or activism that results in them having government assault risks. The Chinese government has strategic goals and most people are unimportant to those. Some people however work on topics like AI policy in which the Chinese government has an interest.

1Pat Myron2y

Leave phones elsewhere, remove batteries, or faraday cage them if you're concerned about state-level actors: https://slate.com/technology/2013/07/nsa-can-reportedly-track-cellphones-even-when-they-re-turned-off.html

[-]ChristianKl4y100

I read about Bir Tawil. It's territory that's currently unclaimed by any country on earth.

If someone would want to fund a new country it seems like a better location than seasteading.

Maybe, both Egypt and Sudan would be willing to recognize a new Bir Tawil state if that state is willing to build roads to Bir Tawil in the territory of both Egypt and Sudan.

6Viliam4y

Wikipedia suggests that some people have already tried this (although most of them not seriously), and were ignored by everyone. I would also expect that if someone tried to actually move to that territory and start building a fence or something, one or the other army would come and kick their butt, regardles of how officially they "do not want" the territory. Also, the countries "not wanting" of Bir Tawil is conditional on their wanting of Halaib, so I would assume that once the conflict over Halaib is resolved, the loser will suddenly "want" Bir Tawil again. I am not saying this can't be done, but I would strongly recommend negotiating about this territory with both governments, trying to reach an explicit agreement like "you will leave me alone regardless of the future status of Halaib", and probably paying them a lot of money to do so. On the other hand, if you could make both Egypt and Sudan recognize you officially, you would have a foot in the door to get recognized by other countries.

2ChristianKl4y

No, nobody did offer to actually build anything. My proposal explicitly spoke about building roads in both Egypt and Sudan. That would be the offer to both governments in return for recognition. Giving those countries free roads in exchange for recognizing the new country is a deal that's worth making for them given that the area is worth nothing to both countries currently.

6Viliam4y

Ah, sorry, somehow I managed to miss that part. That is definitely a good idea, because the roads would provide value to Egypt and Sudan, but also for the new country, so it's win/win.

[-]ChristianKl6y100

Elon Musk seems to have a plan to deploy destructive capabilities to orbit within the next two years that are comparable to the nuclear arsenal of the late forties of the last century.

Little Boy that destroyed Hiroshima had a destructive power of 15 kilotons of TNT equivalents. A napkin calculation on Reddit put BFR to 16.22 kilotons of TNT equivalents.

Refueling in orbit means deploying that much explosive power to rockets in orbit.

There's almost no talk about about the cybersecurity of what he wants to build and it seems doubtful that the process he's currently using takes care of producing structure that keep out determined cyber attackers.

Getting to Mars is nice but it feels like the fact that we haven't had the ethical discussion about proliferation when it comes to Musk is a potential catastrophic error.

4romeostevensit6y

explosive capacity isn't needed when you have rods from god.

2[anonymous]6y

None of those spacecraft will ever reach the bottom of the atmosphere with appreciable orbital velocity remaining, or hit the ground with large amounts of fuel except near to the launch sites.

3ChristianKl6y

Why do you believe that's the case? Why can't a Starship that's full with fuel because it fulled up in space (the infrastructure is necessary for traveling to Mars/Moon) touch earth with a large amount of fuel inside?

1Donald Hobson6y

If a starship full of fuel is in orbit, and gets nudged downward, hitting the earths atmosphere, it gets very hot. If it doesn't have a giant heatshield, it will vaporize the fuel, leading to an explosion in the upper atmosphere. If you used the fuel to slow down, you could reach earth with mostly empty tanks, but still cause some damage if you hit a city.

2ChristianKl6y

It seems to me like fuel wise, a starship has enough full to start from earth, go to orbit and then come down with one tank. Most of the fuel is expended on launch. While you will need to expend some fuel to slow down, I don't see why the starship shouldn't be able to touch earth with a lot of fuel inside.

[-]ChristianKl9mo91

A German legal advice Youtube channel talks about scams via fake voice getting more common and being used against normal people. One of the examples seems to be needing money to make bail.

If you haven't talked about with your parents or grandparents about these kinds of scams, now is the time to find protocols to deal with them.

[-]ChristianKl4y90

It's amazing how mismanged Evernote is.

Their attempt to strategically pivot away from being about remembering information is deeply flawed.

They update their app to a new design and for 3 months the app just crashed when I start it on my phone (I have a Google Pixel 3A which isn't that non-standard).

This Sunday, the app didn't save two notes I made, and now notes can't be saved.

4Dagon4y

Sounds horrible - I'm happy that I mostly use textfiles, and sync them using whatever mechanism works best (currently, Git + iCloud, but that's changed 6-8 times over the last few decades). I find it interesting that you picked "mismanaged" as your root cause, as opposed to "incompetent" or just "failing".

2ChristianKl4y

Releasing a new version when it's very buggy looks to me like a management problem.

4Dagon4y

I don't disagree, but "management problem" is an undifferentiated cause. You can say that everything that seems like a mistake from outside is a management problem. Calling it a QA problem would be more specific (though no more helpful in terms of actions that a bystander can take).

3jp4y

I've had to leave Evernote over the new app, and am so sad about it.

2ChristianKl4y

6 months after writing this, the app can start on my phone but is still unable to create new notes.

[-]ChristianKl2y80

As matter of irony, lsusr decided to censor me from commenting on his posts, so I can't comment on Restricting freedom is more harmful than it seems.

[-]ChristianKl5mo75

For anyone who doubts deep state power:
(1) When Elon's Doge tried to investigate the Pentagon. A bit after that there's the announcement that Elon will soon leave Doge and there's no real Doge report about cuts to the Pentagon.
(2) Pete Hegseth was talking about 8% cuts to the military budget per year. Instead of a cut, the budget increased by 13%.
(3) Kash Patel and Pam Bondi switch on releasing Epstein files and their claim that Epstein never blackmailed anyone is remarkable.

3dr_s5mo

"People can get pressured" or "people can get bribed" or "people sometimes once inside a system discover they are in fact subject to all the same incentives that applied to all the other people inside that system before them" is all you'd get from this, but that's not evidence for anything like a "deep state" unless you meant that term in such a loose meaning that then it would be a trivial discover. Yeah, established organizations have internal politics we don't all know about from the outside. When push comes to shove, the rich donors who have stock in arms sales end up mattering more than Elon Musk or Pete Hegseth.

4ChristianKl5mo

The term deep state originally came from describing a situation in Turkey, where the democratically elected government was powerless in comparison to military. If a newly elected government is unable to execute it's agenda, it's a sign that it's powerless. DOGE lead by Elon Musk cutting costs seemed to be a key part of Trumps agenda, that Trump intended to carry out. The 8% figure from Pete Hegseth was also not just a campaign promise but seeme to be actually part of the Trump administrations agenda. I think Elon Musk spend more on campaign donations to the Trump administration then the combined defense industry. If the influence would just be because rich donors have an interest in certain policy outcomes, why didn't Elon get more for his money? To what extend it's evidence that suggests updating depends on your priors about whether you think DOGE would have actually do anything resembling auditing the Pentagon and whether anything would come out of Hegseth plan to cut the Pentagon budget.

4dr_s5mo

I mean, another option here of course is "those people are full of it, have short attention spans, don't care that much, and/or are so dramatically incompetent at putting out an actual pragmatic plan that all their words are just hot air". But even discarding that, to me the term "deep state" suggests the existence of something that is at least to some extent intentional, clandestine, and has an element of purposeful subversion. If all you mean is "established power structures, interests and practices have inertia and it's not easy to just take a sledgehammer to them as you please, if only because you suddenly realise it's a lot more complex to do that without bringing down the house than you foolishly believed", well, duh.

2ChristianKl5mo

4dr_s5mo

I guess like, a larger organization with some more long term goals? People having friends and associates, exchanging favours or looking out for their own interests is a thing that happens sort of spontaneously. It can lead to some bad outcomes but it's not a particularly interesting insight (if anything, the fact that some societies and organizations can somewhat depart from that is the strange and interesting exception in the landscape of History). I think that's just sloppiness, though. Just like no one in virtually any job environment I've ever been actually respects all the safety and data protection rules and norms. In the government obviously the stakes are much higher, but the people are no more infallible and no less lazy than the guy who sets his password to "password", then writes it on a post-it attached to the screen. Yeah, I mean, obviously people will resist, they will do stuff like malicious compliance or weaponized incompetence to put grist in the gears at every turn if they don't like it. This happens all the time - where there are Republicans, Republicans do it against Democrats too. Again, you can find this frustrating or whatever (I think while a lot of it is frustrating, "anyone who gets in power gets to enact their agenda no matter how insane without any resistance" is also not a desirable condition), but if this is what you'd call "Deep State", then the term means nothing interesting or useful.

2ChristianKl5mo

The Pentagon is a larger organization which does have long-term goals around increasing it's budget and preventing it's budget from being reduced. It also has long-term goals around keeping certain parts of what it does secret that are threatened by DOGE sniffing around. So if I could prove that this is not just sloppy but intentional to reduce information being revealed to congressional inquiries and Freedom of Information Act requests, that would convince you and be a reason to update for you? It's not what "I call Deep State" it's the criteria you proposed.

6dr_s5mo

The Pentagon isn't deep state, it's just state. It would be interesting, sure. I can imagine some of these examples being cases of clandestine communication - as in, you want to say something that you know is illegal, you say it off the record in some channel that isn't monitored. I think it will probably vary case by case when that is the reason though, especially given that if you were doing that you'd still want to use secure channels, just private ones, and some of these instead were ridiculously insecure. So what do you call Deep State? If you mean just "all the administrative and bureaucratic apparatus has in fact inertia and internal politics and does not immediately swerve at the whim of a new elected official" I say, yeah, no shit, that is true, it has been true throughout all of recorded history in every state entity to ever exist, and even better, it's a feature, not a bug, as far as I'm concerned. While the downside is that it'll slow down policies I'd like, it also has a healthy dampening effect against take-over risks. If that wasn't the case then installing a dictatorship would be a lot easier.

2ChristianKl5mo

With Clinton's email server motivations are pretty unclear. If we take Signalgate, using Signal is one choice you can make because you are lazy. Setting the chat to auto-delete after a few weeks is a choice that suggests the intention to avoid the communication becoming a problem later. From what happened at Fauci's NIAID: Morens was stupid enough to write his motivations down, but I would expect that many US government departments run in similar ways.

2ChristianKl5mo

The term deep state was originally used to speak about Turkey's military (and the associated power center). That's what it was coined to describe. There's political power in the military that's separated from the democratically legitimated power. In this case, we did have an administration that had the intention to cut the military budget and audit the Pentagon but the Pentagon was powerful enough to stop that and to instead get their budget increased. It needed a lot more than just inertia and internal politics to accomplish that goal.

4dr_s5mo

I mean, It's the Pentagon. It obviously has all sorts of leverage, as well as personal connections and influence. "If you cut our funding then we won't do X" is enough to put pressure. I'm not saying this is not the case, I'm saying this is... not particularly surprising. Like, anyone who thinks that the true challenge of politics is to figure out the precise orders to give once you're elected, then you can sit back and see your will be enacted as if the entire apparatus of the state was a wish-granting genie is deluded. Obviously the challenge is getting organizations that hold significant power to actually do the thing you want them to. A leftist would talk about the military-industrial complex, or about corporate lobbying (I'm sure the influence of the various big suppliers also matters, as much and more as that of the Pentagon himself). Again, in this sense, if you want to call it "deep state", it's a trivially true thing, and a constant of all polities in history (the military especially! Consider how often personal loyalty of the troops to this or that commander was essentially all that the political stability of a country hinged on). I just don't see much the usefulness of the concept, and I think the word is misleading. None of this is secret or particularly hidden. It's not even just a property of the state. You see the same things play out on a small scale in everyday office politics - the CEO wants one thing, but Team A who's supposed to do it is already busy so they resist it, etc. etc. Organizations at all scales are made of people, people have goals and agency to pursue them. Politics is mostly cat-herding.

4Rafael Harth5mo

I think it definitely is evidence for a deep state (if people couldn't get pressured or bribed, that looks to me like evidence against a deep state), the question is just how much.

2dr_s5mo

That people can get pressured or bribed is the normal state of things we all know about; if that's all the "Deep State" claim boils down to then it's not particularly interesting. You need to operationalize it to something more useful and also falsifiable if you want to take it seriously. Now, if the claim is that these efforts are somehow coordinated and organized on a larger scale than that of interpersonal and/or business relationships, that would be interesting, and a proper conspiracy. But I don't think "people got bribed" is evidence of any kind for that. It's like seeing bank robberies and suggesting there must be a secret society of robbers who all agree to hit banks, rather than banks being the obvious target simply because they're where the money is.

3Rafael Harth5mo

The notion of evidence as rationalists use it is very expansive. B is evidnece for A if and only if P(A|B)>P(A). Or equivalently, if and only if P(A|¬B)<P(A). If people never got bribed (¬B), that sure seems to me like it would lower the probability of a deep state (A). Hence P(A|¬B)<P(A) and therefore P(A|B)>P(A), which means people getting bribed is, in fact, evidence for a deep state. (This is the same with people saying "correlation isn't evidence of causation". Incorrect; correlation is always evidence for causation, because lack of correlation is evidence against causation.) Again, it's the magnitude of the evidence that you can dispute. If you think people will get bribed regardless, then people getting bribed is only very weak evidence for a deep state. But it's still evidence. Just like seeing bank robbers is evidence for a large society of bank robbers. (Because seeing no bank robbers would be evidence against such a society.)

5cousin_it5mo

It feels to me that "evidence of X" as colloquially used might be a stronger phrase than "evidence for X", and almost as strong as "proof of X". So maybe correlation is evidence for causation, but isn't evidence of causation :-)

2dr_s5mo

If something would be equally certain in two cases it can fundamentally constitute no evidence at all, even in the Bayesian sense. Suppose the three alternatives are Honest administrators, business as Usual, and Deep state. We observe some Bribes. Then P(B|U)=P(B|D)=1 and P(B|H)=0. Given appropriate priors then, P(B)=P(U)+P(D). And therefore P(D|B)=P(D)P(D)+P(U) We haven't really updated at all our odds of Deep state being the explanation instead of business as Usual. We have ruled out the Honest hypothesis but I didn't have many illusions about that. It's a fairly extreme perspective - you could get a tiny update if you posited that P(B|U) is slightly lower than 1 or whatever - but I think for all practical matters it does round up to this calculation.

2Rafael Harth5mo

Yes, if you assume that the probability of seeing an observation was 100% under your favorite model then seeing it doesn't update you away from that model, but that assumption is obviously not true. (And I already conceded that the update is marginal!)

2dr_s5mo

I'd say the probability of seeing some resistance or corruption in virtually any administration is damn close to 100%.

[-]ChristianKl10mo71

For all the talk about fraud at USAID, Elon Musk had not provided evidence that any single person did something that's fraud in the legal sense.

All the examples he provided are programs that he considers to be wasteful. Most of those programs are listed at USAspending.gov. They are not secret projects that needed DOGE to go into USAID computers to find out.

Mainstream media journalists should ask at the White House press briefings whether DOGE has found any fraud that it referred to the DOJ for prosecution.

[-]ChristianKl11mo74

Drug legalisation is probably the best way to prevent fentanyl deaths. Many of the people who are fentanyl addicted are addicted because they wanted to buy other drugs buy got them laced with fentanyl. Drug legalisation will allow for quality control and end the lacing with fentanyl.

Now, nitazene which is even more potent than fentanyl gets added and might produce similar effects as fentanyl where people die to nitazene overdose.

If the US wants to actually prevent those deaths drug legalization that allows for quality control is the step forward that would work.

[-]ChristianKl1y*75

When reading posts about AI development I get the impression that many people follow a model where the important variables are the data that, out there in the world, the available compute for model training and the available training algorithm.

I think this underrated the importance of synthetic training data generation.

AlphaStar trained entirely on synthetic data to become much better than humans.

There's an observation that you can't improve a standard LLM much by retraining it by just feeding it random pieces of it's own output.

I think there's a good chan... (read more)

4Viliam11mo

I don't understand these things much, but it seems to me that we have two different kinds of learning here. In chess, the chessboard is the territory, and the AI has a direct access to the territory (it can represent the chessboard in its memory). This is what allows the AI to run experiments and learn from them. It can discover new things by trying random changes and observing the outcome. With LLMs, the territory is "out there", and the AI only has an access to human maps of the territory. It can surpass humans by synthesizing millions of human maps and connecting the dots, but at the end of the day, they are just maps. Which means two problems: * gaps in the knowledge: possible experiments that no one ever did -- and the AI can't do them either * noise: if some information is wrong, AI can't check it, other than comparing to other (unreliable) maps Feeding the AI its own output doesn't address any of these two problems.

4ChristianKl11mo

With o1 and o3, we found out that you get better output if you spent more compute on getting the answer. Letting an LLM work 10 minutes via chain of thought on getting an answer gives you better answers than letting it work 2 seconds on getting an answer. By training on it's own answers you could get the LLM to give the kind of answers it gives with 10 minutes of chain of thought worth of compute in 2 seconds without chain of thought. The difference between a person with an IQ of 90 and one with an IQ of 180 is not in gaps of knowledge or having access to information that's right or wrong but in reasoning ability. As far as the territory of LLMs being "out there", that's true for a subset of the territory. The field of mathematics isn't "out there". You don't need to run any experiments in physical reality to validate claims about math and see whether you get better at creating math proofs. When it comes to computer programming, some tasks require interaction with physical reality but plenty of tasks are "write a patch that includes a functionality that passes all the unit tests and passes some clean code metric". You also have "find a possible input for a function that makes it crash" and a few other tasks where you can automatically evaluate the outcome of what the LLM is doing without needing any human input.

7gwern11mo

There are enormous differences in gaps of knowledge and further, self-selection into niches and lifestyles and skills like looking up information in Google Scholar, between an IQ 90 and a 180 person. Look at vocab norms or simple ordinary trivia questions such as 'does the earth go around the sun or vice-versa?'. You can't do much reasoning about what you don't know about. (This is one of the biggest reasons that 'retrieval heavy' LLMs have underperformed so much. There is no 'small logical core' you can easily cheaply learn free of actual real-world knowledge. At best, like the Phi series, you can steal reasoning from a much larger model like GPT-4 which learned it the hard way and has predigested it for you.)

[-]ChristianKl1y71

With David Sacks being the AI/Crypto czar, we likely won't be getting any US regulation on AI in the next years.

It seems to me like David Sacks perspective on the issue is that AI regulation is just another aspect of the censorship industrial complex.

To convince him of AI regulation, you would likely need to have an idea about how to do AI regulation without furthering the censorship industrial complex. The lack of criticism of the censorship industrial complex in the AI safety discourse now is a big problem because there are no available policy proposals.

4Orpheus161y

Can you quote (or link to) things Sacks has said that give you this impression? My own impression is that there are many AI policy ideas that don't have anything to do with censorship (e.g., improving government technical capacity, transparency into frontier AI development, emergency preparedness efforts, efforts to increase government "situational awareness", research into HEMs and verification methods). Also things like "an AI model should not output bioweapons or other things that threaten national security" are "censorship" under some very narrow definition of censorship, but IME this is not what people mean when they say they are worried about censorship. I haven't looked much into Sacks' particular stance here, but I think concerns around censorship are typically along the lines of "the state should not be involved in telling companies what their models can/can't say. This can be weaponized against certain viewpoints, especially conservative viewpoints. Some folks on the left are trying to do this under the guise of terms like misinformation, fairness, and bias."

4ChristianKl1y

I believe about Sacks views comes from regularly listening to the All-In Podcast where he regularly talks about AI. Sacks is smarter and more sophisticated than that. In the real world, efforts of the Department of Homeland security that started with censoring for reasons of national security ended up increasing the scope of what they censor. In the end the lab leak theory got censored and if you would ask the Department of Homeland security for their justification there's a good chance that they would say "national security".

2Orpheus161y

Do you have any quotes or any particular podcast episodes you recommend? Yeah, I agree that one needs to have a pretty narrow conception of national security. In the absence of that, there's concept creep in which you can justify pretty much anything under a broad conception of national security. (Indeed, I suspect that lots of folks on the left justified a lot of general efforts to censor conservatives as a matter of national security//public safety, under the view that a Trump presidency would be disastrous for America//the world//democracy. And this is the kind of thing that clearly violates a narrower conception of national security.) How to exactly draw the line is a difficult question, but I think most people would clearly be able to see a difference between "preventing model from outputting detailed instructions/plans to develop bioweapons" and "preventing model from voicing support for political positions that some people think are problematic."

2ChristianKl1y

I don't have specific recommendations for the past. I would expect a section in the next All-In Podcast in which David Sachs participates to law out his views a bit. That's the question you would ask if you think the person who's drawing the line is aligned. If you think the people speaking about national security and using that to further different political and geopolitical ends are not aligned, it's not the most interesting question. It sounds to me like you are taking this as an abstract policy issue while ignoring the real-world censorship industrial complex. It's like discussing union policy in the 1970s and 1980s in New York without taking into account that a lot of strikes are because someone failed to pay the Mafia. If you don't know what the censorship industrial complex is, Joe Rogan had a good interview with Mike Benz, who is a former official with the U.S. Department of State and current Executive Director of the Foundation For Freedom Online.

1Noosphere891y

The fundamental problem is that any effective AI alignment technique is also a censorship technique, and thus you can't advance AI alignment very much without also allowing people to censor an AI effectively, because a lot of alignment work is aiming to make AIs be censored in particular ways.

7Jozdien1y

I disagree with the use of "any". In principle, an effective alignment technique could create an AI that isn't censored, but does have certain values/preferences over the world. You could call that censorship, but that doesn't seem like the right or common usage. I agree that in practice many/most things currently purporting to be effective alignment techniques fit the word more, though.

4Noosphere891y

I admit this is possible, so I almost certainly am overconfident here (which matters a little), though I believe a lot of common methods that do work for alignment also allow you to censor an AI.

4ChristianKl1y

If you take early writing of Eliezer, the idea is AI should be aligned with Coherent Extrapolated Volition. That's a different goal from aligning AI with the views of credentialed experts or the leadership of AI companies. "How do you regulate AI companies so that they aren't enforcing Californian values on the rest of the United States and the world?" is an alignment question. If you have a good answer to that question, it would be easier to convince someone worried about those companies having enforced Californian values via censorship industrial complex doing the same thing with AI to regulate AI companies. If you ignore the alignment questions that people like David Sachs care about, it's hard to convince them that you are sincere about the other alignment questions.

2Noosphere891y

A crux here is that I basically don't think Coherent Extrapolated Volition of humanity type alignment strategies work, and I also think that it is irrelevant that we can't align an AI to the CEV of humanity.

[-]ChristianKl6mo62

Asking ChatGPT, Claude or Gemini for help when having problem on the computer is great. Sometimes typing up the problem takes time.

Taking a quick screenshot of an error message and giving that to ChatGPT, Claude or Gemini often means that you don't need to fully explain the problem. It saved me a lot of time. It also allows you to copy over a lot of information from the screen that might or might not be relevant for the LLM to help you.

Windows has: [WindowsKey] + [Shift] + S
Apple has: [Command (⌘) ] + [Shift] + 4

[-]ChristianKl1y61

If Biden pardons people like Fauci for crimes like perjury, that would set a bad precedent.

There's a reason why perjury is forbidden and if you just give pardons to any government official who committed crimes at the end of an administration that's a very bad precedent.

One way out of that would be to find a different way to punish government criminals when they are pardoned. One aspect of a pardon is that they remove the Fifth Amendment defense.

You can subpoena pardoned people in front of Congress and ask them under oath to speak about all the crimes... (read more)

[-]ChristianKl2y60

According to the South China Morning Post's summary of the Xi-Biden talks:

Among concrete results of the summit, the two sides agreed to cooperate on narcotics control and artificial intelligence governance, and resume military-to-military communication. But China voiced its continuing discontent with several US policies it believes hold it back, including export controls, investment reviews and unilateral sanctions.

For anyone who thought that cooperation between the US and China on AI governance is impossible this should be seen as great news.

[-]ChristianKl3y60

According to their Discord, RaDVaC plans to start human trials early next year.

[-]ChristianKl5y60

Running simulations of driving situations is a key feature of how machine learning models for driverless cars get trained.

Maybe a key reasons for why humans dream is to allow us to simulate situations and learn to act in them?

[-]ChristianKl2y*50

While I still don't feel like I understand electrolytes as well as I would like to, I become more convinced that supplementing potassium when one engages in activities that produce sweating is worthwhile.

Over the last year I started using potassium carbonate like a spice and whether or not it feels tasty depends a lot on how much I was sweating in the day before the meal.

Giving that summer comes up, if you aren't already supplementing electrolytes for those days that are warm enough to make you sweat, I recommend you to get some potassium carbonate a... (read more)

2Lorxus2y

The body uses up sodium and potassium as two major cations. You need them for neural firing to work, among many other things; it's the body's go-to for "I need a single-charge cation but sodium doesn't work for whatever reason". As such, you lose plenty in urine and sweat. Because modern table salt (i.e., neither rock salt nor better yet sea salt) contains basically no potassium, people can end up being slightly deficient because we do still get some from foods - lots of types of produce like tomatoes, root vegetables, and some fruits are rich in it, for instance.

2ChristianKl2y

In addition to that from my perspective, I think that if every day of the year you consume the same amount of potassium you (as a typical office worker) likely consume either too much or too little on some days.

1lemonhope2y

"Lo-salt" salt is salt with potassium. That's been my table salt for 5 years.

2ChristianKl2y

That's certainly also an option. I personally found for myself, that I feel intuitively less drawn to NaCl+KCl than to NaCl + K2CO3 (I have both at home). Most supplements that have mixes of electrolytes don't seem to use KCl and so would give you relatively less chloride than the NaCl+KCl mix.

[-]ChristianKl2y5-7

Elon's idea of building a thousand Starships per-year to get to Mars seems ill-thought-out.

Starship is very well designed for bringing objects into orbit and down from orbit but not for the interplanetary journey.

For the interplanetary journey, you likely want to have a ring-space-station that's propelled by ion thrusters.

Having a ring-space-station means that it's easy to produce artificial gravity and generally have the infrastructure to have a good journey for more people.

6Thomas Kwa2y

I don't think you can power the ions with current technology. See this article for power limitations-- 6 kW/kg is required for a 1 month journey, but to be any faster than a Hohmann transfer you'll still need power in the kW/kg range, which we don't have the technology for, either solar or nuclear. In this design half your mass will be argon and most of the rest will be solar panels, which is likely worse than Starship mass ratios to Mars. Maybe you can match Starship mass ratios if you do aerocapture, but it seems implausible to aerocapture a whole ring station, and why would you use future technology just to match current technology? Artificial gravity seems possible with two Starships connected by a cable. You do get more space with a ring station, so maybe it could be luxury or second-generation accommodations.

2mako yass2y

Cole Nielson-cole is working towards designing fiber composit construction stages for space, he has thoughts about this, in short, microwave lasers as energy transmission and rectifying antennas as energy receivers. But he doesn't get into the topic of lasers and I'm pretty sure we don't have that today, right? But I thought the whole interview was great.

3Carl Feynman2y

When you get there how do you get down? You need spacecraft capable of reentry at Mars. There’s no spacecraft factory there, so they all have to be brought from Earth. And if you’re bringing them, you might as well live in them on the way. That way you also get a starter house on Mars. Anyway, that’s the standard logic.

2ChristianKl2y

You need to send some Starships to get down to the surface on Mars but you could likely do that job with a handful of starships. You don't need to produce 1000 starships per year to do that.

7Carl Feynman2y

I’m confused. Suppose your ring-shaped space hotel gets to Mars with people and cargo that weighs equal to the cargo capacity of 1000 Starships. How do you get it down? First you have to slow down the hotel, which takes roughly as much fuel as it took to accelerate it. Using Starships you can aerobrake from interplanetary velocity, costing negligible fuel. In the hotel scenario, it’s not efficient to land using a small number of Starships flying up and down, because they will use a lot of fuel to get back up, even empty. Would you care to specify your scenario more precisely? I suspect you’re neglecting the fuel cost at some stage.

[-]ChristianKl2y50

If China wants to wage war over Taiwan, the situation is easier if all US military is occupied in the war elsewhere in the world.

The US already depleted a lot of its ammunition stockpiles by supporting Ukraine. If the US starts a war with Iran, US military capacity would be further strained.

From the Chinese perspective that might give a good time to fight over Taiwan when a lot of the capacity of the US military is not available to defend Taiwan.

[-]lc2y100

2ChristianKl2y

J. Kyle Bass gave a talk that made me update in the direction of China being serious about taking step to take over Taiwan in the near future:

8Alexander Gietelink Oldenziel2y

They probably will lose a conventional war against the US. This is my current assessment after a moderate amount of research into the topic.

2RHollerith2y

Agree. They probably lose a nuclear war against the US, too, ending with US forces occupying all their seaports or at least imposing such an effective naval blockade that they might as well be occupying the ports.

1FlorianH2y

I guess both countries would lose a nuclear war, if for a weird reason we'd really have one between US and CN

1FlorianH2y

In the grand scheme of things, that would not matter much. If China wants to fully reintegrate Taiwan, it can today, or else simply at lastest in a few years. I guess if China is not doing that in the near future, the main reason will be that (i) there is simply no big enough value in it and/or (ii) there is significant value for the government to have the Taiwan issue as a story for its citizen to focus on/sort of rally-behind-the-flag effect. But less so the effect of US deterrence.

2ChristianKl2y

The problem is that using Taiwan as an area for citizens to focus on is that Chinese citizens expect their government to take steps to reintegrate Taiwan. Careers inside the CCP are built on taking steps to reintegrate Taiwan. If the Chinese start to believe that Taiwan can be taken in a short time frame and Xi does not take it, that's not good for his reelection bid in four years. He needs to tell a narrative about why he didn't move to take it that's compatible with what people in the CCP want to hear. "We didn't take Taiwan because China is too weak and the US is too strong" might not be a narrative that Xi wants to tell.

[-]ChristianKl4y50

StackExchange websites have a feature where questions with a lot of engagement can be tagged in a way that prevents new users from answering them. This is a way to prevent low quality answers.

It seems to me that there are questions like my recent post How would you run the statistics on whether Ivermectin helped India reduce COVID-19 cases? where it would be valuable to have a similar mechanism, as I see a person posting anecdotal links when the question isn't about having anecdotes but searches for a higher level of evidence. In general such a status woul... (read more)

3CraigMichael4y

Is there a way to submit and vote for features? Would support this.

2Pattern4y

Or a way to filter by that.

[-]ChristianKl5y50

I messed up one of my knees by not leaving my flat for 3 weeks followed by a long walk with some inclination.

If COVID motivates you like me to spend much time inside don't overexert yourself by putting your body under much stress at once.

[-]ChristianKl6y50

One of the major problem with getting marketing emails is that we lack good feedback mechanisms to incentivize companies whom we do give our emails because we do want to get some information not to spam us with other information that we don't want to receive.
At the moment we have two options to punish companies who abuse the relationship. We can click on “mark as spam" or we can unsubscribe.
The first version is a punishment as it means that more emails of the company end up in spam folders. Unfortunately, the company usually doesn't know th... (read more)

6Trinley Goldenberg6y

Marketers are already getting much of this data via click through rates and open rates. They care much less about "how much you like an email" and much more about "how much an email is likely to make you buy in the future". The problem of course, is that people who aren't buyers being annoyed by the email is a negative externality. It doesn't affect the marketer's bottom line at all if someone who was never a buyer gets annoyed. It slightly effects them if someone who was a potenjtial buyer gets annoyed, but only if that causes them not to buy in the future (which is reflected in CTR and Open rates). The only way to have marketers not take advantage of a free marketing channel is to better align incentives. One way to do that would be to make it not free, as jacobjacob talked about in another thread. Collective spam filters like in gmail also provide a slight incentive for this, as messages being marked as spam will cause them to be marked as spam in customers' inboxes as well. As you said this isn't perfect because marketers don't know WHICH messages are being marked as spam, but in general this feels decently solved, for instance most email marketing platforms have a "spam score" that will tell you if you're likely to be filtered to the spam filter before you send, using the data THEY have on which messages are marked as spam.

4Raemon6y

Minor note: Jacobian and jacobjacob are different people

2Trinley Goldenberg6y

Whoops, edited.

2ChristianKl6y

In the end they do care about the fact that people buy, but the fact that marketers care about metrics like open rates suggest that it's useful for them to have more information. A lot of emails are send out as a form of content marketing where the goal of the company is to create a trusted relationship which can be later monetized. In those cases it's not easy to measure the effects of an email on sales months down the road. The fact that the marketing platforms have a spam score doesn't mean that the spam score accurately captures the spamminess when it comes to how annoying the email is to customers.

2Raemon6y

On a higher level, email clients could make the "mark as spam" button send information to the sender.

2ChristianKl6y

I think because of marketing and branding reasons that's not a valid move for the companies that produce most email clients.

[-]ChristianKl4y40

One interesting way to reduce maze levels and monopoly power would be to make it harder for industries to consolidate to fewer players.

One possible policy would be to tax buying stakes in limited-liability companies by limited-liability companies. 20% might work.

Hedge funds that play a valuable economical role can do that under corporate structures that don't include limited liability. This would likely reduce the likelihood that individual hedge funds are "too big to fail". The owners of those hedge funds would then have more skin in the game.

2Dagon4y

Why target change rather than level? Taxing organizations by total size or by levels of management might be closer to what you're seeking. Or, more radically, doing away with limited liability for corporations - make officers (and shareholders!) liable for corporate actions. These choices do, of course, also limit the willingness to take risks and overall decrease civilizational capability. Whether you consider that to be additional valuable slack or a significant reduction in overall welfare is a modeling choice :)

2ChristianKl4y

I'm not targeting chance in size. If a company invests in valuable technology and increases in size as a result, it wouldn't face taxes in this proposal. I'm targeting the activity of buying stocks by limited liability corporations. I think stocks should generally be held by private individuals or institutions that are not limited liability corporations. I want companies to take risks by investing in technology and not take risks by buying up stocks of companies. If I look at a company like Pfizer I would want them to reinvest their profits into new research technology instead of buying back their own shares or buying up other companies. Mergers make markets less competitive and can be done by CEO's for reasons that are in the interest of the CEO but neither their company nor society as a whole. A 20% tax would reduce the mergers to those where a really strong case for synergy can be made.

2Dagon4y

Hmm. I model "targetting the merger" as worring about the path, where "targetting the resulting structure" would be preferable, whether it occurs by growth, acquisition, or initial setup. I see "Mergers make markets less competitive" as biasing toward the status quo, and privileging the same result created by non-merger mechanisms. I'm curious whether I'm wrong on this, and you see the merger path to that structure as the main problem, or whether I'm misunderstanding the reasons for your proposal.

2ChristianKl4y

If you have a market where you have a large company with a lot of market power but that company is run relatively badly in a world without mergers that large company will lose. With mergers, it can be possible for the badly run company to just buy off potential competitors. I want people who are able to effectively make investments in technology to be in control of large amounts of capital instead of people who are clever about company politics and merging being in control of that much control. We need companies like Intel who can build a 20 billion dollar microchip factory and for that reason having laws that directly forbid large companies would create a lot of damage.

[-]ChristianKl5y40

Cancer researchers spend the last decade telling everyone "cancer isn't a disease". On the other hand we have antiaging people saying "aging is a disease".

The two strategic positions are interesting to compare and given that cancer gets as much spending and attention it's worth thinking about whether the strategy of the antiaging people is right.

2Viliam5y

The difference is that cancer researchers already have funding, and they need an excuse for why they haven't found a reliable cure yet. Anti-aging researchers need money. Saying "X is a disease" implies that it should be cured.

6ChristianKl5y

There are more downstream effects. One is that it allows companies to put drugs on the market that otherwise wouldn't be allowed on the market because they can use the orphan drug act when the target a specific form of cancer that is below the limit of the orphan drug act. It's more complicated then that. It implies that aging is a cluster that you should be able to diagnose in people, then develop a drug that treats it and get FDA/EMA approval and then having health insurance pay for it. I recently read about Jenna Luche-Thayer's battle for more ICD codes for different forms of lyme and the importance of those. That's a similar position to that of the cancer researchers in a field where there's not much funding. If we would buy the 7 hallmarks model, one conclusion would be that aging is 7 diseases. That means 7 things you can diagnose, get drugs approved for and get health insurance to pay for.

2Viliam5y

Ah, the legal implications of words. "Words as legal hacks" is even crazier version of "argument as soldiers".

[-]ChristianKl5y40

Is a press interview of the German magazin Spiegel Sierk Poetting (Chief Financial Officer and Chief Operating Officer of Biontec) said that Biontec had no room for funding in 2020 and additional money wouldn't have allowed them to scale up vaccine production faster but they have now room for funding.

http://cdn.www.spiegel.de/producing/SPIEGEL_2021_06.pdf (DER SPIEGEL Nr. 6 / 6. 2. 2021 Site 64)

[-]ChristianKl5y40

The public criticism of Russia's vaccination efforts seem strange to me. Claiming that Russia only wants to do early vaccinations because of reasons of national prestigue and not because of health and economic damage of COVID-19 seems to me like too many people still haven't understood that COVID-19 is a serious issue that warrents doing what we can.

6avturchin5y

The vaccine will be available for general public after January 1, and before this it will be available to medics and teachers only - so it will be like phase 3 of the clinical trial.

[-]ChristianKl6y40

It seems that every post gets tagged with world modeling or world optimization. We should likely have a more focused definition or those tags to make them more specific.

[-]ChristianKl6y40

I remember reading a link to a long article this month that was about how the New York times is very narrative driven and that the editors often decide on the narrative of the article before going out to research it. Does anybody know which article I mean?

[-]Thomas Kwa6y120

This one?

It was a shock on arriving at the New York Times in 2004, as the paper’s movie editor, to realize that its editorial dynamic was essentially the reverse. By and large, talented reporters scrambled to match stories with what internally was often called “the narrative.” We were occasionally asked to map a narrative for our various beats a year in advance, square the plan with editors, then generate stories that fit the pre-designated line.

For me, an article linking to this one was the fifth Google result for "new york times narrative driven".

2ChristianKl6y

Yes, thank you. I did search for NYTimes and didn't think of using the fullname.

[-]ChristianKl8mo30

There seems to be papers that show that if you naively train on chain of thought, you train models not to verbalize potentially problematic reasoning in their chain of thought. I however don't see discussion about how to train chain of thought models to better verbalize their reasoning.

If you can easily train a model to hide it's reasoning you should also be able to train models the other way around to be more explicit about their reasoning.

One approach I imagine is to take a query like diagnosing medical issues and replace key words that chang... (read more)

[-]ChristianKl11mo30

With Trump speaking about the US making Greenland a US territory, it would be worth speaking publically about how the US screws their territories with the Jones Act.

It might be a way to build momentum to get rid of the Jones Act.

[-]ChristianKl3y30

With constantly reduced costs of photos from satellites, satellite privacy will likely become an issue in the coming decade.

With current laws, every inch of land that's visible from the sky will get 24/7 surveillance.

Especially for people who don't share their garage with a lot of other people that will mean that everywhere they drive with their car can be public knowledge.

[-]ChristianKl4y30

For all the intensity of advocating that a tough stance toward Russia in the Ukraine conflict is important for deterring China from taking Taiwan, where's the support for Lithuania and Slovenia when they get pursued by China for being pro-Taiwan?

[-]ChristianKl4y30

The Ukrainian war has the potential to turn into a cyber war. Russian cyberattacks on Ukraine have damaged non-Ukranian targets is a good time to think through your OPsec.

Potential cyberwar is a good reason to think through your backup strategy, make sure that you use complex enough passwords with a password manager, use second-factor-authentification and have your software updated.

[-]ChristianKl4y30

It's amazing how the current debates bring people who profess that they believe in science to reject core ideas of the enlightment about science not being about believing authority and attack modern tools of evidence-based medicine like meta-reviews as flawed.

2Matthew Barnett4y

Can you present an example of what you were thinking about here?

8ChristianKl4y

Meta-reviews were an invention 1-2 decades ago, to have a better tool then just authority-based judgements of how to summarize literature. I have seen peer-reviewed meta analysis on ivermectin getting rejected because they differ with statements from authorities like the CDC which feels like rolling back the progress of the last two decades. It used to be that on Skeptics.SE the idea that medical claims should be decided by peer-reviewed papers was accepted.

[-]ChristianKl5y30

Tesla recognizing that Bitcoin is bad for the enviroment shows why Bitcoin will lose to proof-of-stake currencies.

Bitcoin advocates argued that with Tesla buying Bitcoin is a sign that companies in general will do so. We live in a world where any companies that does that is likely going to be downrated on ESG rating while holding technologically more advanced crypto-currency like Polkadot and Ethereum (if 2.0 works) won't lead to ESG downrating.

The proof of stake currencies can provide low transaction fees that make them more suitable as an act... (read more)

[-]ChristianKl5y30

Room humidity matters for COVID-19 transmission. If you are going to spent a lot of time in the same rooms as other people in the next months, invest into proper humidity to reduce your risk of getting ill: https://aaqr.org/articles/aaqr-20-06-covid-0302?fbclid=IwAR3zFZ-UqSjBlc2DJUjHI5yUKTujIW5WyDlwgogmfAAIJtEAxoCas-LkdWc

2David Scott Krueger (formerly: capybaralet)5y

Can you give us the TL; DR on what "proper humidity" means in this context? Google says 30-50% is good (in general). Is the same true for COVID?

4ChristianKl5y

https://www.condairgroup.com/humidity-health-wellbeing/scientific-studies/criteria-for-human-exposure-to-humidity-in-occupied-buildings shows how the effect on humidity on various risk factors and based on it I would suggest 50-60% is ideal for fighting viruses.

[-]ChristianKl5y30

Thinking more about the Russian vaccine is sad. There's no discussion in the media about what risk we should actually expect from the vaccine. The scientists that get asked by the media to comment are only asked to talk about the general policy of clinical trials but not about the underlying biology.

[-]ChristianKl2mo20

I somehow had a cached thought from looking at Metaculus statistics of myself in the past that my performance was more middle of the pack. Yesterday, I looked at the statistics again and it ranks me, Rank: #10 out of 4859 out of Time Period:2016 - 2020.

I'm not exactly sure how I got my impression in the past, maybe I was better at the more long running predictions? Or did Metaculus change something about how it calculates scores?

[-]ChristianKl3mo20

I had an annoying issue with building the Android part of my Flutter App. Long twenty step iterations with GPT-5-pro and o3-pro did not fix the issue. Neither was codex-1 able to fix the issue.

Now, with two tries GPT-5-codex managed to fix my problem. GPT-5-codex frequently manages to work for 100-110 minutes to actually fix problems which the old codex-1 wasn't able to fix well.

OpenAI seems to be surprised that GPT-5-codex usage overall needs more compute and I would guess that's because GPT-5-codex is willing to work longer on problems. It might be that they nerve it temporarily so that it won't as eagerly work for two hours on a problem.

[-]ChristianKl3mo2-7

In his recent All-In talk Elon Musk talked about how he expects all energy by the sun to be harvested sooner or later by AI. This means building a dyson sphere around the sun. The economic reasoning that forces the building of a dyson sphere is quite straightforward.

While a dyson sphere also complete solves our global warming problem on earth, it also has a good potential to create another extinction event on earth.

I wonder whether this might be a scenario that's more useful to talk about when trying to explain the risk of AGI killing humanity.

8Vladimir_Nesov3mo

A Dyson sphere implies the scale of industry that makes any Earth-scale concerns trivial. If the world didn't care about Earth or its inhabitants, they'd be long gone by the time construction of a Dyson sphere is well underway, as a side effect of the smaller projects that came earlier. Conversely, if there was a sufficient level of care for them to persist in some way until that point, then for the same reasons they persisted previously they'd also persist through construction of a Dyson sphere (even if not on the literal Earth with light from the literal Sun). But also, the scale of industry implied by construction of a Dyson sphere makes it a dubious goal, since the Sun isn't producing energy in a way that's at all sensibly optimized or long term stable. Fusion power plants could be doing similar work better, and there is a whole lot of matter trapped in the Sun that could find much better uses. Star lifting the Sun seems very likely the better aim, even if constructing a Dyson sphere would be an intermediate step, mere scaffolding for the ultimately more worthwhile project of fully dismantling the Sun.

[-]ChristianKl6mo2-1

They are saying that naming is hard, but can you be any worse at naming than OpenAI is?

One the one hand, Sora is a text-to-video model. On the other hand, Sora is an web UI that allows both using a text-to-image and a text-to-video model.

One the one hand Codex is a model created years ago and on the other hand it's a web UI with a model released in May 2025.

I have no idea what the o in o1/o3/4 stands for.

2Cole Wyeth6mo

I’ve heard it stands for “Omni”

1ghost-in-the-weights6mo

Confusingly I believe o stands for "Omni" in the context of GPT-4o, since it's "omni-modal". Based on some quick googling, the o in o1/o3/o4 seems to emphasize that o1 was resetting the counter back to 1 (so it's more like zero1).

1ChristianKl6mo

Given that o1-pro and o3-pro are models that can't deal with attachments and analyze images, while GPT 4.5 (that has no o in the name) can, that seems to make the naming convention even more awkward.

[-]ChristianKl1y20

It's my sense that most people have severely reduced mental facilities in the year before they die.

At the same time, the public knowledge suggests that all last ten popes had well working mental facilities before they die.

What's going on here? Does the church have a mechanism to "retire" popes that lose their mental facilities through unnatural death?

Is the church just very lucky?

4Dagon1y

Be VERY careful distinguishing different uses of "most" ("many", "very many", "almost all", "all except a few exceptions"), especially when applying to outliers on other unmeasured dimensions. My expectation is that mental facilities aren't the most critical feature of popes, so they're typically selected such that even a reduction is still sufficient. And there's likely also active obfuscation of older popes' mental acuity, so it's not as obvious to the public. I won't speculate on whether there are other measures taken to make sure that living popes are at least somewhat mentally capable. It's quite likely that they're lucky in that it doesn't get to that point very often. edit: I bothered to actually look for articles on the topic. It seems Benedict XVI did retire in 2013, and didn't die until 2022. And the current pope is denying he's planning to retire, but is open to the idea that a health decline could change that. Previous to Benedict XVI's retirement, it had been ~600 years since a pope had failed to die in office. I think my speculation stands that a pope's duties (especially until very recently, when video coverage makes papal activities and appearances extremely widely public) can be compatible with a LOT of health and cognitive degradation.

[-]ChristianKl2y20

I used to think that while JFK jr. has a lot of unorthodox positions he has some understanding of the basics mechanisms of how governance works.

JFK jr. has been censored a lot and spoke a lot about being pro-free speech. When it comes to the Israel conflict he however advocates for censoring calls for genocide.

He gave an interview where he defended that position by saying that speech that incites violence (like genocide) is not protected by the First Amendment. In reality, only speech that incites imminent lawless action is forbidden by the Brandenburg sta... (read more)

[-]ChristianKl3y20

Is one updates to a pro-ChatGPT account is it possible to use GPT-4 for as many queries as one would have used ChatGPT before?

[-]ChristianKl3y20

I'm trying to understand the quality of German regulation regarding clinical trials better.

It seems that back in 2016 we had a law that allowed a government agency to allow clinical trials even if an ethic commission voted against the clinical trial.

One newspaper article had a quote from the head of the ethics commission of the university of Munich:

Insgesamt geht es um eine Beschleunigung des Verfahrens, und die kann nicht im Sinne der Patienten sein."

Translated: Overall it's about speedup of the process and that can't be in the interest of the patient.

[-]ChristianKl3y20

ChatGPT doesn't want to joke about science:

As a machine learning model, I do not have the ability to create original jokes or humor. I can provide information and support to help answer your questions, but I am not able to generate jokes on my own.
In general, jokes are a form of humor that rely on wordplay, surprise, or incongruity to create a humorous effect. They often involve a setup that establishes a certain expectation, and a punchline that subverts that expectation in a surprising or unexpected way. Jokes can be difficult to create, as they require

... (read more)

[-]ChristianKl4y20

PSA: There are now good FFP3 masks ( https://smile.amazon.de/-/en/gp/product/B00VAT74NG/ ). If your plan doesn't involve getting infected with Omicron this is the time to upgrade your protection.

FFP2/N95 masks are not good enough anymore.

[-]ChristianKl4y20

I have a lot of uncertainty here so, let me write a shortform for now. I'm not sure to what extend the following thoughts are true and I'm happy about comments

It seems that the body has two immune defense levels. One is in the mucosal immune system and there's a second that leads to antibodies in the blood.

SARS-CoV-2 infections usually first are in the upper respiratory tract where the mucosal system provides defense and not in where the immune system that provides antibodies that are active in the blood can fight the infection. https://www.frontiersin.org... (read more)

[-]ChristianKl4y20

Yesterday, I talked with a friend who together with her boyfriend got COVID. He was vaccinated, she wasn't. She had to go to the hospital while he didn't. However both have now the same long-COVID symptoms.

It's an anecdote and I'd really love if someone would actually study the question of how effective our vaccines are for preventing long-COVID...

[-]ChristianKl4y20

I heard a lot about René Girard recently but never read any of his works. Does anyone have a recommendation about what is the best of his books to start with?

[-]ChristianKl4y20

"To 'take over the world'? That must be the natural killer application for a secret clone army... All those clone projects were survivalist projects. They all failed, all of them. Because they lacked transparency."

Radical projects need widespread distributed oversight, with peer review and a loyal opposition to test them. They have to be open and testable. Otherwise, you've just got his desperate little closed bubble. And of course that tends to sour very fast.

Bruce Sterling in "The caryatides"

[-]ChristianKl4y20

Open hypnothesis:

I had my first vaccination with AstraZeneca after which I spent a day in bed. For me second vaccination I had Biontech where I had less energy the next day but not so strong effects as the first one.

It's quite odd that the side-effects of the vaccines are so different from person to person.

One thing that might explain how the side effects are so different is that maybe for some people the vaccine stays mostly at the point where it was injected and for others it travels more through the body which causes side effects.

I had those ideas befor... (read more)

2Viliam4y

I had two AZ shots, the first made me spend 12 hours in bed, the second I barely noticed, and as far as I know, in both cases I did the same thing afterwards (waiting for 15 minutes reading a book, then walking home for 30 minutes). So, who knows, maybe the effects of the second vaccines are just weaker in general.

2ChristianKl4y

On average second shots have higher side-effects because there's a larger immune response.

-15garbageactual4y

[-]ChristianKl4y20

Let me startoff that from my perspective there's a lot of uncertainty about the effectiveness of ivermectin.

At the moment it seems that it's plausible that ivermectin is for many viruses the equivalent of penecilin for bacteria.

Should this turn out to be true, it seems pretty clear evidence against the low-hanging-fruid thesis of why innvoation declined. If we weren't able to detect that an existing drug that we used 4 billion times is the equivalent to penecilin, our ability to pick the low hanging fruid is clearly very low.

2Viliam4y

What is your strongest evidence for ivermectin being useful against covid? I did not pay much attention to this, but asked my friends who did, and they said something like all studies in favor of ivermectin were seriously flawed. Things like "one group using nothing, the other group using ivermectin + some X, and getting better results", where we already have a reason to suspect that X alone does the whole effect. (Which to me sounds like exactly the kind of experiement one would set up if they already expected ivermectin to be useless, but wanted to prove that it was useful.) In other words, is there actually any reason to care about ivermectin other than the fact that someone else has already privileged this hypothesis?

2ChristianKl4y

To not rely to much on my own reading of the evidence, I opened a thread about discussing the quality of pro-and-contra meta-analyses on LessWrong: https://www.lesswrong.com/posts/EAnLQLZeCreiFBHN8/how-do-the-ivermectin-meta-reviews-come-to-so-different In the thread it seems consensus that the pro-Ivermectin meta-analyses is of higher quality. On the other hand the contra-Ivermectin studies is seen as borderline malicious (among others they switched intervention and control number for one study) Trusting the best meta-analyses on a topic is generally a good strategy and the one I'm using here. It's generally quite easy to dismiss evidence by saying "I have abstract standard XY for which I have no structured empiric evidence to justify it being an useful standard, the evidence you provide fails XY."

[-]ChristianKl4y20

Medical researchers:

We see that childhood cancers are associated with PGBD5 which causes a lot of mutations.

What do we do with that knowledge? How about blocking the DNA repair of the mutation that are caused by PGBD5 so that the mutations kill some cancer cells.

I would have guessed that preventing PGBD5 from creating the mutations would be a higher priority.

[-]ChristianKl4y20

I'm playing around with an evolutionary model for transposons and the transposons regularly kill my whole population...

2Pattern4y

Do the transposons ever have positive benefits? Why is your population all connected?

2ChristianKl4y

Transposons increase the mutation rate, so the fitness of organisms changes more when there are transposons in my model. When it comes to that I treat every transposons equally. Otherwise, each transposon has a rate for self replication. Aside of that transposons have no positive benefits but they self reproduce. If the mutation rate is what's useful then there should be pressure for transposons with low self replication rates which I don't see. Transposons work similar to the gene drive ideas for killing of malaria causing mosquitos. However the body does have some defenses. Both the transposons evolve and the defenses evolve and in nature there's an equilibrium. Finding the right parameters that lead to the equilibrium, might produce a model that does predict aging purely based on the fact that transposons exist. Having such a theory would back up https://www.lesswrong.com/posts/ui6mDLdqXkaXiDMJ5/core-pathways-of-aging . From what I read evolutionary models generally don't need group selection to work. If transposons kill every species if it wouldn't be for group selection that would be a major scientific finding. There's the belief that the minimum number of individuals for a specis is 500 over longer timeframes.

1dkirmani4y

This is why it would make sense if there was some (perhaps small) positive effect of transposons on an individual's fitness. (Also, aren't there any less costly ways to increase mutation rate? Maybe error-prone DNA polymerases, or just allocating less resources to DNA repair.)

2ChristianKl4y

It's not costly for a transposon to copy itself. If you start a gene drive to erradicate malaria, it's not in the interest of the individual mosquito to play along but it still happens. A positive selective effect of transposon on the level of the individual is not needed for transposons to have a reason to copy themselves. I'm doing computer modeling and the question of what stops transposons is a harder one then the other way around. Aside from that it seems that humans do use the fact that the have transporase for a few things (but currently not in my computer model). PGBD5 (a transporase) seems to be used in the brain to increase the diversity of brain cells for all vertebrates since 500 My. RAG1 and RAG2 are derived from a transporase and they are important in the immune system to get a diversity of different leukozytes. This means that neither of those can be completely downregulated and when they are active transposons can use them to get copied around. An interesting side-note is human have a lot of different kinds of cells where some cells get cancer much more frequently then others. Leukemia is a common cancer and might be downstream from RAG1 / RAG2. Braincancer might be downstream from PGBD5. Most childhood cancers are downstream from PGBD5 as well, so it's the most costly transporase for fitness.

[-]ChristianKl5y20

China clearly banning human genetic engineering is a interesting news item (I got it from Gwern)

[-]ChristianKl5y20

I feel pretty disturbed right now by https://www.winfried-stoecker.de/blog/die-beste-impfung-gegen-covid-19 . If what Stoecker (a biotech billionaire) is saying is true then for the companies that developed vaccines it was more important to deliever vaccines with fancy technology on which they hold patents instead of just doing the straight-forward well understood way of producing vaccines that we know and that could give us as many vaccines as we wanted.

[-]ChristianKl5y20

I have a new draft for a post on why it makes sense to use rationalist jargon like steelmanning over the existing jargon. This is an experiment whether shortform is a good place to ask for draft feedback:

https://docs.google.com/document/d/1slE6_sR82UsssV6eWRgjHNrHgNDsHPSltmbqIypsISU/edit?usp=sharing

2Trinley Goldenberg5y

What type of feedback are you looking for?

2ChristianKl5y

My goal is to bring the post into a better form before I publish it on LessWrong. Feedback that's helpful for that goal is welcome. When it comes to discussing the ideas of the post, that seems better once the post is finished, so that the discussion will stay on LessWrong.

1supposedlyfun5y

I endorse gjm's final comment 100%. (I wrote a much longer response but eventually decided that it was just repeating what ze said.)

[-]ChristianKl5y20

I have the impression that the rate of new posts on LessWrong dipped over the summer and is now picking up again. Did the warm weather reduce the amount of time people have to write posts and now we all spend more time inside, so we have time to write?

[-]ChristianKl5y20

Winter is coming and given the COVID-19 situation it means that if you want to meet other people savely doing it outside in the cold might be the best way to go about it.

For meetups I'm thinking to have a setup that switches between still group explanation and then pair exercises that can be done while walking around.

As far as clothing goes, I'm very unclear at the moment. What clothes are ideal for being able to be outside in the cold without freezing?

Are there any other concerns about how you might improve outdoor meetups when it's cold?

4stoat5y

For what's actually ideal, I would suggest (if you find it interesting) reading about technical clothing for mountaineering and winter camping and adapting that to city fashion -- but if you want some helpful more affordable tips, what works for me is many layers. For example, long sleeve undershirt and long underwear from Walmart. T-shirt over the undershirt, cheap sweatshirt hoodie over that. Thin pajama style pants over the long underwear. For me anyway, this can be completely comfortable under an outer layer of only jeans and just an autumn jacket up to maybe -15C or -20C. Thin cotton gloves under big mitts or heavier gloves. Thin socks under heavy socks. For items where it's more difficult to layer, such as a toque or scarf (or socks if shoes limit the room), wool is pretty good and is affordable, thick, and sturdy at Army Surplus stores, at least here in Canada. I would also suggest, don't neglect extremities or any body parts. For example, I remember once my thighs being very cold wearing only jeans (no long underwear), even though I had an extremely warm parka and was walking briskly. I've found personally that even fairly thin pajama type pants, if you have say two layers under jeans, can keep you pretty comfortable even up to maybe -30C. Since you have in mind a meetup, I feel that even for example students on a budget, could get extremely adequate winter clothing this way at a Walmart in Canada that would enable them to stand outside inactively for 3-4 hours at up to -25C, say.

2Dagon5y

You should probably give more specific descriptions of the expected conditions than "cold". Answers will differ if you mean high-desert conditions (-20C, windy but dry), cold-ish city conditions (just below 0C, frozen rain or snow), moderate (generally a bit above 0C, often rainy), or something else (Southern California gets down to 10C and cloudy). In a lot of places, you should be thinking semi-outside, rather than full open-air. Pavillions or large covered spaces such as outdoor seating for restaurants will keep the worst of the wind and precipitation out, and many of them have heating elements.

2ChristianKl5y

I don't care about high-desert conditions as Berlin won't hit them. I care about making meetups work in the other contexts and I'm happy about any suggestions for that even if the don't work for all cases.

4Dagon5y

Cool (but not cold ;) ). I've only been to Berlin in the summer, but from travel guides, it seems basically temperate and not overly-damp - there will be occasional snowfall, but no significant accumulation, and many days will have no cold rain to deal with. Standard advice for clothing in variable conditions applies: look to layers, so you can adjust over time, and add/remove pieces when you go indoors or if the sun is out and it's 5-10C warmer or cooler than you expected. A sweater/jumper over your shirt gives you flexibility here. I have a fleece undercoat with a waterproof shell that's great in really cold conditions, and fine as just a shell when it's warm (+5C) and rainy. For meetups, you'll still want to find covered areas - even if no rain is in the forecast, less moving air makes a noticeable difference in comfort. I suspect there's no way to make it pleasant and effective enough to be worthwhile before you can do it in an enclosed (but not crowded) space, but I really look forward to hearing how it goes.

[-]ChristianKl5y20

Schools around the world seem to start using automated grading for tests. If that technology exists, it would be interesting to have a forum that enforces posts to have a minimum score on those grading forms.

[-]ChristianKl6y20

If I put copper tape everywhere, do I still need to take copper supplements when I up my zinc intake?

3jimrandomh6y

Copper tape in your environment is unlikely to meaningfully affect your dietary copper intake.

[-]ChristianKl6y20

I just subscribed to Stiftung Warentest which is Germany's equivalent to Consumer Reports. It seems to me those institutions provide a vital service to allow producer in various categories to compete based on quality when they would otherwise compete based on marketing promises.

Intuitively, it feels easier to pay for physical goods then to pay for information like those reports. I think the information did allow me to buy better soap for washing my hands and I think there's a public good to be done by supporting those institutions and increasing their budgets to do unbiased tests (the Wirecutter is payed by affiliate money in a way that influences their editoral decisions).

[-]ChristianKl6mo10

Today, I had two chats with ChatGPT 4.5.ChatGPT 4.5 know from the memories I gave it that I'm writing a book. I already wrote a post about how LLM make doing theoretical work much easier.

In the first chat I asked "Is there any author that extends the A/B distinction to also cover X?" Then it give me some answer that one author had (AB(combined)/C/D) and some author had (A/B/Z). I asked where there's anyone who has more distinction than A/B/Z and it didn't give me any.

I opened a second chat and asked whether there's an author who does further distinct... (read more)

Moderation Log