@VoxelVexillologist's banner p


Multidimensional Radical Centrist

1 follower   follows 0 users  
joined 2022 September 04 18:24:54 UTC


User ID: 64


Multidimensional Radical Centrist

1 follower   follows 0 users   joined 2022 September 04 18:24:54 UTC


No bio...


User ID: 64

It's not bizarre at all if you remember that ChatGPT has no inner qualia. It does not have any sort of sentience or real thought. It writes what it writes in an attempt to predict what you would like to read.

I don't think I disagree here, but I don't have a good grasp of what would be necessary to demonstrate qualia. What is it? What is missing? It's something, but I can't quite define it.

If you asked me a decade ago I'd have called out the Turing Test. In hindsight, that isn't as binary as we might have hoped. In the words of a park ranger describing the development of bear-proof trash cans, "there is a substantial overlap between the smartest bears and the dumbest humans." It seems GPT has reached the point where, in some contexts, in limited durations, it can seem to pass the test.

One of my heuristics for good persuasive writing involves the number of citations, or at least clear distinct factual references that could be verified, as the clerk is doing for the rest of us here. Broad, general arguments are easy to write, but in my opinion shouldn't be weighted as heavily.

The amusing part here is that I have been doing this for years to weed out political hacks, long predating GPT.

Interesting! Any good papers or summary articles you'd recommend?

I know this is a hobby horse, but once AI is trained on gait recognition and body language of labelled examples of millions of hours of countless criminals’ movements recorded by CCTV, tiny little telltale patterns might well allow for effective pre-crime in the case of almost all premeditated criminal activity. People show their nerves, everyone has a tell, etc.

I have trouble believing there's enough information content present in CCTV streams to uniquely identify individuals confidently. I see how it maybe could work, but it's not something I'd focus on directly. Are human gaits really that different as to be identifiable from distant security cameras? Are they even consistent for a single person day-to-day?

The longer I think about it, I've also started thinking that AI likely scales sub-linearly (logarithmic?) with the size of the training dataset. "But the AI can viably consider a larger dataset than human experts" may be true, but may not generate hugely better results.

This is the equivalent of a Japanese Banzai charge straight into dug-in machine gun emplacements and sighted artillery.

I've long thought that if the Catholics really wanted to win a battle in the Culture War, they should start repeating "anti-Catholic animus" (or perhaps some catchier -phobia or -ism term I'm not going to consider) in the same way that "racism" and "antisemitism" get thrown around. The historical citations aren't really unjustified: the KKK was founded as, among other things, anti-Catholic. All of the historical bias against Italians and Irish immigrants is at least somewhat rooted in anti-Catholic bias, as is some of the bias against Central and South American immigration. The Nazis persecuted Catholics. And they continue to be victims of hate crimes in the US.

On one hand, repetition legitimizes and a constant drone of "we're persecuted" is functionally how various groups on the left have achieved their existing hierarchy -- this seems to bear more relation to the quantity and quality of complaints than to any particular metrics of measurable oppression. On the other, I respect that Catholics absolutely could claim (some degree of) martyrdom in the Year of Our Lord 2023 but choose not to because silent stoicism better aligns with their principles.

And yet when asked to apply the same logic to that same sort of men in America, an impenetrable mental block descends and makes it impossible for even the same commentators to reach the same conclusion.

Part of the issue here is that America defines nationality not along ethnic lines, but by citizenship (so does France, although this appears less in Anglophone discussions). Swede refers both to citizenship and generally to a specific subset of Scandinavian heritage. I'd say these map generally to language groups, but that's not fully the case (there are Swedish-speaking Finns): identity is a complex and often locally-defined concept, and Americans have generally embraced the "Melting Pot" outlook that "American" is not an ethnic group.

The problem is that "Those men are not [Americans]. They will never be [Americans]" is not obviously a true statement: one can absolutely (with appropriate vetting, which is hopefully not open to gang members shooting each other in malls) choose to become an American. One can become a citizen of Sweden, but our linguistic blurring of Swede referring to both heritage and citizenship makes "becoming a Swede" less obviously correct.

The NYPD claims that its rollout of body cameras to "all Police Officers, Detectives, Sergeants and Lieutenants regularly assigned to perform patrol duties throughout the city" was completed in August 2019, for a total of around 24,000 cameras. There are approximately 6400 subway cars in NYC.

If we can reasonably expect every police officer to carry a body camera on duty in 2023, we could absolutely put a camera on every subway car. A little searching suggests that every car in the Tokyo subway has had cameras since at least 2020. Like the body cameras, at some point it becomes an obvious choice to not provide surveillance, although it's not completely obvious to me that 2023 is that point.

What are the specific statements she made you think constitute defamation?

In a '98 NBC interview she called the sexual assault allegations against her husband a "vast right-wing conspiracy".

During her NBC interview, when Lauer said, "So when people say there's a lot of smoke here, your message is, where there's smoke..." Clinton interrupted.

"There isn't any fire," she said. After the allegations and motivations are dissected and the truth emerges, she predicted, "some folks are going to have a lot to answer for."

This was in an NBC interview, not a courthouse (like Trump's comments), and appears to directly suggest that all of the women are liars, which appears at least modestly similar to Trump's supposedly-defamatory remarks.

Ultimately, I think I have to conclude that denying allegations should probably enjoy specific privilege from defamation concerns.

I do think there are many Republicans who would not like to see mass deportations, but I think 80%+ of Republicans would like to see a secure border.

The (admittedly moderate) Republicans I know generally would agree with this: the sentiment is generally in favor of enforcing the laws on the books, with the caveat that changing the laws to reflect reality is acceptable -- for example, by massively expanding work visas to provide a legal basis for those who currently cross as illegal immigrants to find work, which would require facing the thorny issues of what conditions would be applied for those visas. But there's a concern that the generally uncontrolled state of the border allows all sorts to cross: not just cartels, and not even just Central/South Americans. Take a look at the 2020 statistics (last page): there are literally hundreds of citizens of places as far away as China, Ghana, and Bangladesh, and Romania apprehended by Border Patrol crossing the Southern border (many of these may present valid asylum claims under existing law, but that's not clear from the data and is honestly a pretty poor route to encourage even if true). Most developed nations aren't opposed to or incapable of tracking names and dates of those that enter the country.

The equivalents in 1975 were saying that the Cold War would inevitably end in nuclear annihilation. This was a terminally unhelpful position

IMO this is a fair comparison, although the Cold War MAD scenarios were explicitly designed to cause annihilation. The Bulletin of the Atomic Scientists, probably the premier Cold War doomerism group, is practically a laughing stock these days because they kept shouting impending doom even during the relatively peaceful era of 1998-2014, finding reasons (often climate change, which is IMO not likely apocalyptic and is outside their nominal purview) to move the clock towards doom. Do you think they honestly believe that we're closer to doomsday than at any point since 1947? We supposedly met that mark again in 2018 and then moved closer in 2020 and again in 2023.

There are all sorts of self-serving incentives for groups concerned with the apocalypse to exaggerate their concerns: it certainly keeps them in the news and relevant and drives fundraising to pay their salaries. But it also leads to dishonest metrics and eventually becomes hard to take seriously. Honestly, the continued failure of AI doomerists to describe reasonable concerns and acknowledge the actual probabilities at play has made me stop taking them seriously as of late: the fundamental argument is basically Pascal's wager, which is already heavily tinged with the idea of unverifiable religious belief, so I think actually selling it requires a specific analysis of the potential concerns rather than broad strokes analysis. Otherwise we might as well allow religious radicals to demand swordpoint conversions under the guise of preventing God The One Who Operates The Simulator from turning the universe off.

As a counterexample, I think the scientists arguing for funding for near-Earth asteroid surveys and funding asteroid impactor experiments are quite reasonable in their proclamations of concern for existential risk to the species: there's a foreseeable risk, but we can look for specific possible collisions and perform small-scale experiments on actually doing something. The folks working on preventing pandemics aren't quite as well positioned but have at least described a reasonable set of concerns to look into: why can't the AI-risk folks do this?

I am pretty firmly of the opinion that any post-scarcity society we can build looks less like UBI socialism and more like massive deflation in the pricing of essentials driven by automation and capital investments.

There are plenty of commodities that used to be expensive and are now basically free. Salt and pepper used to be valuable commodities, but now are tossed carelessly in paper packets into food packaging and largely discarded. Within the last century we've driven food prices down to where basically nobody in the West dies of starvation because they can't afford food: there are plenty of charities that together manage to make sure everyone is fed, although I'll concede the nutrition is often lacking.

But I also think the hedonic treadmill is a powerful thing and we can relatively easily convince ourselves that things haven't objectively changed: to me post-scarcity seems doomed to always look like the distant future, but is actually an asymptote we can steadily approach.

It was rescinded by Betsy DeVos in 2017.

IIRC the controversial "refereeing of adolescent relationships" portion was driven by a requirement to review sexual harassment and sexual assault allegations with a preponderance of evidence (civil) standard, which put school administrators in a position of having to establish parallel judicial systems because an act that didn't meet "beyond a reasonable doubt" in criminal court can absolutely meet a preponderance of evidence.

IMO punishing students with things like expulsion needs higher than a 51% standard of evidence.

Because Title IX has been interpreted to require universities to referee adolescent relationships! Title IX created the problem (via campus administration), and Title IX "fixed" the problem (via the judiciary)

Worse: the requirement was codified by a now-rescinded Dear Colleague letter from the Obama administration requiring schools adopt these policies.

At least in rural areas it's pretty frequently legal to engage in target shooting on private property: as far as I know in Texas it's legal to discharge firearms on your own property outside of city limits provided you're at least 300 feet from neighboring occupied buildings. Within city limits it's generally a local law issue. Rifle and shotgun shots (presumably mostly for sport or hunting) are not an uncommon sound if you start wandering backroads.

Which would make this another example of "just enforce the law you losers" cases.

As far as I can tell, the suspect in question wasn't in the US legally, and thus couldn't have legally acquired the firearm in question.

I imagine these kinds of things still happen in the US but aren't federally funded. And US university people know anything about this?

The typical workaround is that you can host a "women in [field]" event, but you can't restrict who actually attends. To some extent everyone knows what's expected, but I do recall my local Society of Women Engineers chapter was pretty explicit about recruiting all comers, so it's not all a wink and a nudge.

It is no exaggeration to say that most of STEM innovation in US academia is now being carried out by foreign-born people.

This may be true overall, but I think it's somewhat exaggerated because what we discuss as "STEM innovation" is colored by high-profile tech companies that generally broadcast their research far and wide. If you're talking OpenAI and Google, sure. But there are significant classes of employers that are restricted to either citizens or at least permanent residents (green cards). SpaceX isn't hiring Chinese nationals. The largest employer of mathematicians in the US only hires citizens and generally holds its research very close to the chest, as do the national labs.

I think an unbiased sample of "STEM innovations" is harder to measure in real-time than it sounds. But also that concern about US research output is not necessarily misplaced.

Keep in mind: there is absolutely no sex/gender distinction in our local language.

There is barely one in English either, to be honest. It seems to have been shoehorned retroactively because the sex descriptors are adjectives -- "female" as a noun is, er, quite objectifying as used, and I can see why it upsets some feminists -- and the gender descriptors are nouns: "woman [career]" is awkward too.

those corporate HR policies he’s setting aren’t getting implemented in places like Texas and that middle management is simply lying because they don’t want to look bad.

I think this is really a function of workplace culture and class rather than geography. Most of your college-educated, white-collar workforce is cognizant enough to recognize that "no homo" in the office probably won't fly. But I doubt that the truck drivers, technicians, assembly line workers, and even janitorial staff are really watched by the liberal panopticon so closely. For many working-class gigs like restaurants, the working language (Spanish, most frequently) isn't necessarily understood uniformly by the educated left anyway.

But I could just as easily write this about dropping "fuck" in every other sentence, which is also a distinctive class marker.

Does reddit's userbase make it significantly less valuable (and possibly even worthless)?

Broadly, there appears to be a decent market segment for a pseudonymous broad-market shitposting service. Unfortunately, it's not terribly lucrative: Reddit isn't public, and although rumors of IPOs circulate occasionally, it doesn't seem hugely profitable for the amount of content it hosts. This seems pretty in-line with similar competitors: Usenet was never really commercialized, Digg engaged in some heroic self-immolation, and 4Chan and Twitter have never been hugely profitable. The traditional strategy for making money on such a platform heavily relies on advertising, but this seems to falter in practice because (1) pseudonymous users can't easily be served targeted ads and (2) many advertisers don't want to appear next to random shitposts. Charging users for access doesn't work for broad market appeal because it breaks pseudonymity and places a high barrier to entry for new users.

Of the other big social media platforms, Facebook escheded psuedonyms early on and was able to break into targeted advertising (and benefited from early self-censoring of controversial content), and YouTube and TikTok manage to exercise pretty strong editorial control of the content firehose.

Reddit's problem is that they entered a market segment that seems doomed to be minimally-profitable and can't really afford large investments in improving the quality of posts on the platforms.

There is a sincere group of people that believe data should be free and shared are widely as possible Almost as a terminal goal in and of itself. It may seem ridiculous to us who are steeped in capitalism, but this mindset exists.

I don't know that they're wholly incompatible: I enjoy capitalism because it is observably a local maxima in improving quality of life over most command economy alternatives. But in practice this usually means efficient capital investments producing supply-side gains by reducing prices: I can afford a nicer car than my parents had growing up (largely because even entry-level modern cars have better features than a 1993 Honda Civic), better computers (my phone can run circles around a 386DX), and even better media (even terrible streaming platforms are better than 4 fuzzy over-the-air channels).

It's not an immediate goal, but I personally think that anything resembling a post-scarcity society looks on a long time scale like capital investments in scale and efficiency driving prices towards zero. For media in general, the marginal cost of an extra view is effectively zero already, but I look askance at publishers who push mass-market media and aggressively target high profit margins. While Once Upon a Time in Shaolin is interesting as a piece of art, I find its business model pretty repugnant.

I don’t know what would stop the FBI.

Directly lying to Congress under oath never seemed to hurt the career of James Clapper, who is still employed by CNN as a security analyst. He later apologized after the Snowden leaks, claiming that he "simply didn't think" of the Patriot Act, but was never punished.

DEI does not attempt to answer any of these comprehensive questions.

I think proponents of this law would disagree, and point at the fairly clear examples of "good and evil" and "duties of conscience" appearing in the relevant canon. Although you have a point that DEI lacks a unified view of meanings for life and death.

Is it clear that "TPTB" are actually directing the scrubbing? It seems possible to me that many of the intermediaries like Discord and even Telegram have reasons to avoid hosting actually-classified content of their own volition, from "remain in the good graces of the DOJ" to "we're patriotic and Support Our Troops". Those posting originally also have reason to clean up what they can, although I doubt they can actually hide their actions from the DOJ.

AFAIK, the legal penalties for classified documents are almost exclusively due to violating the SP-312 NDA the government requires for access. I don't know what teeth, if any, are actually relevant to non-signatories (The New York Times writers, for example) in peacetime.

In keeping with that, I will smugly note that I don’t drink that shit anyway and I’ll be cracking an IPA from a real industry underdog - Lagunitas(tm), a tiny subsidiary of a little-known international parent company.

I know at least a few people who actively avoid beers produced by AB InBev and Heineken (owns Lagunitas) and all their subsidiaries. This is harder than it sounds, because it's not always obvious on the packaging. I'm not going to say that I do so exclusively myself, but I do prefer locally-owned brands where possible.