GeneralMayhem

I'm pretty sure there are exactly two reasons for Trump going on about Greenland:

1. Putin suggested it to him at some point, because he knew it would cause chaos among the Western alliance bloc.

2. Mercator projection. It looks big on a map, therefore he wants it.

The second one is along the same lines as his obvious confusion about the word "asylum" - if you watch how he talks about refugees, he obviously doesn't know the difference between "mental asylum" and "political asylum", and thinks that people seeking refuge in the United States are all nuts. The man is really, really, really evil, but he's also really, really, really dumb.

or

    git stash push <filename>

PDFs are the opposite of machine-readable if you want to do anything other than render them as images on paper or a screen. They're only slightly more machine-readable than binary executables.

I hate, hate, hate, hate, hate the practice of using PDFs as a system of record. They are intended to be a print format for ensuring consistent typesetting and formatting. For that, I have no quarrel. But so much of the world economy is based on taking text, docx (XML), spreadsheets, or even CSV files, rendering them out as PDFs, and then emailing them around or storing them in databases. They've gone from being simply a view layer to infecting the model layer.

PDFs are a step better than passing around screenshots of text as images - when they don't literally consist of a single image, that is. But even for reasonably-well-behaved, mostly-text PDFs, finding things like "headers" and "sections" in the average case is dependent on a huge pile of heuristics about spacing and font size conventions. None of that semantic structure exists, it's just individual characters with X-Y coordinates. (My favorite thing to do with people starting to work with PDFs is to tell them that the files don't usually contain any whitespace characters, and then watch the horror slowly dawn as they contemplate the implications.) (And yes, I know that PDF/A theoretically exists, but it's not reliably used, and certainly won't exist on any file produced more than a couple years ago.)

Now, with multi-modal LLMs and OCR reaching near-human levels, we can finally... attempt to infer structured data back out from them. So many megawatt-hours wasted in undoing what was just done. Structure to unstructure to structure again. Why, why, why.

As for universality... I mean, sure, they're better than some proprietary format that can only be decrypted or parsed by one old rickety piece of software that has to run in Win95 compatibility mode. But they're not better than JSON or XML if the source of truth is structured, and they're not better than Markdown or - again - XML if the source is mostly text. And there are always warts that aren't fully supported depending on your viewer.

If your problem is humans making arbitrary and nonsensical decisions about how you can do your job, then you have a non-technical problem, and it's unlikely that any technical solution will solve it.

I find that the best way to set up test cases, regardless of language, is usually to use string constants in the proto text format (https://protobuf.dev/reference/protobuf/textformat-spec/). For arrays, and especially for oneofs, it's way less verbose than how things are represented in Go, C++, or Java, and generally at least on par with the Python constructors. Maps are the only thing that suffer a bit, because they're represented as a list of key-val pairs (like in the wire format) instead of an actual map. Your language's compiler won't help you debug, but the parse-text-proto function can point to the source of the issue on a line/character level.

With Go generics - and equivalent in most other languages - you can write a 5-line helper function that takes a string in that format and either returns a valid proto value (using the generic type param to decide which type to unmarshal) or `t.Fatal()`s. You would never do this in production code, but as a way to represent hand-written proto values it's pretty hard to beat.

The real issue is that the sand isn't orderly sorted. At a micro level, it's billions and trillions of individual interactions between atoms that create the emergent behavior of solid grains of sand packing reasonably tightly but not phasing through each other.

Democrats have not had a filibuster-proof majority in the Senate since before Reagan, and they're constantly up against the asymmetry that breaking things is a lot easier than building them. The side that wants government not to work has a huge structural advantage in any argument (even apart from their structural electoral advantages thanks to letting empty fields carry as much weight as cities) since they "win" by default.

It only takes a few years or even months of Republican rule to install judges that will deem any regulations unconstitutional, to allow time-bounded authorizations to lapse, to fire large numbers of public servants who keep the administrative state running, or simply to use the bully pulpit to tarnish the reputation of federal bureaucracy so that individuals will refuse to interact with them. Building those institutions as credible, competent forces takes somewhere between years and lifetimes.

> He claims that an uptick in autism incidence is correlated with a lot of environmental things, one of them being that the number of childhood vaccines has increased

That is a factual lie. That is not what he said.

In 2005, he wrote an article (https://www.govinfo.gov/content/pkg/CRECB-2005-pt10/html/CRE...) that was published in Salon - since retracted by the publication, but never disavowed by him - in which he said, among other things:

> It was only after reading the Simpsonwood transcripts... that I became convinced that the link between thimerosal and the epidemic of childhood neurological disorders is real.

The idea that thimerosal in vaccinations causes autism, or any other neurological problem, has been thoroughly debunked by every serious researcher who's ever looked at it. I'll just link to Wikipedia's page on the topic (https://en.wikipedia.org/wiki/Thiomersal_and_vaccines) since that's easier than collecting the same list of sources again, but you can ask the CDC, the WHO, the NIH, etc., etc.

> There is plenty of academic literature around [ivermectin and hydroxychloroquine]

Yes, and all of it points to them having no effect on COVID. Sources: Mayo Clinic (https://www.mayoclinic.org/diseases-conditions/coronavirus/i...), NIH (https://www.ncbi.nlm.nih.gov/search/research-news/12053/), WHO (https://www.who.int/news-room/questions-and-answers/item/cor...).

> I'd imagine the same thing is true about seed oils

So you know nothing, much like RFK. You're right, the situation is the same: there is no actual evidence, and this is a nonsense meme that's been spread on social media by non-experts. There are fewer academic studies, because the panic about seed oils is newer (and therefore there hasn't been time for academia to exhaustively disprove it), but that's not how this works: if you're making the extreme, uncommon claim, the burden of proof is on you to provide evidence, not on the rest of the community to disprove it. Sources: American Heart Association, citing a professor of medicine at Stanford (https://www.heart.org/en/news/2024/08/20/theres-no-reason-to...).

> every talking head that mentions RFK makes it look like he's just some weirdo that thought up all of these kooky ideas

This is also a lie. This is the fourth or fifth time you've said this in this thread, and it's not true. Nobody, neither on this thread nor in the wider media landscape, thinks he invented them himself. Everyone knows he's parroting random memes or conspiracy theories he found online. That is not an excuse.

> I (and respectfully, probably most of the commentariat) aren't actually qualified to tell you either way.

You're right! That's why I trust the actual medical experts, who categorically disagree with RFK and his ilk, and tend to describe them as "quacks", "cranks", and "misinformed". If you are as unqualified to judge the evidence as you say you are, then the only even remotely logical response is to rely on experts to interpret it for you. Why are you trusting some random person with a a famous last name over every single medical researcher? Do you apply this logic to any other topic - "I'm not an expert, therefore, instead of listening to the experts, I'll listen to the first person I see talk about it on TV"?

-----

There is a reason why people abbreviate this whole conversation to "brain worm" and move on. Nothing RFK says about health is founded in reality. It is all crank nonsense, and it takes a special kind of gullibility or motivated reasoning to believe it when the actual sources are right there, just a Google search away. He's the equivalent of the schizophrenic who stands on the corner screaming about UFOs and bodysnatchers, just with slightly more decorum. In order to even believe that he's acting in good faith, let alone that he's right, you have to be willfully blind to evidence that is right in front of you. You have to choose to believe one weirdo over the entirety of modern science. That is simply not how intelligent people behave.

Bullshit.

Show me one example where he's said something like that and he was citing a real source. And I mean an actual source that he should be respecting. If he just parrots this stuff from other cranks on Twitter, that's not exactly the defense you think it is.

I find it very hard to take this kind of comment in good faith. You know as well as I do what RFK says, and why saying that it won't instantly kill you, while obviously true, only benefits misinformation.

But ok, fine. Raw milk is not better for you than pasteurized. It does not have magic healing properties. And it has a much higher risk of introducing pathogens into an industrialized food supply.

But see, this is the problem. Raw = natural = good takes only a word or two. And if challenged, the crank can fall back on the motte-and-bailey argument of "what do you mean, of course milk is good for you!" But if you're not aligned with the lunatic fringe, any incomplete or even fractionally incorrect answer is a gotcha. Debunking nonsense is exponentially harder than spewing it to begin with, and then concern-trolling about "oh, only if it's contaminated!" makes the situation worse, not better.

So your answer to multiple specific examples is "nah, the other guys probably do it too, trust me bro"? Yes, Trump went beyond all recent presidents, in pretty much every way.

Sabotaging the FBI's background check is absolutely without comparison. It was corrupt and inexcusable.

While yes, all presidents will tend to appoint justices they agree with, you cannot in good faith say that there is any comparison between Jackson and Sotomayor on the one hand, and Kavanaugh and Barrett on the other, in terms of qualifications to sit on the bench. And that's just at the Supreme Court level - the whole affair in Florida with Aileen Cannon is another level of obscene.

The "official acts" decision is completely without legal historical merit, and was made up out of whole cloth to allow Trump's appointees to protect him from any consequences from the judicial branch (remember that whole idea of three co-equal branches of government?). No other president has dared make so bold a claim, both because the idea that the Court should be subservient to the president is clearly at odds with how the American government has worked for almost its entire history, and because they didn't have personal crimes to cover up.

I do have a vendetta against Trump, but you have the cause and effect backwards. I don't think he's a bad president because I hate him, I hate him because I think he's a bad president who is dangerous to me personally, to the United States as an institution, and to the continuation of the human race. But perhaps even more than that, I hate him because he has enabled and legitimized pathetically transparent bad-faith arguments like this.

Not for the justices, but for the institution of the Supreme Court. He files frivolous lawsuits and appeals designed to give his appointees the opportunity to legislate from the bench (e.g., "official acts"). He appoints blatantly partisan judges at all levels of the federal judiciary. And he sabotaged the FBI's inquiry into Kavanaugh's history, which is standard for any appointee at that level, by having any concerns be routed to the White House instead of handled by actual investigators. In short, he's demolished any pretense that the Court exists to enforce laws fairly, and has turned it into an unapologetic arm of the MAGA Republican party.

> if the norms mean giving all of our tax dollars to NATO for nothing in return

We do not give our tax dollars to NATO, at least not in any meaningful way. NATO's entire budget as an organization is about $4B/year, which includes valuable shared command/control systems. For the most part, we fund the American military, and we commit to using it in concert with our allies in certain scenarios.

In exchange, we get incredibly valuable hard and soft power. We get access to land in Europe to use as bases, which are staging areas for potential worldwide threats (e.g., an imperialist Russia). We get shared intelligence. We get goodwill with the rest of the West, so that they'll join our trade pacts. We get commitments of Polish tanks and British spies and French manpower if there ever to be a hot war, so that the US can focus on what it does best (air and naval superiority).

But also, you're the only one who brought up NATO. There are myriad unrelated norms that Trump broke the first time around, and will certainly break further this time, that make the institution of "the American government" less able to serve its purpose. Norms like, a president can't pardon himself. A president can't use his position to direct foreign powers to patronize his own businesses. A president can't summon a violent mob to Washington to overturn an election. A president can't conspire with state legislatures and militias to disregard the results of their states' elections. A non-sitting president can't steal classified documents, and can't have ongoing secret communications with a foreign power. A president keeps special counsel at arm's reach. A president shouldn't use tax policy to explicitly punish states that don't vote for him.

Personalized != contextualized. You could have a search engine that uses geolocation without building any sort of cross-request profile on the individual making the search.

Google does not check in protoc-generated code. It's all generated on demand by Blaze/Bazel.

> same-sex marriage prop (which I believe is purely symbolic...

Same-sex marriage is currently outlawed by the California constitution as a result of Prop 8 from 2008. That clause is void as a result of the Hollingsworth and Obergefell decisions, but there are multiple members of the Supreme Court who have explicitly said that they would like to overturn Obergefell, so it's a good idea to get ahead of the potential catastrophe by taking the bad law fully off the books, rather than relying on a capricious and extremist court to stick to a rights-defending decision for any amount of time.

Event correlations are a typical one. Think about ad tech: you want every click event to be hydrated with information about the impression or query that led to it. Both of those are high-volume log streams.

You want to end up with the results of:

``` select * from clicks left join impressions on (clicks.impression_id=impressions.id) ```

but you want to see incremental results - for instance, because you want to feed the joined rows into a streaming aggregator to keep counts as up to date as possible.

That's true in the general case, but there are usually ways to make it not matter.

If you have a vector clock with snapshot reads, then you may not know consistently what the system looks like now, but you can know what it looked like as of some arbitrary previous point in time. If your use case doesn't require read-modify-writes (e.g., metric calculations where the source of truth is out of your control and you just need to aggregate deltas), then that's good enough - you can find a set of results that are in sync with each other, even if they're not fully up to date.

If you have limited read-modify-writes, but you can at least engineer synchronous transactions on subsets of the data (e.g., single-row-level transactions) with asynchronous replication and fan-out of side-effects, then you can turn more complex problems into something that looks like the first case.

Those tricks alone are enough to remove most of the weirdness with eventual-consistency, but the broader point is that distributed system design is more about learning a toolbox of tricks to handle and describe certain patterns, of which you may have any number in any combination in any given project, than it is about "solving" the problem once and for all.

Sure, but now you have the string constant as a symbol, which you can either grep for (in which case you're delayed by one search, not the end of the world if you were going to unwind callstacks anyway) or, if you have an LSP, you can jump directly from it to users...

> I don't have the resources on my laptop to compile a kernel

That level of "argument" from a federal judge should be grounds for instant removal. It's bafflingly stupid, and nakedly partisan and outcome-motivated. Some ideas are just categorically bad, and the regulator needs to be able to say that, rather than pretend that the answer is always "halfway in between".

For his next decision, will he forbid the SEC from regulating all insider trading, since they didn't find a way to target "specific, harmful" patterns? And yet somehow I suspect he wouldn't bat an eye if the FDA were to decide that all abortifacients were disallowed, instead of looking at each one individually...

And of course, all of that is ignoring that he's flagrantly lying about the facts of the case, in at least two ways:

1. The summary is misstated. The FTC did not find that all non-competes were invalid. It specifically carved out cases where the balance leans toward non-regulation - namely, non-competes for executives.

2. Some of the "arbitrary and capricious" reasoning claims that no states have passed non-compete bans as broad as what the FTC did, and therefore none of their quantitative evidence can matter. That is, again, blindingly stupid on the face of it - but it's also factually false, because California has had just as broad a ban on non-competes for many decades.

No, it's a triple-negative. Judge struck down (1) ban (2) on non-(3) competes.

As a result of this ruling, non-competes are now enforceable; workers are not allowed to compete with ex-employers.

But the Eiffel Tower is an art project, not something of actual utility...

There's a branch for "Does the HRE exist" (in the top left corner), which basically says that "territorial maps" aren't well defined before the mid-modern period.

What makes this tough is that LLMs can show logical thinking and self-correction when specifically prompted (e.g. "think step by step", "double-check and then correct your work"). It seems unlikely that they can truthfully self-reflect, but I don't think it's strictly impossible.

Yep, this is a big part of it. Intelligence and consciousness are barely understood beyond "I'll know it when I see it", which doesn't work for things you can't see - and in the case of consciousness, most definitions are explicitly based on concepts that are not only invisible but ineffable. And then we have no solid idea whether these things we can't really define, detect, or explain are intrinsically linked to each other or have a causal relationship in either direction. Almost any definition you pick is going to lead to some unsatisfying conclusions vis a vis non-human animals or "obviously not intelligent" forms of machine learning.

It's a real mess.

0x9c (156 dec) is still a very small number, all things considered. To me that sounds like attempting to access an offset from null - for instance, using a null pointer to a struct type, and trying to access one of its member fields.

The argument would be that that conceptual model is encoded in the intermediate-layer parameters of the model, in a different but analogous way to how it's encoded in the graph and chemical structure of your neurons.

I am highly skeptical of LLMs as a mechanism to achieve AGI, but I also find this paper fairly unconvincing, bordering on tautological. I feel similarly about this as to what I've read of Chalmers - I agree with pretty much all of the conclusions, but I don't feel like the text would convince me of those conclusions if I disagreed; it's more like it's showing me ways of explaining or illustrating what I already believed.

On embodiment - yes, LLMs do not have corporeal experience. But it's not obvious that this means that they cannot, a priori, have an "internal" concept of reality, or that it's impossible to gain such an understanding from text. The argument feels circular: LLMs are similar to a fake "video game" world because they aren't real people - therefore, it's wrong to think that they could be real people? And the other half of the argument is that because LLMs can only see text, they're missing out on the wider world of non-textual communication; but then, does that mean that human writing is not "real" language? This argument feels especially weak in the face of multi-modal models that are in fact able to "see" and "hear".

The other flavor of argument here is that LLM behavior is empirically non-human - e.g., the argument about not asking for clarification. But that only means that they aren't currently matching humans, not that they couldn't.

Basically all of these arguments feel like they fall down to the strongest counterargument I see proposed by LLM-believers, which is that sufficiently advanced mimicry is not only indistinguishable from the real thing, but at the limit in fact is the real thing. If we say that it's impossible to have true language skills without implicitly having a representation of self and environment, and then we see an entity with what appears to be true language skills, we should conclude that that entity must contain within it a representation of self and environment. That argument doesn't rely on any assumptions about the mechanism of representation other than a reliance on physicalism. Looking at it from the other direction, if you assume that all that it means to "be human" is encapsulated in the entropy of a human body, then that concept is necessarily describable with finite entropy. Therefore, by extension, there must be some number of parameters and some model architecture that completely encode that entropy. Questions like whether LLMs are the perfect architecture or whether the number of parameters required is a number that can be practically stored on human-manufacturable media are engineering questions, not philosophical ones: finite problems admit finite solutions, full stop.

Again, that conclusion feels wrong to me... but if I'm being honest with myself, I can't point to why, other than to point at some form of dualism or spirituality as the escape hatch.

There's a big problem of information asymmetry, though. "Security" is a very noisy metric, and not a quality that's easily evaluated by 99.9% of your customers. Outside of specialty domains like national defense (and even then...) customers don't generally care about "secure", they care about "not obviously insecure", and almost all products will appear to be "secure" enough right up until the moment that they're visibly or publicly compromised. If you either stay under the radar or get lucky long enough, you can turn profits for quite a while before the chickens come home to roost.

Technically, SCOTUS, as the highest court, isn't beholden to anyone's precedent. This year's SCOTUS is just as legitimate as last year's, so they always have the power to overturn past decisions. Lower courts have to follow what higher courts decide, but SCOTUS has no higher authority. And sometimes, like in Brown vs. Board of Ed (which ended legal segregation), it's a very good thing for the Court to overturn its past decisions.

But in practice, to non-lunatics, stare decisis (the legal principle that says not to overturn, or even consider, topics that have already been decided in the past without an extremely good reason) is an incredibly important prior to bring into any discussion. If the court actually uses its power to completely rewrite the rules of how government works on a whim - and let's be clear, that's what this decision does - then there's no way for anyone to ever make a plan. Nothing is stable.

Unfortunately, at least 5/9 of the current Supreme Court are either lunatics or blatantly corrupt. Chevron was decided unanimously for a reason. There is no way to administer a modern state without that concept - which is why right-wing extremists are so happy to see it gone, because they don't want the state administered.