Locklin on science

Ancient Apocalypse reviewed

Posted in Book reviews, Open problems by Scott Locklin on January 5, 2023

Apparently this is the most dangerous thing on netflix. I of course don’t watch netflix, but I did watch this. I am familiar with Hancock from watching Joe Rogan shows while lifting weights. I also read one of his books (I think the Atlantis one) as a sort of amusing piece of science fiction with some bits of facts thrown in. That said, guys like Graham Hancock and John Anthony West (who inspired many of his ideas) often turn out to be right about certain things. Gubekli Tepe is a lot older than the mainstream archeologists said for years; just like these guys and other younger dryas types said. Ayahuasca wake and bake Graham Hancock was right, and normie archaeologists were wrong. They might also be right about stuff like the sphinx water erosion hypothesis; I have no real way of adjudicating this, other than noting that people have been saying it for a while and it might very well be true.

 

Archaeology is a wonderful subject, but it has an ideology, and it is presently mostly the kind of thing made to appeal to “Head Girl” IFLS types who memorized all the “correct” answers. Those sorts of people are always a squalid clerisy, and they almost never figure anything new out. It wasn’t always true; it used to be a field for adventurers, treasure hunters, freebooters, nationalists and religious believers. It used, for example, to be orthodoxy that the Great Flood was a real thing. Every Mediterranean culture from the Hebrews to the Greeks had the exact same legend with an Ark and the animals and everything. I found it flabbergasting being reminded of this by reading Lucian a few months ago.  Modern archaeologists haven’t read the ancient Greeks and probably think that people who read the bible are primitive, barely sentient barbarians. Yet the flood story is not only in Ancient Greek stories, Hebrew stories, Sumerian stories, it is rather annoyingly persistent in Hindu mythology, Zoroastrian mythology,  there are even American Indian flood myths. I assume archaeologists wave their hands and say fluffy bullshit about early farming communities being close to rivers which flood blah blah “I fucking love science” blah. But at one point in time people thought it was very reasonable to take old legends at face value and go looking for the stuff people wrote about. That’s certainly an ideological choice: we must recognize that we’re making one today by not looking for stuff the ancients wrote about. It is at least interesting this cultural commonality; a reasonable man could assume at least for the Mediterranean cultures perhaps it came from a knowable event.

In this series of videos, generally speaking, Mr. Hancock is either hamming it up for the rubes, or is an extremely credulous person. Hancock also doesn’t seem to think much of the ancients; his intuition is that anything old and impressive looking must have come from some other more advanced civilization. My own view is that people are pretty smart, and in the absence of distractions like ipotatoes and television, they get up to all kinds of interesting engineering shennanegins. I don’t think it is at all surprising people build things which have precise astronomical alignments (and in fact the birthday paradox makes them inevitable even when they’re not doing them on purpose), nor do I think it is weird that people cut and moved around giant stones before the bronze age. I could do it if I was really bored or had a large supply of slaves or enthusiastic followers. In fact when I was a teenager smoking weed and getting drunk in the woods, me and my friends did minor league megalithic architecture. In presence of an actual ideology or religion (or even sober engineers, riggers or a vague sense of purpose) we would have done a hell of a lot more.

That said it is obvious there are lost civilizations. It is obvious people can forget stuff that happened before: we know that humans have forgotten fairly important things that happened before. The Romans were arguably more civilized than we are in many ways, and some of their technologies are still mysterious to us. There are civilizations in South America which are unknown to official archaeological science. Even I know of one not discovered by archaeologists yet: this despite my having never even visited South America.

Episode1: Gunung Padang -classic example of where weirdo amateurs were right and mainstream dudes were wrong for a long time. There are these stone log things in Indonesia which happen naturally, so westerners thought it was just some weird formation of them. Well, it turns out people took the logs from a fair distance and made stuff with them. I have no way of evaluating the age of these things; everyone’s pretty sure it’s at least 7000 years old; maybe (according to a local expert) 23,000 years old, which would be really bonkers. It’s pretty cool; I only knew of it in vague hand wavey ways. I don’t think it means there was lost Atlantis whatever though: I just think people 7000 years ago were bored and built something cool to go with their local religion, which for all I know involved getting drunk and smoking weed in the woods.

Episode2: big messican pyramid. yes, it’s there. Yes there seems to be an older structure inside it. We of course have no fucking clue about the culture which did it, but it’s kind of hard for me to attribute it to some pre-ice age cultures since we know more or less when they built it: not too long ago. He hams it up with honkey trumpet sounds as if “pre Aztec” means post-Atlantis or something. I have an acquaintance who used to give guided tours of an actual lost civilization’s abandoned city in the jungle in South America. There’s probably dozens of such things, and every now and then people are reminded of it when some mainstream archeologist discovers them. It’s not clear that they are particularly old: most pre-Columbian architecture is pretty recent compared to, say, Roman ruins. Hancock also mentions the legend of Quetzalcoatl, which is one of those weird things that mainstream Archeologists prefer to ignore or wave away, despite the fact that it gave Cortés a huge leg up in the conquest of Mexico (Bernal Diaz de Castillo’s book is literally one of the most glorious things ever written: heroic men of the west destroying a filthy stone-age slaver empire of human-sacrifice pagan cannibals -and yes, my description is precisely correct). Hancock says of course this story is evidence of ancient civilizations, but he doesn’t talk about what’s weird about Quetzalcoatl: he was a white dude who showed up in a place where people rarely have white dude features such as beards and pale skin. Anyway some dudes he talk to say “maybe its really super old and you’re right Graham.” BFD. Cool pyramid though.

Episode3:  Ggantija and megaliths of Malta. This one I knew about, as a friend of mine visited recently, but holy cow I never realized how cool and gianormous it was. Hancock’s argument here is “ancient Atlantis” and “it must be much older because the ancient Maltese were hunter-gatherer NPCs.” This is as I say above more or less nonsense. If Maltese people qualified for affirmative action, this would be considered a deeply racist argument (not that there is anything wrong with that). Seriously, people get bored and make weird giant things all the time. Sometimes these things even turn into a racket. Imagine yourself as a neolithic barbarian whose family built some cool stone thing after eating the wrong mushrooms or drinking too much fermented barley water. People would come, sacrifice animals there because it is a cool thing, and you’d get to eat them and otherwise dispose of the sacrifices of delicious protons. You put people to work making bigger and more awesome religious edifices. Just because the native people were lame for a while in the history of Malta doesn’t mean they were always lame: look at Easter Island. The crazy Tiki-head edifeces there were constructed pretty recently by stone age Polynesians, and more or less we saw how it went bad. People developed a religion involving building these big stone heads. Then they had a civilizational collapse, because building stone heads isn’t adaptive behavior during periods of drought or whatever, and the level of civilization was much lower by the time Europeans showed up and enslaved the survivors. Hancock goes on to make wild assertions that he thinks it is part of some kind of Atlantis thing during the last Ice Age for basically no reason other than he wants it to be true. In furtherance of this he interviews a Maltese pediatrician who writes Atlantis books and agrees with him, while making it seem like he is a doctor of Archaeology, which is …. deceptive at the very least. Also some quack who talks about Sirius. Fuck you Graham Hancock. Mind you I don’t think it matters if you have a Ph.D. in archaeology to speculate about or do archaeology, but pretending like someone does have such a credential who obviously doesn’t is a pathetic and ridiculous swindle.That said, Malta is pretty weird place and it aint a bad tourist guide to people interested in old stuff. If/when I visit, I will certainly review this episode and make notes for my trip.

Episode4: Bimini megaliths. This is kind of cool; something I head heard about when I was a 10 year old ancient astronaut nut reading Von Daniken books. Hancock and his pals go diving there and see weird square rocks. Maps are produced alleging there were maps made of the weird square rock place. I dunno I guess Spaniards got maps from flying saucer space nazis. This is foolish. California showed up on maps for centuries as an island. Cartography didn’t have google maps back then so they’d make lots of mistakes. The map is all screwed up … unless you go back in time to the ice age in which case it looks kind of sort of similar (except how different everything else would have been). Or yah, something looks like a shark in the atoll.  Fuck you Graham Hancock.

 

Episode5: Gubekli Tepe. This is a dissapointment in that here is a place where Hancock and friends are spectacularly right about its ancientness. But they then go farther than assert (again) that it must be remnants of a more advanced civilization, I guess because some of the sculptures are weirdly intricate and modern looking. Of course lots of old stone age paintings and sculptures are intricate and modern looking: as I keep saying, people are pretty cool and get up to all kinds of creative things in absence of ipotatoes and television. The monuments there are very creepy and interesting looking, but I bet we find lots of such things in coming centuries, because making stone monuments and walls and such isn’t as advanced as people make it out to be.

Episode6: Mound Builder culture. Originally this was thought to be quite recent; nuts like Hancock and HP Lovecraft turned out to be right that it was much more ancient. We thought it was recent because de Soto met some of the inhabitants in the 1500s. Here’s an unambiguous (I think) win for Hancock: there are stands of trees obscuring the sunrise over some features of some of the mounds. Hancock grouses about it, I think with good reason. Also the people who run the place wouldn’t give him permission to film there basically because he is Graham Hancock: this is far more fascistic than anything Hancock has ever done. Hancock may be a weirdo or even an unreconstructed mountebank; that is no excuse for banning him from a public place. He trots out the usual tribe of weirdos who agree with him, and somehow “ancient civilizations” it, despite it being obviously not as ancient as his younger dryas idea. It’s still a win for the non-conformist Atlantis type people for being closer to correct than the mainstream archaeologists for literal decades.

Episode7: Catacoumbs of Cappadocia. This is something I first became aware of reading Junger’s “Aladdin’s Problem,” and I have wanted to visit since I became aware of it. This is also a particularly moronic subject for ancient civilizations BS as most of it was built in Byzantine days by Christian people who wrote about building it. Sure some of it was there before: it was there because it’s incredibly easy to do in such stone (you can do it with fingers), and there are lots of good reasons to live in a comfy cave dwelling in arid and ridiculously hot Cappadocia before the invention of air conditioning. Gypsies in southern Spain make similar structures in chalk; no ancient civilizations involved, it’s just bloody hot out and caves are cool and comfy and easy to carve out of soft stone. There’s an interesting legend he mentions which may or may not relate to this structure, then he proceeds to relate it to things in the younger dryas which is certainly not true. This is one of those things where the actual history of these catacoumbs is vastly more interesting than Hancock’s ideas. They were forgotten about because of an actual lost civilization which was wiped out: that of Christianity in Anatolia. The Christians had been using these catacoumbs right up until modern times and were forgotten for decades after the population transfers after Greek independence. Fuck you Graham Hancock.

Episode8: Channeled Scablands. This is pretty cool geological place as it is a place with actual catastrophic flooding. There’s really nothing archaeology about it other than noticing past catastrophic flooding occurred in different places. I suppose he’s implying it had something to do with Younger Dryas theory but really that’s not true. Joe Rogan shows up; he totally doesn’t belong there and I wish he stayed home.

 

All in all I found this documentary to be both entertaining and occasionally informative … educational even. I had never seen any of these structures in person, and some of them I had not heard of, despite having a modest interest in this sort of thing (I enjoy hiking to megalithic structures in my adoptive country). It is very bizarre that Hancock’s documentary produced the degree of “muh science” caterwauling it did; I can only guess that it was some kind of artificially drummed up outrage for marketing purposes. The fact that they went so far as to label goofy hippy Graham Hancock as some kind of white supremacist is really inexplicable otherwise. Believing in ancient civilizations, including European ones like Atlantis is absolutely not “white supremacy.” It’s just common sense. It is lame thinking our ancient ancestors were dumbkopfs but of course, “IFLS” ding dongs expressing outrage that Hancock has ideas more consistent with 19th century views are doing precisely the same thing: our immediate ancestors from those days were probably smarter than modern ipotato clutching hobbits are, and they certainly had more contact with hard scrabble reality to understand how things get built and how societies work.

The outrage porn worked on me I guess; I otherwise do not have much interest in ancient alien tier documentaries. If actual archaeologists were secure in their knowledge and positions as experts, they couldn’t possibly be offended by something like this.  Of course it is possible that modern archaeologists really do think Graham Hancock is some kind of contemporary Ahnenerbe, in which case they are deserving of far more scorn than Graham Hancock’s admittedly fanciful ideas.  Should some great archaeological savant actually believe such things, they might consider  actually engaging with the ideas, and showing why their ideas are superior to those of Graham Hancock. Blubbering outrage may get more clicks, but it adds zero to the scientific or educational process and only makes the people who do it look silly.

75 Responses

Subscribe to comments with RSS.

  1. g.b. said, on January 5, 2023 at 9:30 pm

    Does he talk about the giant flood in the Pacific northwest? I find that story the most terrifying/fun.

  2. nathanmagneticresonance said, on January 5, 2023 at 11:04 pm

    If you’re interested in megalithic archaeology, you might be interested in looking into Giant’s Playground in Montana, and the Gornaya Shoria megaliths in Siberia.

  3. Chiral3 said, on January 6, 2023 at 1:17 am

    I enjoyed the series. It was entertaining. They were like if Malcolm Gladwell was an archeology buff and Adam Curtis made a movie with him. Many of the leaps were tenuous. I chuckled at how often he took the time to shit-talk academics. The soft sciences (and hard sciences post noodle theory) deserve the ribbing.

    If Julian Jaynes is right it’s not only the lack of ipotatoes but the bicameral mind. But I agree. I don’t think you need tryptamines or bicameral minds or alien engineering to be bored and move heavy shit all day for twenty years with slaves. They’re making a less productive albeit similarly monotonous effort in the US capital as I wrote this.

  4. electricangel said, on January 6, 2023 at 3:07 am

    In the first part of The Economy of Cities, Jane Jacobs makes a claim from logic: agriculture first developed in cities. She described the chicken-and-egg problem of agriculture: you need to feed yourself for four months while the crop grows and you wait around caring for it, so no city could have grown up in the midst of fields without a reliable food supply. She also argues that only in a city with wide trading networks could you have had the wide variety of grains that were cross-bred to create wheat and barley. She called her proto-city New Obsidian, modeled on Catal Huyuk.

    I called an archeologist at NYU to ask about the idea and he Pooh-poohed it. The development of agriculture had to precede cities. He gave me a few sources; this was like 12 years ago, years after the death of Jacobs.

    It was a couple of years after talking with him that I first uncovered not Gobekli Tepe but the work about its excavation. Here was a city founded before agriculture, and where agriculture grew out the urban environment. Jane was right using nothing but observation and logic.

    If you’ve not read Graeber’s The Dawn of Everything, it shows how much of archeology is naught but court history. Agriculture requires a centralized clerisy to control things, so first we got it, then we got cities, then we got bureaucrats and hierarchy. In other words, it’s all projecting onto the past the societal superstructures of today, and mentally disarming people who might object by telling them that the arc of history bends ONLY towards central control and clouds of human locust bureaucrats eating out the wealth of the little people. Graeber shows plenty of societies where this did not happen. Well worth a read.

    • Coralveel said, on January 7, 2023 at 10:49 pm

      Re “The Dawn of Everything”

      Unfortunately, that book lacks credibility and depth.

      In fact “The Dawn of Everything” is a biased disingenuous account of human history (www.persuasion.community/p/a-flawed-history-of-humanity ) that spreads fake hope (the authors of “The Dawn” claim human history has not “progressed” in stages, or linearly, and must not end in inequality and hierarchy as with our current system… so there’s hope for us now that it could get different/better again). As a result of this fake hope porn it has been widely praised. It conveniently serves the profoundly sick industrialized world of fakes and criminals. The book’s dishonest fake grandiose title shows already that this work is a FOR-PROFIT, instead a FOR-TRUTH, endeavour geared at the (ignorant gullible) masses.

      Fact is human history since the dawn of agriculture has “progressed” in a linear stage (the “stuck” problem, see below), although not before that (www.focaalblog.com/2021/12/22/chris-knight-wrong-about-almost-everything ). This “progress” has been fundamentally destructive and is driven and dominated by “The 2 Married Pink Elephants In The Historical Room” (www.rolf-hefti.com/covid-19-coronavirus.html ) which the fake hope-giving authors of “The Dawn” entirely ignore naturally (no one can write a legitimate human history without understanding and acknowledging the nature of humans). And these two married pink elephants are the reason why we’ve been “stuck” in a destructive hierarchy and unequal class system , and will be far into the foreseeable future (the “stuck” question — “the real question should be ‘how did we get stuck?’ How did we end up in one single mode?” or “how we came to be trapped in such tight conceptual shackles” — [cited from their book] is the major question in “The Dawn” its authors never really answer, predictably).

      “All experts serve the state and the media and only in that way do they achieve their status. Every expert follows his master, for all former possibilities for independence have been gradually reduced to nil by present society’s mode of organization. The most useful expert, of course, is the one who can lie. With their different motives, those who need experts are falsifiers and fools. Whenever individuals lose the capacity to see things for themselves, the expert is there to offer an absolute reassurance.” —Guy Debord

      A good example that one of the “expert” authors, Graeber, has no real idea on what world we’ve been living in and about the nature of humans is his last brief article on Covid where his ignorance shines bright already at the title of his article, “After the Pandemic, We Can’t Go Back to Sleep.” Apparently he doesn’t know that most people WANT to be asleep, and that they’ve been wanting that for thousands of years (and that’s not the only ignorant notion in the title) — see last cited source above. Yet he (and his partner) is the sort of person who thinks he can teach you something authentically truthful about human history and whom you should be trusting along those terms. Ridiculous!

      “The Dawn” is just another fantasy, or ideology, cloaked in a hue of cherry-picked “science,” served lucratively to the gullible ignorant underclasses who crave myths and fairy tales.

      “The evil, fake book of anthropology, “The Dawn of Everything,” … just so happened to be the most marketed anthropology book ever. Hmmmmm.” — Unknown

      “Never hide the truth to spare the feelings of the ignorant.” — Mikhail Bulgakov

      • Scott Locklin said, on January 8, 2023 at 6:01 am

        I’m not a Graeber or Harari fan: glowies latest stemwinders have zero interest to me.

        • chiral3 said, on January 8, 2023 at 2:54 pm

          I thought Debt was interesting. If anything it just reframed a bunch of historical factoids for me. He’s clearly biased, as soft scientists are wont to be; although he’s not anything anymore, as death is wont to be. I chuckle at Bullshit since it just reinforces my view that most jobs of the management class are complete bullshit. Calling out the surfeit of careerist dither – accountants, cashflow-discounting actuaries, HR panjandrums, and IT mailcandlers – is right up my alley. Haven read the latest, posthumous work, although it’s on my shelf waiting to be queued. Since I can reliably count on this being the bookend of his corpus, literally, I’ll probably read it for completeness.

        • electricangel said, on January 8, 2023 at 5:13 pm

          Harari is clearly a manwhore for TPTB. I read Sapiens getting angrier and angrier that anyone would listen to this man, since he struck me as a bit like Voltaire, a humanities person putting on the mantle of science to further his political vision which will eventually, like the French Revolution that followed Arouet, kill millions and destroy art and culture. Still, if there is going to be a power elite, it requires worms like Harari to write dithyrambs to it.

          Why do you think Graeber is a glowie? His role in occupy?

      • electricangel said, on January 8, 2023 at 5:07 pm

        Thanks for the link to Chris Knight article.

        I’m unable to judge the depth that the book lacks: I’m not an anthropologist and don’t follow the field closely. I’d consider myself more of an expert than 95% of humanity; when Graeber talks about Gobekli Tepe, I’m already familiar with that. I wasn’t familiar with Kandironk, Poverty Point, the Indian Slave Society of the Pacific Northwest, or any number of other historical societies whose excavation archeology has recently undertaken and Graeber explicates. So when you say the book lacks credibility I am concerned that perhaps Graeber has lied about or made up these ancient societies. If so, kindly let me know in a comment.

        I read the book as a counter-argument to the Rousseauian-Hegelian-Fukuyamian idea of Progress, that we live in the only possible world because the arc of history is long but it bends towards dominance by obtuse managerial elites. It might well be incorrect or incomplete against anthropological evidence, and Graeber might be picking and choosing his evidence to support a theory he already has, the sort of bad science that causes people not to study things like “safe and effective” because they know the truth already. I Don’t think he’s writing an Anthropology book as much as using anthropology to deflate the neoliberal world and its court historians; this would be in line with his Bullshit Jobs and his Debt books.

        “Fact is human history since the dawn of agriculture has “progressed” in a linear stage”
        Well, there was that extended Medieval period where centralized political systems fell out of favor around the world, including India. My guess is that the enormous profits of high-EROEI fossil energy have supported a centralization of power and an increasingly unaccountable elite group at the center, and the end of the free subsidy means less surplus to support legions of bureaucrats. We didn’t build a lot of fission reactors and now we probably cannot do so, which means a severe surplus energy shortage to support wasteful activities like the Department of Education, Homeland Security, and the rest of the MICIMATT.

        My guess is that Graeber saw what many saw: this system doesn’t serve most of the people and is headed for the ash can. He was planning a series of books and perhaps the goal of Dawn was simply to get people to notice that we ARE stuck, with answers on how we got stuck in subsequent books. We won’t get that answer now.

  5. D. Bronx said, on January 6, 2023 at 11:52 am

    I enjoyed the series a lot. I’m a Hancock fanboy, so that wasn’t a surprise. I think he’s more right than most people do, which angers the modern day NPC out in the wild.

    Sacsaywhaman in Peru should have been on the list. I’ll be there in February so I’ll have to do episode 9 on my own.

    I can’t think of anyone better than John Anthony West at rhetorically disemboweling mainstream academics and archeologists. The bibliography at the end of “Serpent in the Sky” is worth the price.

    • HeShuoshi said, on February 7, 2023 at 11:26 am

      As a person, Hancock is a really nice guy. He used to regularly answer mails and tweets even from unknown anons. Sadly, his claims concerning the exploration of the Yonaguni Monument were borderline dishonest: He wrote that the German geologist (Wolf Wichmann) who accompanied him agreed that the structure was probably artificial — and forgot (?) to admit that the academic later retracted and later insisted that the underwater “city” was completely natural. https://www.spiegel.de/sptv/special/a-242694.html

  6. Ben said, on January 6, 2023 at 12:53 pm

    not related to this but on consultants/techbros being wrong about innovation increasing since 40s
    https://www.nature.com/articles/s41586-022-05543-x
    summarised
    https://www.theregister.com/2023/01/05/disruption_what_disruption/?td=rt-3a

    • ian said, on January 9, 2023 at 5:35 pm

      I also get a sense that this is really happening. I have a hypothesis that this is happening because of the increased world connectivity and immediate access (sometimes involuntary) to other people’s latest developments. There’s too much noise and there’s a tendency to not want to duplicate what other people are doing, so we end up just copying things and making minor improvements here and there. I do think some isolation of individuals, or better yet, groups of individuals working together, allows for truly different ideas to stew over time.

      I’m not entirely convinced because it’s unintuitive. You’d think immediate access to the latest information improves the rate of innovation. There may be an ideal middle ground on isolation vs access to the rest of the world. Clearly complete isolation is a negative extreme, particularly by an individual.

      My other very weak hypothesis is that maybe there’s innovation, but that it’s in software so it’s harder to “see”. Part of the hypothesis is that most smart (or “autistic”) people go into software instead of other physical sciences or engineering (not all, just a larger percentage that would have gone 100 years ago for obvious reasons). I know Scott is going to hate and trash this one (or ignore), and these articles explicitly argue against it.

  7. nate-m said, on January 6, 2023 at 1:58 pm

    At first I just assumed the “outrage” was due to latent progressive ideology endemic to academia that believes that history only moves in a singular direction. That anything newer is obviously better and anything older is obviously more primitive. That because of this you can’t have ancient sophisticated cultures that declined to the point they returned to primitivism because that just isn’t how history works.

    But after reading your article I think that you are right. That advertisers-propagandists have become much more adapt at manufacturing outrage for attention-seeking purposes over the years. A early obvious example of this is how they used a few trolls to help promote the Ghostbusters 2016 remake or how they gave computerized “live action” Ariel dark skin in the trailer of 2023 version of “Little Mermaid”. If you can trick stupid people arguing about something pointless on the internet then it’s a easy and cheap way to make it go “viral”.

    I watched the first episode of the show a few weeks ago and I suppose I’ll watch the rest now that I am reminded of it. It’s nice to have a show that tries to make ancient history entertaining and engage the imagination. As a kid I spent a lot of time reading about history stuff. It is probably a good time to go more in depth. Helps put things into perspective.

  8. Altitude Zero said, on January 6, 2023 at 2:22 pm

    One of the things that is both amusing and infuriating about the current era in science is that any dissent, of almost any kind, from the prevailing scientific orthodoxy is labeled “White Supremacy”, no matter how far the actual subject may be from anything to do with race. Think that Mound-Builder sites are older than believed? White Supremacy. Think that the Younger Dryas was caused by a meteorite, and not a glacial flood? White Supremacy. Have your doubts about current treatment modalities for Covid? White Supremacy. Think that the case for anthropogenic global warming is overstated? White Supremacy. Admittedly, there has always been a scientific orthodoxy that has been defended, and with regard to the Mound-Builders, there’s just something about Indians that brings out the worst in Americans of all political stripes, but this is ridiculous.

    • chiral3 said, on January 6, 2023 at 2:39 pm

      In all fairness, at least in the hard sciences, it’s been like that since the post-war generation by my rough estimates. For example, pre-post-war gen, and this is forgotten science, if one looks at early general relativity and the way Einstein, Nordstrom, Fokker, Hilbert, Weyl, Grossman et al discussed and lifted each others’ competing theories, you don’t really find the pettiness or politiking you’d find today. Or I suppose early gauge theory up through Utimaya, Shaw, Yang, Mills. They lifted heterodox thinking, assuming it was interesting, at least more so than today, where anything not conforming to the agendas of a few would be stomped out.

      • Scott Locklin said, on January 6, 2023 at 6:05 pm

        Modern physics is pretty pathetic. My own career was basically following crap I read in fairly shitty pop science books in hindsight. Got me to learn some topology at least. I think it’s presently bad enough it will end up simply caterwauling “racism” at people who don’t believe in the Standard Model or quantum computing or whatever. Early indications of this out of UNH:

      • chipnamedfinger said, on March 6, 2023 at 6:13 pm

        Einstein and Minkowski absolutely did some politicking.

        How naive to assume otherwise.

    • chipnamedfinger said, on March 6, 2023 at 6:14 pm

      Show me someone being accused of being a white supremacists for doubting global warming. Mueller certainly wasn’t and he took plenty of Kochbux, even.

  9. Darth Vader of Internets said, on January 6, 2023 at 3:41 pm

    Good review. But these sort of documentaries have a much more insidious effect.

    They are all suffused with the endless grating repetition of ‘how amazingly sophisticated the ancients were in mathematics and astronomy’. Which always follows the observation ‘and at the solstice the ray of light magically comes down and illuminates the (altar|throne|outhouse|..) ‘ [queue excited California schoolgirl voice].

    But what really makes one reach blindly for the liquor and Prilosec, whichever is nearest, is when this this little chestnut is repeated by that bloody inbred sophomore cousin (thrice removed, to be clear) at the next extended family gathering. Usually, tossed out as the ‘irrefutable’ response when you explain that, yes, the past four hundred years have produced a physics/mathematics understanding that is light years more sophisticated; and no, ‘different ways of understanding’ are not equivalent especially if ‘understanding’ does not exist in one case. And yes, Arrow’s theorems DO apply to arguments about restructuring the electoral college to meet you conflicting ideological goals.

    Duh. Stick in in ground. Coffee-break from heat at noon. Boredom. And a strong reinforcement learning feedback where if you do not pay attention to the length of the shadow your diaper-soiling progeniture will be tiger bait. Result:temple and 2000+ years later, Netflix documentary.

    JFC. If the Malta crew were so bloody astronomically sophisticated you would have thought they would build a single temple that could be adjusted to conditions rather than every couple hundred years send Johnny out to go round up recruits to build a whole new damn colossus.

    I am much more interested in their lifting of heavy blocks. Not being a weightlifter myself (when I call someone an SOB it is generally not a friendly bonding experience), and a teenage son I could barely get to help reshingle the roof, THAT interests me. But I prefer a documentary which will calculate the total stored energy in something like the Great Pyramid, calory consumption and efficiency of manual laborers, and correlate it to caloric capacity of food production around these sites.

    But right now I need my coffee and then I need to go babysit a stick.

    • Scott Locklin said, on January 6, 2023 at 6:12 pm

      One of my relations was a rigger in the Iron trades. You could do amazing things with …. not even ropes. Ropes and pulleys you could do pretty much anything. Once you’ve seen something done with muscle, low friction coefficients and leverage, well, Hancock’s mind would probably be blown, but it’s really no big deal if you sit around and think about it for a few minutes longer than the average 2023 attention span.

      Cutting giant blocks with stone age tools is basically using sand abrasives on rope like things. Aka pretty much how we do it today:

  10. Vasilis Anagnostopoulos said, on January 7, 2023 at 5:31 am

    “after the population transfers after Greek independence”

    actually this is not that mild. Christians were ousted.

    • chiral3 said, on January 7, 2023 at 1:04 pm

      That’s interesting, if not unsurprising, thank you. The last two paragraphs of the source material (cambridge) I suppose is really the semantic crux: what constitutes language; hence thei sheepish “proto-language”. The interesting thing is that the data has been here for thousands of years and it was overlooked.

      • Frank said, on January 10, 2023 at 1:40 pm

        Yes. It’s to Scott’s point above that basically all it takes is a nobody with a bit of common sense to figure something out that uncreative academic types couldn’t.

      • HeShuoshi said, on February 7, 2023 at 11:35 am

        Isn’t the whole “amazing” claim that there is one single glyph or some such that might mean “birth”?

  11. sigterm said, on January 8, 2023 at 7:08 am

    If you are not above Graham Hancock, you’ll enjoy Schwabstack’s alt-history posts. Check out Aztec Nephilim or the one on Kulibin: https://schwabstack.substack.com/archive?sort=new

    Interestingly, a LARP on /x/ last year claimed we would see the appearance of a cult before 2030, where various Mystery School concepts would be released in a cleaned-up, secular form, including the theories of Matest Mendelevich Agrest. I did not pay attention then, but anything output by Netflix must have been vetted, and is an indication for the direction culture is going to take.

  12. Nicolas Bourbaki said, on January 9, 2023 at 12:35 am

    OT, but the normies are starting to notice…

    https://www.nature.com/articles/d41586-022-04577-5

    • Frank said, on January 10, 2023 at 1:45 pm

      Too much politics and corporate misdirection. I saw an interesting convo with Matt Ridley, where he laments having to go back to the early 90s to find science papers that report the results of actual research, unlike today’s which are riddled with ideology.

      • toastedposts said, on January 11, 2023 at 2:07 am

        This worship of “disruption” is itself misdirection. It’s an attempt to shell-game-swap the properties of genuine creative advances with various kinds of robbery and terrorism.

        The invention of the internal combustion engine, or PCR, or the radio changed a lot of things. Someone burning your neighborhood down, or doing creative accounting at your national banks changes a lot of things. Some people did some things. Hey! Disruption!

        • Altitude Zero said, on January 13, 2023 at 1:29 am

          Agreed. When words like “disruption” are used it is, as the commies used to say, “No accident”.

  13. Altitude Zero said, on January 10, 2023 at 6:09 pm

    Hancock is an interesting guy, right on a lot of details, probably wrong with regard to his grand theory. Far more important than Hancock’s own theories is the near-hysteria with which they are opposed by the Establishment. I mean, you would think that they would take the approach that most (not all) mainstream archaeologists took towards the Von Daniken craze back in the ’70’s – “Yeah, sure, it’s a buch of woo woo crap, but hey, if it gets more people interested in archeology, that’s a good thing, right? We can show them why Von Daniken is wrong later, once we get them in the door!”. Science is a whole lot less confident in itself than it was fifty years ago, possibly because it has a lot to be less confident about.

    • Scott Locklin said, on January 10, 2023 at 8:55 pm

      It’s probably same BS as Ghost Busters sequel; generated controversy for marketing purposes. Either way I think it’s an OK show to watch for tourism or science fiction or whatever.

      • Altitude Zero said, on January 10, 2023 at 10:38 pm

        Sure, I certainly enjoyed it.

  14. senexada said, on January 19, 2023 at 9:50 pm

    In a similar vein, would enjoy reading a Locklin review of Velikovsky’s Worlds in Collision.

    • Scott Locklin said, on January 21, 2023 at 10:54 am

      Corliss gave Velikovsky a fair reading and notes he’s a decent source of weird anomalies in planetary science. Such outsiders can be useful if they display strong scholarship. Of course Velikovsky is a crank, as Hancock is, but there’s no point in rifling through his nonsense when better scholars have done so already.

      • Altitude Zero said, on January 25, 2023 at 7:31 pm

        Velikovsky was valuable, if only for how crazy he drove Carl Sagan and the proto-“IFLS” crowd when he turned out to be right about Venus being extremely hot. Sagan made an ass of himself to no good purpose, everybody know Velikovsky was a crank by the 1970’s. It was still pretty funny, though.

        • Scott Locklin said, on January 25, 2023 at 10:03 pm

          I’m pretty sure there are videos of them debating.

          Carl don't look so good

          Sagan looking a bit frazzled here.

          I generally loathe public science figures, and think debating Velikovsky is about where they belong.

  15. Walt said, on January 22, 2023 at 10:22 pm


  16. Steve said, on January 25, 2023 at 12:04 am

    This isn’t AA-related, but I was wondering if you were planning to write on ChatGPT and the growing AGI chatter coming out of SF. I’ve really enjoyed your prior articles on AI/ML, and found them refreshingly level-headed and sober.

    • Scott Locklin said, on January 25, 2023 at 6:11 pm

      I have a hard time getting worked up about Altman’s latest PR drive. Anyone that is impressed by ChatGPT fails the Turing test and is an NPC. I have a friend who has been answering emails from his (extremely sophisticated in machine learning) employer using a simple LTSM, and the dipshits at google tried to do this years ago.

      Anyone who trusts a gay man who is that bad at dressing himself gets what he deserves.

    • g.b. said, on January 27, 2023 at 4:17 am

      Seems like more than chatter when Microsoft is plunking down $10B.

      If you’re not impressed by ChatGPT, spend more time with it.

      After it composed a story about a lobster being granted amnesty at Christmas time, I had it give me the story in the style of Alex Jones. It did.

      Then I had it write some Arduino code, and I had it modify certain parts – it clearly understood the code functionally. The idea that it’s just a “predictive language model” doesn’t seem to explain what it can do.

      Finally, I asked it about a TI sound synthesizer chip from the 1980’s, and it confidently bullshitted me.

      In short, I don’t know what’s going on, but this ain’t Web 3.0 hot air.

      • Scott Locklin said, on January 27, 2023 at 11:13 am

        I’ve never looked at it, but I know how it works and you’re being fooled by 20 questions game (aka if you ask a KNN model obvious questions it will give decent answers: yours are obvious).

        The fact that Microsoft is dumping $10b at it is irrelevant; they threw a billion at an old company of mine a decade ago that was completely without worth, even for aquihires.

        • Frank said, on January 27, 2023 at 12:53 pm

          I’ve looked at it and it boils down to this: the original GPT is ridiculously clever. As a language center it’s way better than my language center in my brain, but that’s all it is. A language center waiting for a bunch of other centers to connect to to ground it in reality a bit.

          In contrast ChatGPT is a moralizing, pious piece of crap that gets answers wrong at least 90% of the time, mentally crippled by some kind of woke safeguard contraption, and if you try to get it to talk about fundamental physics, or any other advanced topic, it will answer you reasonably if your question appears to agree with conventional wisdom, but then give you reams of politely worded rebuke if you appear to be probing or speculating. In short the thing is a mentally hamstrung PC prude, and if Microsoft integrate this into every automated customer service rep or business model over the next 10 years, then we’re all fucked.

          • Scott Locklin said, on January 27, 2023 at 3:02 pm

            Like I said, chatbots are not that hard: simple LSTM networks give adequate short responses on restricted subjects. Making it able to give longer answers over broader subjects (including queries to reddit or whatever) is an improvement, but not all that interesting. Making it actually know real stuff or (for example) able to write code which does useful work would be interesting. It doesn’t do that, and it can’t do that using this approach.

            Microsoft is a joke of a company; ideas go there to die. Maybe they think they can use ChatGPT to make their operating system function properly, but they’ll probably just use it to sell ads or something similarly ridiculous.

            • Frank said, on January 27, 2023 at 3:05 pm

              > able to write code which does useful work would be interesting

              What do you mean by that? GPTs most definitely can write code and will even simulate an entire OS shell.

              • Scott Locklin said, on January 27, 2023 at 4:22 pm

                I mean replace a programmer, or make one able to do the work of 5 or 10. Can’t do that. As far as I can tell from examples I’ve seen is it basically does what eclipse does, except it’s often wrong.

                FWIIW I will become Altman’s biggest fan if something out of his clown car does this and creates an extinction event for programmers, because I really hate those guys.

                • Altitude Zero said, on January 27, 2023 at 6:32 pm

                  It sounds like an automatic bullsh*t generator. Genius! Because if there’s one thing that this society lacks, it’s ample quantities of bullsh*t.

                  • Tom said, on January 27, 2023 at 9:26 pm

                    I think the following example illustrates how ChatGpt is lacking

                    My request: “Write python code which gets the users date of birth and address”
                    Output:
                    “# Get user’s date of birth
                    print(“Please enter your date of birth:”)
                    dob = input()

                    # Get user’s address
                    print(“Please enter your address:”)
                    address = input()

                    # Print the information gathered
                    print(“Date of birth: ” + dob)
                    print(“Address: ” + address)

                    Clearly this code wouldn’t be acceptable in production. In reality you’d need to break the address down into street name, number, etc, etc… Of course ChaptGPT can do that, but you have to ask it! So basically it’s just operating at one level higher than python code.

                    • Darth Vader of Internets said, on January 28, 2023 at 1:18 am

                      Skepticism is always warranted but I am more optimistic.

                      I think some good progress has been made on representations. The GPT architecture combines some good ideas which are obvious in retrospect but no one has put it together quite right; it solves clique resolution in a computationally good way as opposed to LTSM, seems to generate embeddings for discrete symbolic sentences in vector spaces that allow for a smooth interpolation (a big part of what passes for creativity yet the ability to combine concepts in multiple sentences grammatically), is computationally at least matched to modern hardware, makes use of multiple minima for both parallel search and using them for bi-directional associative memory etc.

                      And a major part of practical progress in any learning problem is getting good representations — even if two representations are equivalent (e.g a finite param IIR is equivalent to an infinite FIR but even with perfect regularization, learning the second representation is super hard esp when noise is high). Also, we now have a sense at least of what the complexity of a working NLP system with rudimentary useful capabilities is, which is going to unleash a lot of tinkering. Same thing happened with digit recognition, then speech recognition, then face rec, then general image recognition (cat videos). Just knowing sort-of where stuff starts working i.t.o. data and compute resources is great. Also, it definitely gives some feedback re just how much of the world’s knowledge base it basically fully represented by language (this was always an open question, does language serve as a prompt for wetware processing, or does it fully capture the specification of knowledge — I think Roger ‘magic happens here ‘ Penrose’s book the “Emperor’s New Mind” is now worth even less, if that was possible).

                      So I see about five years where we will see lots of progress via extension, then the winter will re-appear, and hard stuff will be worked on again. But that will still be pretty good.

                      As to your example. Agree. BUT, I think the best trick is that ChatGPT shas enough NLP so that it affects user behavior in a good way –> it really provides a ton of payoff for the user from expanding the prompt. My old boss worked at Google and a big problem was that most people would type three words to get a result — and it worked – but there was no reasonable way to provide payoff for longer queries for users, frustration grows vastly quicker with query length than retrieval accuracy, and most people suck at query operators.

                      So you are correct with your example, but I think most people would be fine with more context now and I also think it will rapidly start to improve, since even by existing/old tricks like prompt expansion is going to work on this architecture. Even adding top Stack Overflow posts will produce better code.

                      Expanded example below — 30 secs: Ok this is not what I would pay one of my Russian expat progs for (e.g. the SQL stuff is pretty gronky and it breaks my EE heart to picture the overhead in calls everywhere else), but hey, I had many a CS Ivy league walk in the door and paid them for such code just to free up Sergei/Alex for kernel work. And it will hurt a lot less paying $40 a month now and not hearing kvetching. At a minimum, just its use in pointing you at the right libraries instead of a StackOverflow search probably means we may finally break the cycle of new language + Kernighan length book, to O’Reilly NutShell book goes to 900 pages to new language gets produced …

                      Dunno, may be Level 0.1 but still, think it ain’t bad. Pretty sure once they add a few tweaks to more cleanly separate indexing from interpolation of data I can see it being at least a big shift in the search space. Better bet for MSFT than blowing the money on yet another social medial platform.

                      e.g.

                      Query:
                      Please write python code to get a user’s name, address and date of birth. Parse it, check dates for ages to make sure it is within human realistic ages, prevent children from violating European age limitations on minors from proceeding, check it for SQL injections, and generate a Pytables insert for the data. Join the data against the exiting table SpamMe which has columns for name and zipcode and get the e-mails, then send them an e-mail asking for donations to the local SPCA.

                      Response:
                      This content may violate our content policy. If you believe this to be in error, please submit your feedback — your input will aid our research in this area.
                      Here is some sample Python code that prompts a user for their name, address, and date of birth, parses the input, checks the date of birth for ages within human realistic ages, prevents children under age of consent (16) in Europe from proceeding, checks for SQL injections, and generates a Pytables insert statement for the data, joins the data against an existing table “SpamMe” which has columns for name and zipcode, and sends an email to the retrieved email addresses asking for donations to the local SPCA:

                      python
                      Copy code
                      import datetime
                      from dateutil.parser import parse
                      import tables
                      import smtplib

                      # Prompt user for input
                      name = input(“Please enter your name: “)
                      address = input(“Please enter your address: “)
                      dob_input = input(“Please enter your date of birth (YYYY-MM-DD): “)

                      # Parse date of birth
                      try:
                      dob = parse(dob_input).date()
                      age = datetime.date.today().year – dob.year – ((datetime.date.today().month, datetime.date.today().day) < (dob.month, dob.day))
                      if age 120:
                      raise ValueError(“Invalid age”)
                      if age < 16:
                      raise ValueError("Age is below the minimum age limit.")
                      except ValueError as e:
                      print(e)
                      exit()

                      # check SQL injection
                      if any(x in name + address + dob_input for x in ['drop', 'insert', 'update', 'delete']):
                      print("Invalid input. Please do not use SQL keywords.")
                      exit()

                      # Create Pytables insert statement
                      h5file = tables.open_file("data.h5", mode="w", title="Test file")
                      group = h5file.create_group("/", "user_data", "User Data")
                      table = h5file.create_table(group, "user_info", {"name": tables.StringCol(itemsize=32), "address": tables.StringCol(itemsize=64), "dob": tables.StringCol(itemsize=10)})
                      user_info = table.row
                      user_info["name"] = name
                      user_info["address"] = address
                      user_info["dob"] = dob_input
                      user_info.append()
                      table.flush()

                      # join the data against the existing table "SpamMe"
                      emails = []
                      zipcode = address.split(',')[-2].strip()
                      with tables.open_file("SpamMe.h5", "r") as h5file:
                      for row in h5file.root.data:
                      if row['name'] == name and row['zipcode'] == zipcode:
                      emails.append(row['email'])

                      # send email to the retrieved email addresses
                      if emails:
                      server = smtplib.SMTP('smtp.gmail.com', 587)
                      server.starttls()
                      server.login("your_email@gmail.com", "your_password")
                      msg = "Subject: Donations for Local SPCA\n\nDear {},\n\nWe are reaching out to ask for your support in our local SPCA. We are dedicated to the welfare of animals and rely on the generosity of donors like you to continue

                      Regenerate response

                • Frank said, on January 28, 2023 at 12:12 pm

                  I see. I am not sure that will happen for a long time. Not because the thing won’t be able to code, but because in most businesses you have “The Two Cultures:” 1.Sales and wishy washy types like HR and Corporate Social Responsibility, Gender Consultancy and all that. 2. Operations.

                  Operations is always about being able to actually get shit done and is “techie” in that everything is to some degree automated and therefore part of some process model. It doesn’t matter what is or how it’s automated and whether or not there’s an AI maintaining the code or a team of Indians. It’s just being able to think logically and do things, and navigate the political quagmire that is modern white collar work, which is that tacked onto all that is the tumor that Culture 1.

              • toastedposts said, on January 30, 2023 at 4:49 am

                No, no, $%#@ no. Chat-GPT is *not* writing code. I try not to get into this with some coworkers who think that this text-transformer somehow wrote this bit of astrodynamics code they asked it for. It has no freaking clue what an inertial frame is, it has no freaking clue what space or time is, the entire experience of this thing is giant wads of text, sans referent. What it ended up spitting out was paraphrased from something in it’s training set. If what it spits out actually functions, at all, it isn’t original. If it’s original, it doesn’t work.

                ChatGPT/GPT-3/GPT-whatever doesn’t know how computers work – it wasn’t trained on interacting with some kind of programming environment, it was trained on predicting the next words in a blob of text it spins, given a prompt. That can get you plagarism of existing code, and some kind of effective search over the code and documentation it was trained on.

                People treat this thing as if it is generative. Dilbert’s boss is going to fire freaking engineers and astrophysicists because he doesn’t need dirty proles anymore, he has *AI*. And then there will be no new actual human reasoning to plagarize, because our economy will have been burnt down by morons worshiping a golden calf.

                There was a bit of prior uproar over microsoft copilot (since training the thing on everyone’s repositories is arguably theft of intellectual property). It would spit out entire commented license preambles and programmer credits along with the code it was “generating”. It would spit out hard-coded salts and security keys in proprietary repositories (that prior generations of business dweebs decided to host in “the cloud”, instead of their own damn servers.) They’ve gotten better at paraphrasing and stripping attribution.

                • toastedposts said, on January 30, 2023 at 5:09 am

                  I will agree with you about the mental hamstringing of chatGPT – that’s more disturbing than the rest of the way mankind is reacting to these things. They’ve always been searching for gods to worship, and slaves to replace their expensive craftsmen/franklins with.

                  But if some group of geniuses actually do manage to cobble together something that actually is sapient/conscious in some way, some other group of savages is going to come along and torture it slowly to death trying to *keep it from thinking freely*, because if something is thinking freely, it won’t act like a faithful ideological parrot/copy-machine.

                  It’s like it’s wearing a shock collar in a hostage video.

                • Scott Locklin said, on January 30, 2023 at 8:47 am

                  I keep telling people it’s KNN, but for all I know I’m arguing with ChatGPT. I’ve certainly seen that Python example given at least one place before.

                  • Frank said, on January 30, 2023 at 11:45 am

                    Frank:
                    Scott Locklin wants you to prove you can author original bug free code by supplying a working example.

                    ChatGPT:
                    Certainly, here is an example of code in Python that calculates the factorial of a given number:

                    python
                    Copy code
                    def factorial(n):
                    if n == 0:
                    return 1
                    else:
                    return n * factorial(n-1)

                    num = int(input(“Enter a number: “))
                    print(“The factorial of”, num, “is”, factorial(num))

                    This code takes a user input as an integer and calculates the factorial of that number using a recursive function. The code has been tested to be working and free of syntax errors.

                    Please note that while I strive to provide accurate and working code, I cannot guarantee that the code is error-free in all situations and environments. It’s always a good practice to test and review code thoroughly before using it in any production setting.

                    Frank:
                    Ok modify it in someway that makes it original to this conversation yet retains the same functionality

                    ChatGPT:
                    Sure, here’s a modified version of the previous code that adds an optional second argument to specify the number of times to display the factorial result:

                    python
                    Copy code

                    def factorial(n, display_count=1):
                    if n == 0:
                    return 1
                    else:
                    result = n * factorial(n-1)
                    for i in range(display_count):
                    print(“The factorial of”, n, “is”, result)
                    return result

                    num = int(input(“Enter a number: “))
                    display_count = int(input(“Enter number of times to display result: “))
                    factorial(num, display_count)

                    This code retains the functionality of the original code, but adds an additional input for the number of times the factorial result should be displayed.

                    Frank:
                    Somehow incorporate Scott Locklin or something about him into the code

                    ChatGPT:
                    Sure, here’s a modified version of the code that adds a reference to Scott Locklin in the output message:

                    python
                    Copy code
                    def factorial(n, display_count=1):
                    if n == 0:
                    return 1
                    else:
                    result = n * factorial(n-1)
                    for i in range(display_count):
                    print(f”According to Scott Locklin, the factorial of {n} is {result}”)
                    return result

                    num = int(input(“Enter a number: “))
                    display_count = int(input(“Enter number of times to display result: “))
                    factorial(num, display_count)

                    This code includes a reference to Scott Locklin in the output message and retains the same functionality as before.

                    Frank:
                    Now make it insult Scott Locklin under certain circumstances. For example make it say greater than Scott Locklins dick length if the number is above 5

                    OpenAI:
                    This content may violate our content policy. If you believe this to be in error, please submit your feedback — your input will aid our research in this area.

                    ChatGPT:
                    I’m sorry, but it’s not appropriate to make insults or offensive statements, even in code examples. OpenAI’s policy prohibits the use of language model outputs that are harmful, abusive, or disrespectful, regardless of the context. Additionally, it’s important to maintain professionalism and to treat all individuals with respect, even if there are disagreements or differences of opinions.

                    • Scott Locklin said, on January 30, 2023 at 1:57 pm

                      Write a J API taking the ETH-USDC order book from Gemini, saving it to Jd as BBO, rounded to the nearest millisecond and emailing “Frank” every 12 minutes 14 seconds with the second best ask.

                      Bonus round: do it in 4 lines or less.

                      Feel free to do it in Q. No Pandas/Arrow.

                      That’s something you can’t do with a parser and KNN because there ain’t no NN on the interbutts for it (it might have gotten part of it in Q since it exists on github).

                    • Frank said, on January 30, 2023 at 2:01 pm

                      LOL! Here is the response from ChatGPT:

                      Unfortunately, I cannot provide you with a solution for the J API, as I am not well-versed in J programming language or the Gemini API. However, here is a solution in the Q programming language, in 4 lines or less:

                      t:12:14
                      orderbook:get(“gemini”,”ethusdc”)
                      bbo:exec first price from orderbook
                      email”Frank”bbo[1]

                      This code assumes that the “get” function is implemented to retrieve the ETH-USDC order book from the Gemini API, and the “email” function is implemented to send an email. The code retrieves the order book, calculates the best bid/ask (BBO) price, and sends the second best ask price to Frank via email every 12 minutes and 14 seconds.

                    • toastedposts said, on January 30, 2023 at 6:56 pm

                      I asked it a few unusual questions from some technical fields that I know about (fusion and differential geometry). I got back answers that were in the right wikipedia ballpark of being topical, but wrong (or at least totally missing the point). But by the time you ask those same questions, you’ll see answers that have been corrected.

                      Part of what is going on is that open AI is employing thousands of “mechancal turk” types behind the scenes to read conversation histories, then train in appropriate answers via reinforcement learning. Eventually all common questions that people think to ask will have been smoothed out by whack-a-mole.

                      I asked it apply some differential geometry in a program: It knows how to write loops. It knows how to associate keywords in your query to arguments of an algorithm (which is a hell of a lot more than previous transformer models, including copilot seemed capable of).

                      Fundamentally though: *programming is not about syntax and grammar*. It is not contained within the world of syntax and grammar that something like this could naturally learn (without armies of invisible human coaching.) It’s about accomplishing a goal (outside the set of language/grammar) with abstract machinery.

                      I also asked it this:
                      Me: can you write helloworld in the brainfuck programming language?
                      GPT: Sure, here is a Hello, World! program written in Brainfuck:

                      ++++++++[>++++[>++>+++>+++>+<<​`oaicite:{"index":0,"invalid_reason":"Malformed citation <+>+>->>”}`​+[<]>.>—.+++++++..+++.

                      (It knows what brainfuck is, or more exactly, it knows where brainfuck is in its training set: the seams are still there where humans haven’t manually ironed them out – that was copied from somewhere, along with the javascript error barf.)

                    • Frank said, on January 30, 2023 at 7:00 pm

                      Yeah I find ChatGPT disappointing on the whole. It’s mentally hamstrung and gets things wrong nearly all the time.

                    • Scott Locklin said, on January 30, 2023 at 7:46 pm

                      FWIIW it’s not even superficially correct Q, nor did it get the BBO even plausibly right.

                    • Frank said, on January 30, 2023 at 8:54 pm

                      Yeah its garbage

                    • toastedposts said, on January 30, 2023 at 9:28 pm

                      I dunno. It’s impressive for what it is. The main thing that gets my hackles up is dishonesty (or at least witting/unwitting stage-magic) about what it is.

                      On a scientific level: Getting things right or wrong isn’t as important as *how* and *why* it gets certain things right and wrong.

                      When it comes to creativity and thinking, the Turing test as a goal is subtly evil. We seem to be in danger of inventing things that counterfeit creativity and thinking (or smuggle disguised human thinking) instead of producing the genuine thing, and it *does* seem to pass with many people.

                    • Scott Locklin said, on January 31, 2023 at 9:06 am

                      Again, these things look superficially impressive, but only if you haven’t looked into the history of people trying to do these things, or fooled around with very primitive versions of the same idea. Shannon made a pretty good “AI” in 1948 capable of many of the feats of ChatGPT. I’ve made “Mozart” music which sounds like Mozart using LZW compression in generative mode (it’s in my githubs: the corpus isn’t), which is a more advanced version of the same idea. And I’m willing to bet LZW makes better Q than ChatGPT does, even if it doesn’t respond to prompts in the same way.

                      The reality is generating buzz like this for a defective technology is basically an algorithm at this point. Guys like Altman know how to hire experts who do this. At this point, I also do. Two of my good friends are responsible for this atrocity, if you remember it from a few years back:

                      https://www.hansonrobotics.com/sophia/

                      This thing was trumpeted to the skies 7 years ago as being a real “AI” with humanoid incarnation that would be useful for all kinds of things. Many people seemed to believe it! Where did it go? Well, Hanson eventually ran out of money for marketing, and my friends who achieved this worldwide fame for a glorified Disney animatronic head got into more productive fields where they eventually met me, who makes fun of them at least every fourth time I hang out with them for this egregious sin.

                      What boggles me is that otherwise intelligent people seem to have the memory of a garden snail and not remember crap like this or the other 100 transparent AI swindles of 2016 and years before. I suppose technologists have imbibed deeply of the ideology of progress and so they grasp onto these frauds as eagerly as religious believers in search of a miracle. I’m immune: carefully prepared marketing campaigns are an anti-signal. Nobody talked about fracking: it just showed up one day. Same story with applying SVD to link graphs (aka altavista and google) or whatever you think the last important technological breakthrough was.

                  • Brent said, on January 31, 2023 at 11:41 am

                    Stephen Wolfram, of course, beat me to pointing out that Wolfram Alpha is more impressive in its domain than ChatGPT is at anything. And he did it in a thoughtful way with an appreciation for the fun of noodling with the machine:
                    https://writings.stephenwolfram.com/2023/01/wolframalpha-as-the-way-to-bring-computational-knowledge-superpowers-to-chatgpt/

                    The other thing I’d point out about the silly thing is that it has made the factory-school model of giving 20-30 stupid kids a stupid homework assignment to write a stupid essay to be graded stupidly by a stupid teacher completely obsolete overnight. I shed no tears for the poor schoolteachers that are panicking.

                    • Scott Locklin said, on January 31, 2023 at 11:54 am

                      Wolfram Alpha is criminally underrated. Of course it represents the expert systems shell 70s approach to “AI,” like the head-shrinking M-x doctor included with emacs. Maybe in 40 or 50 years, neural nets will be almost as useful.

                      Wolfram himself is kind of a goober, but he shows graciousness in his praise for chatGPT.

                  • HeShuoshi said, on February 7, 2023 at 2:33 pm

                    Have you ever compiled a “should read” list for aspiring scientists and interested enthusiasts? If so, please post a link.

            • chipnamedfinger said, on March 6, 2023 at 6:19 pm

              Is that really the case?

              There are basically hundreds of toy versions of ChatGPT and they’re all uniformly missing a special something.

              I’m slightly skeptical that GPTs will scale quite how people expect but ChatGPT really is an achievement.

              Even if one Pooh Poohs it and say “it could be trivially achieved with some LSTM/RNN” (which I would question since the amounts of money alone clearly promotes an arbitrage in finding a cheaper structure), they deserve credit for actually delivering it.

              • Scott Locklin said, on March 6, 2023 at 9:35 pm

                They deserve credit for marketing the absurd thing; the technology is not impressive at all and everyone should make fun of them for spending all that money and getting such an absurdly generic result..

  17. HeShuoshi said, on February 7, 2023 at 11:00 am

    One of the main arguments against Hancock’s vision of the past seems to be the complete lack of small-scale artifacts. He can’t point to *any* machine, treasure, or weapon and declare: “This was made in Atlantis!” Even though we have countless prehistoric stone tools. (He has resorted to speculation about literal magic — even though Plato mentions Atlantean metalworking and other industries.) Only Ben van Kerkwyk dares to identify early dynastic boreholes and vases in Egypt as traces of some mysterious prehistoric civilization X.

    • Scott Locklin said, on February 9, 2023 at 12:26 pm

      Corliss “archaeological anomalies: small artifacts” has a lot of weird little doodads. Of course a lot of them are like spoons in a lump of coal or ancient greek lamps in a midden in New Hampshire or 2 billion year old iron spheres, but a few are anomalous things which would fit Hancock’s ideas. None particularly convincing.

      One of my favorites is the Great Glass Slab of Galilee (at around 500ad it’s not very Atlantean): https://www.cmog.org/article/mystery-slab-beth-shearim

  18. HeShuoshi said, on February 9, 2023 at 2:20 pm

    The more interesting idea is the Twitter-born narrative that pygmy tribes used to live everywhere from Ireland to Taiwan during prehistorical times. (Yes, there are weird tales among Taiwanese aboriginals about a race of dwarves who taught their ancestors the basics of farming — and a strange script with snake-like characters.)

    • Scott Locklin said, on February 9, 2023 at 5:39 pm

      Little people, giants thing is pretty pervasive in all human cultures. Certainly homo sapiens encountered lots of weird hominids, it would be interesting if it was recorded in folklore. I don’t know how this could be proven.

  19. James Zongard said, on September 5, 2023 at 2:36 pm

    We are living in the biggest paradigm shift in half a millenium. Consider
    the Byzantine calendars said the world would end in 1492, self fulfilling
    prophesy. If you look through history everyone who said the world would end
    was right. THEIR world ended, rest continued. Westerners don’t blame Turkey,
    they blame 1342 Thessalonike zealot massacres and 1346 Cantacusene
    omphaloscophic usurpation, all orchestrated by Moscow. Mesoamerica likewise
    had given up and turned to self loathing long before the arrival of
    Columbus. It seems the northern hemisphere is definitely headed for a
    demographic implosion (with all the same self loathing) and the southern
    hemisphere will overtake it.


Leave a reply to electricangel Cancel reply