The Decline in the Transmission of Scientific Ideas

Enrico Berkes and Ruben Gaetani, The Decline in the Transmission of Scientific Ideas, NBER, July 2026.

Abstract: We document that the diffusion of new scientific ideas beyond their field of origin has declined substantially over the past four decades. This contraction is closely linked to increasing spe- cialization in scientific language: research that employs more technical terminology tends to be adopted less broadly. We develop a theory of scientific discovery in which the diffusion of new ideas depends on the degree to which potential adopters can understand and process them. When introducing their discoveries, scientists face a tradeoff between technical com- munication targeted at their immediate peers and more accessible language meant to reach broader audiences. As knowledge accumulates and research at the frontier builds on deeper layers of prior work, this tradeoff increasingly favors specialized language, limiting diffusion. Policy interventions that align scientists’ incentives can broaden adoption and increase the social value of scientific research.

H/t Tyler Cowen.

TO WAR! Part 1: War and America's National Psyche

Once again I"m bumping this to the top of the queue. And for obvious reasons. Once again a foolhardy President has taken the country into a war with no obvious end in sight, just a squandering of human live and productive resources.

* * * * * 

As part of a special WAR EDITION of New Savanna I'm reproducing a set of notes I wrote up during the 2000 Presidential Election. I first published them on NS back in November of 2011 and I'm republishing them now in recognition of yet another turn in the long-spinning wheel of American mythology. Yet another bump to the top, as I'm thinking of this stuff, this time in connection with the relationship between war and nationalism.

Everything is connected to everything else and the causal forces meeting in the historical present stretch back into the past without end. Figuring out where to start is not easy. My sense is that we need to focus our attention on the dissolution of the Soviet Empire in the late 1980s. That left the nation without a national scapegoat, thus radically altering the nation’s psycho-cultural landscape. We no longer had Ronald Reagan’s Evil Empire to kick around.

As some of you may know, my thinking on these matters has been strongly influenced by an essay Talcott Parsons published in 1947 on “Certain Primary Sources of Aggression in the Social Structure of the Western World”. Parsons argued that Western child-rearing practices generate a great deal of insecurity and anxiety at the core of personality structure. This creates an adult who has a great deal of trouble dealing with aggression and is prone to scapegoating. Inevitably, there are lots of aggressive impulses which cannot be followed out. They must be repressed. Ethnic scapegoating is one way to relieve the pressure of this repressed aggression. That, Parsons argued, is why the Western world is flush with nationalistic and ethnic antipathy. I suspect, in fact, that this dynamic is inherent in nationalism as a psycho-cultural phenomenon.

For the most part I have used Parsons, and others as well, in arguing about the nature of racism in the USA. While Africans were brought to this country for economic reasons it seems to me that during, say, the 19th century African Americans increasingly assumed a dual psychological role in the white psyche. On the one hand, they were a source of entertainment. On the other, they were convenient scapegoats, as became evident with the lynchings that emerged during Reconstruction and continued well into the last century. That is to say, African America served as a geographically internal target for the ethnic and nationalist antipathy Parsons discussed.

Thus we have the thesis in Klinkner and Smith, The Unsteady March (U. Chicago, 1999). They argue that African Americans have been able to move forward on civil rights only during periods where the nation faced an external threat - the Revolutionary War, the Civil War, and the major wars of the first half of the 20th century. When the external danger had subsided, gains were lost. From my point of view, they’re arguing that, when external danger looms large and demands attention, the citizenry can focus aggression there and so ease up on the internal colony. Beyond this, of course, it becomes necessary to recruit from the colony to fight the external enemy, both physically and propagandistically - be kind to your black citizens when you fight the Nazis, etc.

  

Vietnam was the last major war of the Cold War period. As it receded into the past, a political backlash set in and affirmative action came under attack. That’s the situation we faced when the Soviet Empire collapsed. With the major external threat suddenly collapsed, there was a crisis of aggression - I’m reminded of the phrase “conservation of aggression” coined by Robert Wright. The fall of the Evil Empire deprived a great many people of an object for aggressive impulses. What then, happened to that aggression?

It got directed elsewhere. My sense is that the political rhetoric on a number of issues heated up in the wake of the fall: gun control, abortion, the arts, gays, affirmative action, violence in the media. A number of these issues come under the rubric of the so-called “culture wars”. Each of these issues was already on the political agenda, and had been there for some time.

Sexy music had been inspiring pulpit denunciations and legislative action since the early decades of the 20th century. Movies have been problematic since the beginning and the NAACP put itself on the political map by organizing protests against “Birth of a Nation.” But, it seems to me, that the scope of politicized cultural contest broadened. [If I’m correct, then this could be verified empirically by doing content analysis of periodicals, looking at opinion polls, crime rates, etc.]

Perhaps the most interesting redirection, however, was into the so-called War on Drugs. Political concern about drug use is not, of course, new. It goes back to Prohibition - which, was, of course, intimately linked with that objectionably sexy music - and got redirected by and in reaction to the counter-cultural 60s and 70s. However, it is my impression that the current effort ramped up in the wake of the Soviet collapse.

This war on drugs has had substantial material consequences: increased law enforcement and court activity, a considerable increase in the prison population and, of course, in the prison industry. Our prisons now have a relatively large population of non-violent offenders who are disproportionately black, taken off the voting rolls as felons, and available for labor in various prison-based enterprises. I do not know whether or not the increase in the economic “weight” of the prison sector is roughly equal to the losses suffered by the defense sector. I would, of course, like to know.

Regardless of how those numbers work out, my basic point is simply that the end of the Cold War changed the psycho-cultural system in a major way. Psycho-cultural aggression had to be redirected and much of it was redirected at targets within the country, rather than externally. That redirection is the central political phenomenon of the 90s and is responsible for much of the ugliness and programmatic futility of current politics.

Addendum 2011: Obviously, this internal redirection of aggression continues. It may well have gotten worse – I wonder what ‘sentiment analysis’ would turn up if applied to documents written over the past decade? The problem facing us, of course, is not that of finding suitable targets for this displaced aggression. No, we have to restructure our way of living so as to reduce our need to have such targets. That’s a tough one, and one that will take a generation or three.

Addendum 2014: Obviously the Arab world is a perfect target for this repressed aggression. We've got a long-standing and well-crafted Orientalist mythology of the exotic and crazy Arab Other. I fear we're going to be banging on this nail for a long time.

 

Note 2026: The second post in this series has a clip from the Marx Brothers movie, Duck Soup, that can be taken as a comment on how Trump arrived at the decision to attack Iran: TO WAR! Part 2: A Marx Brothers Analysis of America's War Craziness.  

Howard’s Jam II: Go with the flow nation

I’m bumping this to the top of the queue on general principle and because the next edition of Howard’s jam is coming up this May 17th. You might want to check out this earlier post: Knockin’ on Heaven’s Door (at Howard’s Party).

Jamming has come up over at Morton’s joint, so I thought I’d say a little more about how things work at Howard’s Annual Birthday Bash and All-Hoboken Jam Session.

First let me reiterate that Howard’s been living in Hoboken for some time, 25 or more years I believe. He’s been active in the Hoboken music scene for at least that long. He knows a lot of people and, wouldn’t you know, many of them are musicians. Thus the people who attend this bash know one another, some better than others, naturally.

For the past three or four years Howard and a few others have had a jam session that would met every two or three weeks, usually at Howard’s place. That group has been fairly stable. Howard plays guitar and sings. So does Pete. Doug – who’s played with Howard longer than anyone else in the current group – plays bass; he sings as well. Pavel plays drums and is a Russian working in the USA and living here with his family. Vinnie plays harp (that is, harmonica), though he wasn’t able to make it to the party. I play trumpet and bells (cowbell, agogo).

We perform every once in awhile, but mostly we just jam. Over the years we’ve developed a repertoire of most rock and folk tunes, including a few turns written by members of the jam (Howard, Pete, me). While we do have routines on tunes we play regularly, they aren’t particularly rigid. And there are tunes were the idea is take the jam wherever the moment leads us. These tend to be Grateful Dead tunes; Fire on the Mountain is a favorite. And sometimes stuff just happens.

That’s how things began last Saturday, the night of the party. The front room on the first floor had been designated and set up as the music room. There was a drum set, microphones and speakers, amplifiers, and a keyboard. Sometime between 8:30 and 9:00 PM us core jammers were all there, as were other people, milling about. Howard started playing a simple two-chord riff on his guitar. Pete and Doug joined in – I forget in which order. Nor do I remember when Pavel started in on drums.

 

But there they were, noodling away. People were entering through the front door, walking through the room to the kitchen (& the food and the liquor). Talking, munching, milling. And I was sitting on the couch in the front room, oiling the valves on my trumpet, setting out my bells.

Are they just messing around? Is that a tune they’re playing? Sounds like SOMETHING but I don’t recognize it. What should I do?

I joined in. Just made up something that seemed to fit. And we played whatever it was for awhile and then stopped. Then, I believe, we played one of the tunes we’d been working on. But I forget which one.

And then another. More people came in. Some of them had their instruments. Which they got out, one by one. Got ‘em out, set ‘em up. Joined in. Pretty soon we were playing this or that song that wasn’t part of the regular repertoire. But someone knew it, otherwise we wouldn’t have been playing it, right?

The crucial thing is that WE was no longer the core jam group, the five guys I mentioned up there in the third paragraph. Other people were playing – guitar, keyboard, electric fiddle, and various shakers and rattlers – and singing. Oh yes, the singing.

Somewhere in there three women sitting on a wide bench started doing routines – synchronized hand gestures, leaning this way and that, just like Motown groups did. Who planned that? No one.

That’s how these things go. Here and there around the room other people were playing bells and rattles. Lots of things happen. No one plans anything. Well, not much.

Pretty soon the women started doing simple kicks, while remaining seated. Of course. Before you knew it, Guitar Dave (that’s how I think of him) was wailing away on his acoustic and joined them in their kicks. About this time I was into some kind of trumpet solo, and I – from my seat on the couch – also join up with the kick brigade. Which was ‘kinda awkward, and must have looked ridiculous. But who cares?

It was fun.

Somewhere each and everyone of us from the core group put down our instruments, went into the kitchen, and got some food, and some liquor. And other people filled in on the music. We weren’t actually necessary since there were many musicians in the crowd. They really didn’t need us. The real point, I suppose, is that there is no firm distinction between “they” and “us.”

Somewhere during the evening, this happened (I’m copying a note that Doug had posted to Facebook):

Group mind can be used for good. For ex: late Sat. @ Howard’s. The trumpet player (you) were playing repeating sixteenths on A. I picked it up on bass. Then - then we got on the same page and walked it up the progression, driving the previously listless vamp up a notch. The jam took off. All of us listening and hearing and doing: group mind - used for good. Another reason to adore making music.

You listen, you follow along, you play, someone follows you, and so it went. All night long. It worked because we were comfortable with one another. And because we shared a wide repertoire of songs and riffs. Whatever happened, you could somehow fit in and go with that particular flow.

As for that opening piece that we did, it turns out it wasn’t a tune that anyone knew. When Howard started playing that riff, he was just playing that riff. Getting the fingers loosened up. Checking the feel of the guitar, the sound of the room. We jammed it into, well, a tune.

Now, can we reconstruct what we did, and do it again? Who knows? Maybe, yes. Maybe, no. Something from nothing.

* * * * *

That guy in that picture up there. That’s Howard, though not on this occasion. That was last summer and an outdoor concert by a group he plays in, The Gordy’s. This guy here, that’s Guitar Dave:

And the people in the next photo, they’re members of The Gordys too. And they’re doing something silly. I bet they didn’t plan that routine:

As the concert moved along, people got up and started dancing. Like this:

The blonde in front is Howard’s mother. The woman behind her is Howard’s wife. Howard’s somewhere off to the left, playing his guitar.

For the next photo I think I was standing more or less behind the band and looking into the audience. But how do you draw a line between the band and the audience?

See that red-haired woman to the left of center? She was at the party Saturday night. She was one of the women doing the Motown routines.

These guys are Howard’s sons:

And that’s the moon over Manhattan:

Understanding transformers and communication among sperm whales

Mar 5, 2025
Allen School Colloquium Series Title: Discovering & Engineering the Computation Underlying Large Intelligent Agents
Speaker: Pratyusha Sharma (MIT)
Date: March 3, 2025

Abstract: The richness of language and intelligent behavior has often been attributed to latent compositional structure. Can we build tools for discovering how deep networks learn and represent this latent structure implicitly? And more importantly, can we use this knowledge to improve generalization in largely structure-less general purpose models or refine our understanding of the world they describe? In this talk, I present three perspectives to answer these questions. I will discuss experimental methods to functionally characterize the space of learnt solutions in LLMs and demonstrate how this understanding can be used to improve their empirical generalization in a gradient free manner, sometimes by as much as 30% points on language understanding benchmarks. Following that, I show how to decipher the structure of another (black box) language-like system, the naturally arising communication system of sperm whales in the wild, discovering for the first time a unique combinatorial communication system. Finally, I apply insights from these results to equip embodied agents with a latent language of thought–-hierarchical and compositional — and show how it can enable long-horizon reasoning and planning in these systems.

Bio: Pratyusha Sharma is a Ph.D. student in the Computer Science and Artificial Intelligence Lab at MIT, advised by Antonio Torralba and Jacob Andreas. She studies the interplay between language, sequential decision making and intelligence in natural and AI systems. Before this, she received her Bachelor’s degree from the Indian Institute of Technology, Delhi. Her research is published in interdisciplinary journals like Nature Communications, etc. and in academic conferences across machine learning, natural language processing, robotics, and marine biology. Her research has also been featured in articles in the New York Times, National Geographic Magazine, BBC, etc. She was recently a speaker at TED AI and was selected as a Rising Star in EECS, Data Science, and GenAI.

Her webpage, with links to her papers.

The most rapid change in human written communication ever?

The past 18 months have seen the most rapid change in human written communication ever

By. September 2024, 18% of financial consumer complaints, 24% of press releases, 15% of job postings & 14% of UN press releases showed signs of LLM writing. And the method undercounts true use pic.twitter.com/V1N6wKgTKZ

— Ethan Mollick (@emollick) March 1, 2025

Download the paper here.

Communication stripped to the barest minimum?

Hays, D. G. (1973). "Language and Interpersonal Relationships." Daedalus 102(3): 203-216. The following passage is from pp. 204-205:

The experiment strips conversation down to its barest essentials by depriving the subject of all language except for two pushbuttons and two lights, and by suggesting to him that he is attempting to reach an accord with a mere machine. We brought two students into our building through different doors and led them separately to adjoining rooms. We told each that he was working with a machine, and showed him lights and pushbuttons. Over and over again, at a signal, he would press one or the other of the two buttons, and then one of two lights would come on. If the light that appeared corresponded to the button he pressed, he was right; otherwise, wrong. The students faced identical displays, but their feedback was reversed: if student A pressed the red button, then a moment later student B would see the red light go on, and if student B pressed the red button, then student A would see the red light. On any trial, therefore, if the two students pressed matching buttons they would both be correct, and if they chose opposite buttons they would both be wrong.

We used a few pairs of RAND mathematicians; but they would quickly settle on one color, say red, and choose it every time. Always correct, they soon grew bored. The students began with difficulty, but after enough experience they would generally hit on something. Some, like the mathematicians, chose one color and stuck with it. Some chose simple alternations (red-green-red-green). Some chose double alternations (red-red-green-green). Some adopted more complex patterns (four red, four green, four red, four green, sixteen mixed and mostly incorrect, then repeat). The students, although they were sometimes wrong, were rarely bored. They were busy figuring out the complex patterns of the machine.

But where did the patterns come from? Although neither student knew it, they arose out of the interaction of two students.

Language is primarily for communication, not thinking

Fedorenko, E., Piantadosi, S.T. & Gibson, E.A.F. Language is primarily a tool for communication rather than thought. Nature 630, 575–586 (2024). https://doi.org/10.1038/s41586-024-07522-w

Abstract: Language is a defining characteristic of our species, but the function, or functions, that it serves has been debated for centuries. Here we bring recent evidence from neuroscience and allied disciplines to argue that in modern humans, language is a tool for communication, contrary to a prominent view that we use language for thinking. We begin by introducing the brain network that supports linguistic ability in humans. We then review evidence for a double dissociation between language and thought, and discuss several properties of language that suggest that it is optimized for communication. We conclude that although the emergence of language has unquestionably transformed human culture, language does not appear to be a prerequisite for complex thought, including symbolic thought. Instead, language is a powerful tool for the transmission of cultural knowledge; it plausibly co-evolved with our thinking and reasoning capacities, and only reflects, rather than gives rise to, the signature sophistication of human cognition.

Carl Zimmer has an article about this in the NYTimes. After giving some background, specifically noting we have a language network in the brain that's distinct from other networks:

The researchers then scanned the same people as they performed different kinds of thinking, such as solving a puzzle. “Other regions in the brain are working really hard when you’re doing all these forms of thinking,” she said. But the language networks stayed quiet. “It became clear that none of those things seem to engage language circuits,” she said.

In a paper published Wednesday in Nature, Dr. Fedorenko and her colleagues argued that studies of people with brain injuries point to the same conclusion.

Strokes and other forms of brain damage can wipe out the language network, leaving people struggling to process words and grammar, a condition known as aphasia. But scientists have discovered that people can still do algebra and play chess even with aphasia. In experiments, people with aphasia can look at two numbers — 123 and 321, say — and recognize that, by using the same pattern, 456 should be followed by 654.

If language is not essential for thought, then what is language for? Communication, Dr. Fedorenko and her colleagues argue. Dr. Chomsky and other researchers have rejected that idea, pointing out the ambiguity of words and the difficulty of expressing our intuitions out loud. “The system is not well designed in many functional respects,” Dr. Chomsky once said.

Zimmer's article links to a number of other studies.

If language is for communication, what does that imply about LLMs?

Noam Chomsky famously believes that language originated to facilitate thought, but then came to be a medium of communication. Others believe the reverse, that it originated as a facility for communication which turned out to facilitate thinking. That is certainly my view.

If that is so, then one would think that language is structured to facilitate communication. Communication is serial, one token at a time, token after token. That would imply that language is structured to facilitate next-token prediction. That would in-turn imply that the relational structure of semantics would evolve to facilitate mapping between the linear structure of the language string and the multidimensional structure of meaning. You want to be able to efficiently project multidimensional semantic structure onto a string and to reconstruction multidimensional semantic structure from a string.

How are LLMs trained? By next token prediction. That is to say, the training regime mirrors the primary communication constraint governing the structure of language. So it is with text generation as well. Language is spoken one token at a time, and so LLMs generate texts, one token at a time.

The tasks that LLMs have trouble with, such as planning and arithmetic, ARE NOT primarily communicative in nature. They are tasks for thought, for reasoning.

Whale vocalization, phonemes? (was Disney there first?)

Carl Zimmer, Scientists Find an ‘Alphabet’ in Whale Songs, NYTimes, May 7, 2024.

Ever since the discovery of whale songs almost 60 years ago, scientists have been trying to decipher their lyrics. Are the animals producing complex messages akin to human language? Or sharing simpler pieces of information, like dancing bees do? Or are they communicating something else we don’t yet understand?

In 2020, a team of marine biologists and computer scientists joined forces to analyze the click-clacking songs of sperm whales, the gray, block-shaped leviathans that swim in most of the world’s oceans. On Tuesday, the scientists reported that the whales use a much richer set of sounds than previously known, which they called a “sperm whale phonetic alphabet.”

People have a pho-ne-tic alphabet too, which we use to produce a practically infinite supply of words. But Shane Gero, a marine biologist at Carleton University in Ottawa and an author of the study, said it’s unclear whether sperm whales similarly turn their phonetic sounds into a language.

“The fundamental similarities that we do find are really fascinating,” Dr. Gero said. “It’s totally changed the way we have to do work going forward.”

Clicks, not melodies:

Sperm whales don’t produce the eerie melodies sung by humpback whales, which became a sensation in the 1960s. Instead, they rattle off clicks that sound like a cross between Morse code and a creaking door. Sperm whales typically produce pulses of between three and 40 clicks, known as codas. They usually sing these codas while swimming together, raising the possibility that they’re communicating with one another.

ChatMOBY-DICK?

Microphones deployed in the Caribbean are capturing ocean sounds 24 hours a day, and scientists are programming computers to learn how to pick sperm whale songs out from the background noise.

Dr. Andreas and his colleagues are also training artificial intelligence programs similar to ChatGPT. After listening to the sperm whale songs, these models might learn to recognize not just rubato and ornamentation, but other features that scientists have missed.

The hope is that computers will then be able to compose whale songs of their own, which could then be played to the whales.

That effort leaves other experts skeptical. Luke Rendell, a marine biologist at the University of St Andrews, in Scotland, worries that the A.I. models assume whale songs are a kind of language, rather than something more like music.

“I’ve no doubt that you could produce a language model that could learn to produce sperm-whale-like sequences,” Dr. Rendell said. “But that’s all you get.”

There's much more at the link.

Meanwhile, back in 1946 Disney release Willie the Operatic Whale. Here's a clip (watch it now, before Disney snatches it away):

Power, Persuasion, and Influence in The Valley

Henry Farrell has a new post at Crooked Timber: Dr. Pangloss’s Panopticon (Feb. 27, 2024). He's replying to Noah Smith's negative review of Acemoglu and Johnson, Power and Progress. 

On power:

What Acemoglu and Robinson are saying is something quite different than what Noah depicts them as saying. For sure, they acknowledge that persuasion has some stochasticity. But they stress that it is not a series of haphazard accidents. Instead, under their argument, there are some kinds of people who are systematically more likely to succeed in getting their views listened to than other kinds of people. This asymmetry can reasonably be considered to be an asymmetry of power.

Under this definition, power is a kind of social influence. Again, it is completely true that it is extremely difficult to isolate social influence from other factors, proving that social influence absolutely caused this, that, or the other thing. But if Noah himself does not believe in the importance and value of social influence, then why does he get up in the morning and fire up his keyboard to go out and influence people, and why do people support his living by reading him?

I imagine Noah would concede that social influence is a real thing! And if he were actually put to it, I think that he would also have to agree to a very plausible corollary: that on average he, Noah Smith, exerts more social influence than the modal punter argufying on the Internet. Lots of people pay to receive his newsletter; lots of other people receive it for free. That means that he is, under a very reasonable definition, more powerful than those other people. He is, on average, more capable of persuading large numbers of people of his beliefs than the modal reply-guy is going to be.

This understanding of power is neither purely semantic nor empirically useless. Again, it may be really difficult to prove that Noah’s social influence has specific causal consequences in a specific instance. But the counter-hypothesis – that Noah’s ability to change minds, given his umpteen followers, is the same as the modal Twitter reply guy – is absurd. Occasionally, random people on the Internet can be temporarily enormously influential. Sometimes, super prominent people aren’t particularly successful at getting their ideas to spread. But on average, the latter kind of people will have more influence than the former. We can reasonably anticipate that people with lots of clout (whether measured by absolute numbers of followers, numbers of elite followers, bridging position between sparsely connected communities or whatever – there are different, plausible measures of influence and lively empirical debates about which matters when) will on average be substantially more influential than those with little or none. This means, for example, that it will be very difficult for ideas or beliefs to spread if they are disliked by the highly connected elite.

Now in fairness to Noah, Acemoglu and Johnson don’t help their case by using a wishy-washy seeming term like “persuasion.” But if you think about “persuasion” as some combination of “social influence” and “agenda control,” you will get the empirical point they are trying to make.

Core claims:

Acemoglu and Johnson’s core claims, as I read them are:

1. That the debate about technology is dominated by techno-optimists [they actually write this before Andreessen’s ludicrous “techno-optimist manifesto” but they anticipate all its major points].
2. That this dominance can be traced back to the social influence and agenda setting power of a narrow elite of mostly very rich tech people, who have a lot of skin in the game.
3. That their influence, if left unchecked, will lead to a trajectory of technological development in which aforementioned very rich tech people likely get even richer, but where things become increasingly not-so-great for everyone else.
4. That the best way to find better and different technology trajectories, is to build on more diverse perspectives, opinions and interests than those of the self-appointed tech elite, through democracy and countervailing power.

Since I more or less endorse all these claims (I would slightly qualify Claim 1 to emphasize mutually reinforcing pathologies of tech optimism and tech pessimism), I think that Power and Progress is a really good book, in ways that you won’t understand if you just relied on Noah’s summary of it (I note that this book and my own with Abe Newman are both shortlisted for a very nice prize, but that is neither here nor there in my opinion of it). I haven’t read another book that lays out this broad line of argument so clearly or so well. I haven’t read another book that lays out this broad line of argument so clearly or so well. And it is a very important line of argument that is mostly missing from current debates. Noah speculates that the book hasn’t gotten much attention because it is lost amidst the multitudes of tech pessimistic accounts. My speculation is that it has gotten less attention than it deserves because reviewers and readers don’t know quite how to categorize it, given that it approaches the issues from an unexpected slant.

The panopticon:

The panopticon may indeed have efficiency benefits. People can get away with far less slacking, if it works as advertised. But it also comes with profound costs to human freedom. And the technologies that are at the heart of the book’s argument – machine learning and related algorithms – bear a strong and unfortunate resemblance to Bentham’s panopticon. They too, enable automated surveillance at scale, perhaps making hierarchy and intrusive surveillance much, much easier and cheaper than they used to be. As Acemoglu and Johnson note:

The situation is similarly dire for workers when new technologies focus on surveillance, as Jeremy Bentham’s panopticon intended. Better monitoring of workers may lead to some small improvements in productivity, but its main function is to extract more effort from workers and sometimes also reduce their pay

This is, I think, why Acemoglu and Johnson worry that machine learning might immiserate billions, another claim that Noah finds puzzling. Acemoglu and Johnson fear that it will not only remake the bargain between capital and labour, but radically empower authoritarians (I think they are partly wrong on this, but that authoritarian machine learning could instead lead to a different class of disasters: pick yer poison).

The post is longish, but excellent. I made the following comment, about power in Silicon Valley:

Excellent, Henry, at least as far as I got. I made it about half-way though before I just had to make a comment. I'm thinking about the piece you and Cosma Shalizi did about the culture of Doomerism, which is very much a Silicon Valley phenomenon. And is also very relevant to any discussion of ideas, influence, persuasion, and POWER. I was shocked when such a mainstream magazine as Time ran a (crazy-ass) op-ed by Eliezer Yudkowsky.

I am reasonably familiar with his work. I have several times attempted to read a long piece he published in 2007 about Logical Organization in General Intelligence. I've been unable to finish it. Why? Because it's not very good. It's the kind of thing a really bright and creative sophomore does when they've read a lot of stuff and decide to write it up. You read it, think the guy's bright, if he gets some discipline, he could do some very good work. Well, 2007 was awhile ago, but as far as I can tell, he still doesn't have much intellectual discipline and certainly doesn't have deep insight into current AI or into human intelligent. But Time Magazine gave him scarce space in their widely read magazine.

That's power. Now, as far as I know, he's not again been able to place his ideas in such a venue. But even once is pretty damn good.

How'd that come about? Well there's a story, one I don't know in detail. But the story certainly involves money from Silicon Valley billionaires. He's been funded by Elon Musk and, I believe, by Peter Thiel (who's since become disillusioned with some of those folks). There's a lot of money coming into and through the world centered around LessWrong (which, BTW, has the best community-style user interface I've seen) from tech billionaires.

On technological trajectories, Acemoglu is one of a team of writers of a 2021 report out of Harvard, How AI Fails Us. Here's the abstract:

The dominant vision of artificle intelligence imagines a future of large-scale autonomous systems outperforming humans in an increasing range of fields. This “actually existing AI” vision misconstrues intelligence as autonomous rather than social and relational. It is both unproductive and dangerous, optimizing for artificle metrics of human replication rather than for systemic augmentation, and tending to concentrate power, resources, and decision-making in an engineering elite. Alternative visions based on participating in and augmenting human creativity and cooperation have a long history and underlie many celebrated digital technologies such as personal computers and the internet. Researchers and funders should redirect focus from centralized autonomous general intelligence to a plurality of established and emerging approaches that extend cooperative and augmentative traditions as seen in successes such as Taiwan’s digital democracy project and collective intelligence platforms like Wikipedia. We conclude with a concrete set of recommendations and a survey of alternative traditions.

That is much better than the view that dominates AI development today. Moreover, I believe it to be more technically feasible. But, despite the Harvard imprimatur, it doesn't have nearly as much power behind it as the Silicon Valley view of "a future of large-scale autonomous systems outperforming humans in an increasing range of fields."

Busy Bees

See "Busy Bee Brain."

Moby Dick speaks (has vowels)

Previously it was thought that mostly the number of clicks and their timing matters.

We identify two discrete patterns that we call the a-vowel and the i-vowel. We also observe clear trajectories in formant frequency that we call diphthongs. pic.twitter.com/2zPrKE193f

— Gašper Beguš (@begusgasper) December 5, 2023

The study of rhetoric as a route to cognitive empathy?

John Bowe, An Ancient Solution to Our Current Crisis of Disconnection, NYTimes, Nov. 27, 2023.

While rhetoric had its detractors, starting with Plato, Bowe points out that it was

the cornerstone of education until the 1700s.

Across Western Europe, students from about the age of 12 onward learned logic, social skills, critical thinking and speech techniques as a single, integrated discipline by means of a 14-step verbal and cognitive curriculum known as the progymnasmata.

Exercises began with simple recitations and enactments of fables and short stories. Later drills trained students to compose and deliver short speeches of praise and blame and, eventually, long discourses on complex themes. By writing with the intent of performing for others (rather than writing objectively for the page), students learned the art of blending fact with opinion. By mastering the techniques of persuasion, students became proficient at spotting others’ manipulative use of language.

Bowe goes on to argue:

My interest in rhetoric began in 2010, during a chat with my extremely reclusive Iowa step-cousin. He’d lived alone until the age of 60 in his parents’ basement with no friends, no girlfriends, then surprised the entire family by meeting someone and getting married. I asked him how he’d mustered up the courage to approach his future wife, given the depths of his isolation. “I joined the Toastmasters,” he said, referring to what is likely the world’s largest organization devoted to teaching public speaking. He’d never seen a therapist or taken meds. One or two dozen hours of speech training changed his entire life.

I’ve since learned that this is what speech training does. When speakers put themselves in their listener’s place, they find it easier to explain themselves. The confidence that we can make ourselves known and understood is transformative.

Apparently, scientists agree. Hannah Hobson, a lecturer in psychology at the University of York who has studied the connections among language, communication and mental health, especially among neurodiverse youth, has found repeatedly that the inability to express feelings or ask for help can often correlate with existing or developing mental health issues among youth. Conversely, she told me, improved communication skills correlate with youngsters’ emotional development and mental well-being.

Bowe doesn't elaborate on that part, putting yourself in your listener's place, but that's also called cognitive empathy, something Robert Wright is interested in. He even believes that it can save the world. Well, it's more like he believes that it's necessary, though likely not sufficient.

Interpersonal synch is stronger face-to-face than in video chat

Face-to-face interaction elicits more inter-brain synchronization than video-chat interaction. Moreover, associations with behavior emerge only during f2f interaction — with right temporal-temporal synchrony linked to shared gaze and right frontal-frontal to engagement & empathy. pic.twitter.com/oIQkgOJkVX

— Guillaume Dumas (@introspection) October 16, 2022

Thanks and kudos to my amazing collaborators @Lin_Scw @yoniphd @AmirDJV @YaaraShapira and of course @ruthfeldmanlab 🙌 Also grateful for the continuous support of @IVADO_Qc @FRQS1 @Med_UMontreal @CR_CHUSJ 🙏

— Guillaume Dumas (@introspection) October 16, 2022

Investigating entrainment of spontaneous activity of brain and body physiology between subjects [synch!]

Our new paper:
Systemic physiology augmented functional near-infrared spectroscopy hyperscanning: a first evaluation investigating entrainment of spontaneous activity of brain and body physiology between subjectshttps://t.co/6TxmRhbRFV#fNIRS @SfNIRS #hyperscanning #neuroscience pic.twitter.com/0OuUyayexN

— Felix Scholkmann (@FScholkmann) May 5, 2022

From the linked paper:

Abstract

Significance: Functional near-infrared spectroscopy (fNIRS) enables measuring the brain activity of two subjects while they interact, i.e., the hyperscanning approach.

Aim: In our exploratory study, we extended classical fNIRS hyperscanning by adding systemic physiological measures to obtain systemic physiology augmented fNIRS (SPA-fNIRS) hyperscanning while blocking and not blocking the visual communication between the subjects. This approach enables access brain-to-brain, brain-to-body, and body-to-body coupling between the subjects simultaneously.

Approach: Twenty-four pairs of subjects participated in the experiment. The paradigm consisted of two subjects that sat in front of each other and had their eyes closed for 10 min, followed by a phase of 10 min where they made eye contact. Brain and body activity was measured continuously by SPA-fNIRS.

Results: Our study shows that making eye contact for a prolonged time causes significant changes in brain-to-brain, brain-to-body, and body-to-body coupling, indicating that eye contact is followed by entrainment of the physiology between subjects. Subjects that knew each other generally showed a larger trend to change between the two conditions.

Conclusions: The main point of this study is to introduce a new framework to investigate brain-to-brain, body-to-body, and brain-to-body coupling through a simple social experimental paradigm. The study revealed that eye contact leads to significant synchronization of spontaneous activity of the brain and body physiology. Our study is the first that employed the SPA-fNIRS approach and showed its usefulness to investigate complex interpersonal physiological changes.

You think it, the computer writes it (sorta' – it's NOT reading your thoughts, rather, your hand motions)

#AINews #AI #SwissCognitive

AI decodes neural signals to convert mental handwriting to text, enabling communication for patients with spinal cord injuries, strokes, and other conditions.https://t.co/JvkDceURDT

— SwissCognitive – The Global AI Hub (@SwissCognitive) May 26, 2021

Willett, F.R., Avansino, D.T., Hochberg, L.R. et al. High-performance brain-to-text communication via handwriting. Nature 593, 249–254 (2021). https://doi.org/10.1038/s41586-021-03506-2

Abstract: Brain–computer interfaces (BCIs) can restore communication to people who have lost the ability to move or speak. So far, a major focus of BCI research has been on restoring gross motor skills, such as reaching and grasping or point-and-click typing with a computer cursor. However, rapid sequences of highly dexterous behaviours, such as handwriting or touch typing, might enable faster rates of communication. Here we developed an intracortical BCI that decodes attempted handwriting movements from neural activity in the motor cortex and translates it to text in real time, using a recurrent neural network decoding approach. With this BCI, our study participant, whose hand was paralysed from spinal cord injury, achieved typing speeds of 90 characters per minute with 94.1% raw accuracy online, and greater than 99% accuracy offline with a general-purpose autocorrect. To our knowledge, these typing speeds exceed those reported for any other BCI, and are comparable to typical smartphone typing speeds of individuals in the age group of our participant (115 characters per minute). Finally, theoretical considerations explain why temporally complex movements, such as handwriting, may be fundamentally easier to decode than point-to-point movements. Our results open a new approach for BCIs and demonstrate the feasibility of accurately decoding rapid, dexterous movements years after paralysis.

How do we fact-check social media?

Avi Tuschman, Want to solve the misinformation crisis? We already have a proven solution at our fingertips, Fast Company, 4.8.21:

The most dystopian feature of our time is not that we face formidable challenges; in a different era we might have had enough shared beliefs to navigate a bloodless presidential transition, vaccine hesitancy, racial tension, and even climate change. Today, however, our body politic suffers from an informational infection that hinders our ability to adequately respond to these serious threats.

What do we do? We need a third entity:

How can we improve our information systems to save lives? As NYC Media Lab’s Steve Rosenbaum has pointed out, neither the tech platforms nor governments can be trusted fully to regulate the internet, “So it’s like we want this magical entity that isn’t the government, that isn’t Facebook or YouTube or Twitter.”

Rosenbaum is absolutely correct: solving the misinformation crisis requires a “magical” third entity that lacks any incentive to manipulate information for economic or political ends. However, it is the tech companies that could build a sufficiently fast and scalable system for distinguishing facts from falsehoods. Such a solution is not only theoretical—its key components are already well developed and proven.

For example, Wikipedia has become a trusted source of reliable information.

Social media platforms can leverage Wikipedia’s strengths to reduce their weakness. They must provide an open-source fact-checking space in their content-moderation systems. Doing so is not only an ethical responsibility; it’s also a smart move to get ahead of the regulatory hammer.

Here’s how it could work: A tiny percentage of social media content contains viral misinformation deleterious to public health. Tech companies should start by implementing policies to make such content eligible for open-source fact-checking. The platforms could use several mechanisms to pass suspect content to a distributed review process. Here, fact-checking users would utilize the same open-source software and mechanisms that have successfully evolved on Wikipedia to adjudicate verifiability. The “visible process” of fact-checking would occur on a MediaWiki, ideally governed by a multi-stakeholder organization. The facts themselves—the “ground truth”—would be English-language Wikipedia text, from articles that meet minimum authorship and editorship thresholds.

There’s more at the link.

But what about telepathic communication? [more on the incoherence of brain2brain thought transfer]

Think of this as a pendant on and further exploration of my ongoing discussion of the incoherence of the idea of direct brain-to-brain thought transfer.

I’ve been watching Start Trek: Enterprise. There’s an episode in season 4, #14 The Aenar, where we meet a species capabable of telepathic communication with one another. That seems to be a minor theme in the Star Trek universe. Of course we have the Vulcan Mind Meld, which made its appearance with Spock in the original series and shows up a few time in Enterprise, but it seems to be assymetric; the party who initiates the meld can read the other’s mind, and even place thoughts in it, but not vice versa. Then we have the telepatic conversations between Deanna Troi and her mother in The Next Generation.

These conversations are presented to us as that, verbal interchanges; except that the parties don’t move their lips and may not even be in one another’s presence. The voices are simply in voiceover; we are to understand that no one hears anything, that this is telepathic?

Is that how telepathy would work? I’m not suggesting that the makes of these programs are seriously proposing telepathic communication as a human possibility any more than they are proposing warp drive of subspace communication as serious possibities. It’s pure fiction, not a real (in some way or other) proposal such as those Llinás, Koch, and Musk (seem to) have made about brain-to-brain communication. If telepathy were possible, is this how it would go, silent conversation? After all, there are silent conversations taking place all around us on cell phones and the like. We don’t hear them because they’re being carried by radio waves, which we cannot hear. Would that be possible with telepathic waves?

My first reaction – and this is important – upon posing the question (last night) is, sure. There’d have to be some physical means of sending and receiving telepathic waves but, given then, such silent conversation would be possible. After I’d thought about that for awhile, though, I began reconsidering.

Just what signal is being sent and received? The waves that make up real conversation arise when we send neural impulses to various muscles – in the tongue, jaw, truck, etc. – to drive vocalization. They are then picked up in the ear, which transduces them into neural impulses. But none of that would happen in the hypothesized telepathic communication. What gets telepathically broadcast? Perhaps the motor impulses. How would the other party interpret them, with the auditory cortex? Maybe what gets sent and received is some modally neutral signal, neither motor nor auditory? But of course, if telepathic communication were real, brains and nervous systems would have evolved accordingly. In addition to motoric sending and auditory receiving, there’d be telepathic sending and receiving.

What is it that would be telephatically send and received? Would it be telepathic word forms, that is signifiers, or would the actual meanings be sent (the signifieds)? I think it would have to be the signifiers, and only the signifieds. Why, because word meanings aren’t going to be neat little informatic packets like word forms. That is, the existence of words as discrete entities invites us to believe that word meaning are gathered up in discrete packets of meaning. But there is no reason [that is, beyond our desire that things be so simple] to think things work like that.

Rather word meanings reside in neural nets where the meaning is a function of relationships in the net. Word meaning is not separable from the net itself. That, alas, is not something that is easily explained in a blog post or, for that matter, at all. At this point I think you have to actively work with semantic or cognitve network models to get a sense of what’s going on. Sydney Lamb exaplains that in Pathways to the Brain (1999) and I say a bit more in my post, 2. The brain, the mind, and GPT-3: Dimensions and conceptual spaces (August 2, 2020).]

As a crude analogy, imagine a pond as the repository of meaning. Word forms (signifiers) are pebbles dropped into the pond. When a pebble hits the pond’s surface ripples travel outward from the point. As successive pebbles hit the service each of them sends ripples as well and the ripples interact with one another. It’s those interactions of the ripples that embodies the meaning of the word stream (that is, the succession of pebbles).

At this point we seem to have lost the thread of our original inquiry, about telepathic communication. That’s how it goes. Thinking these things through is tough and you cannot predict where things will go.

On the incoherence of the idea of brain-to-brain thought transfer [once again]

Once again I find myself thinking about the (im)plausibility of direct brain-to-brain communication of thoughts, most recently hyped by Elon Musk in conjunction with his Neuralink project. What I’m wondering is why Musk hasn’t (seemed to have) thought about the most obvious objections to the project, and not only Musk, but others who have proposed such linkage (Christof Koch and Rodolfo Llinás). I’ve addressed that issue once before (Brain-to-brain thought transfer @ 3QD) but I’d like to take another shot at it.

Let’s start by observing that the ideas of “thoughts” and “thinking” are commonsense ideas and have not been defined in neural terms. There is thus a conceptual GAP between them and the terms used talk about brains, even in the most casual way. To connect the two realms one must think it through. And just what does that mean? What items in each realm must be considered?

It’s all well a good to say that thoughts and thinking exist in the brain, but that assertion, with which I occur, doesn’t tell us how those things exist in the brain. How many neurons does it take to support a single thought – one, 10, 89, 3000, more? Do they have to be in the same region of the brain or can they be spread out? If so, how far, a hemisphere, the whole brain? How do thoughts flow through axons? Can a whole thought be squeezed through a single axon? These questions may seem (faintly) ridiculous, but are they? What is clear is they aren’t the terms in which neuroscientists investigate the brain.

And that’s fine, really. I have no problem with that. But it does mean that the relationship between thought/thinking and neurons is left undefined. Talk of brain-to-brain linkage, however, takes place in terms of neurons and brain tissue. Such talk tends to be rather loose and vague, though talk of more modest neural interfaces (which, for example, allow for neural control of prosthetic limbs) is quite precise; such things, after all, have actually been built.

Here’s what I suspect is going on. When these people – Musk, Koch, Llinás – think about the issue they start with the existence of various technology connecting brains with outside devices of one kind or another. These exist in one form or another. They then imagine having a whole lot of them (Koch writes of 10s of millions) running between two brains. That sets up (some kind of) a communication channel between the brains. They then apply what cognitive scientists call the conduit metaphor and, viola! direct brain-to-brain thought transfer. Of course they don’t explicitly think, “and now we apply the conduit metaphor” – that’s not how these things go, is it? – they just do it.

What is the conduit metaphor? (See this post for a more careful explanation of it.) It is the idea that we communicate with one another through some kind of conduit, often imaginary. After all, when we converse – surely the prototypical case of human communication – there is nothing between us but the air. Note, though, that all we exchange directly in communication are word forms. The ideas aren’t themselves in those word forms. Rather, ideas are associated with them in the processes of comprehension or production. But those ideas exist only in our heads (minds, brains), not in the signals.

When we apply the conduit metaphor to this artificially constructed brain-to-brain communication channel, thought transfer is automatic. Note in particular that the problem I identified – in this post at 3 Quarks Daily and earlier at this post here at New Savanna – is that a brain has no way of telling whether a neural impulse comes from within itself or from another brain, or, for that matter, an investigator or surgeon zapping the brain with an electrical current. Regardless of where it comes from, an impulse is an impulse is an impulse. They’re all alike. If neural impulses can’t be separated into mine and thine, then how can there be thought transfer?

But – and here’s the important point – this problem doesn’t arise in ordinary speech communication. You know who you’re talking to and there’s no trouble distinguishing their words from yours. Thus there is nothing in conduit metaphor to tell you to check who’s sending a given signal. The conduit metaphor simply isn’t rich enough to handle the problem of direct brain-to-brain communication.

What has Elon Musk been smoking? No, Elon, you're NOT developing technology that will make language obsolete.

In a recent conversation with Joe Rogan Elon Musk said his company, Neuralink, is developing technology that may well make language obsolete. Here's how the Independent excerpted those remarks:

“You wouldn’t need to talk,” Musk said, adding: “We could still do it for sentimental reasons.”

He went on: “You would be able to communicate very quickly and with far more precision … I’m not sure what would happen to language. In a situation like this it would kind of be like The Matrix. You wanna speak in a different language? No problem, just download the program.”

Asked how long it would take for his firm to develop technology advanced enough to do it, Musk said between five and 10 years, “if the development continues to accelerate”.

Color me skeptical.

I've talked about that sort of thing in an old post, Why we'll never be able to build technology for Direct Brain-to-Brain Communication. This link will take you to a number of posts about technology for brain-to-brain communication; Musk is hardly the only one working on it.