Time Well Spent - 12/12/2023

Holiday Edition

For the first time ever, parents going through IVF can use whole genome sequencing to screen their embryos for hundreds of conditions. Harness the power of genetics to keep your family safe, with Orchid. Check them out at orchidhealth.com.

---

Your time is finite. Your phone and the internet stand ready to help you squander it. Here are my latest picks for spending it well instead. Feel free to add more in the comments. 

Books, what else?

The last year in technology circles has been the “Year of AI.” OpenAI’s November 30th, 2022 release of ChatGPT triggered a fresh debate about the possible imminent emergence of “Artificial General Intelligence” (AGI) and the existential risk to humankind that might entail, along with the untold economic possibilities. Companies like Google, Facebook and Amazon went into high gear, hustling to catch up to OpenAI, an upstart partnering with Microsoft. Almost a year later OpenAI saw turmoil, as CEO Sam Altman was fired and rehired within ‌the space of one rollercoaster week. This was a massive business story because while the AI sector is worth $150 billion today, by 2030 it is projected to be a $1.3 trillion dollar industry (in 2023 in purchasing-power-parity terms, the whole American economy is worth $25 trillion dollars). Over time, AI will eat the world, embedding itself within the computing systems of the modern information economy just as the internet has.

The past year’s AI revolution has had a major cultural impact, with reactions dividing people into three broad camps. The first, largest group tends to view it as just another technology, like the internet, telephone or railway. This camp is sanguine about AI because they do not consider AGI, whatever that will look like, a major likelihood or threat. Instead they class AI as just another utility that will alter economic relationships and improve the technological matrix of our world—a mechanical step forward, not a daring leap into some dark philosophical unknown. Most of the public falls into this camp,  and, well, arguably most computer scientists as well.

The two other camps are in radical opposition to each other, though both take the existential implications of AI and the likelihood of AGI, deadly seriously. One faction believes that AGI is a grave existential threat to humankind, portending a species-level menace on the scale of Skynet in the Terminator film franchise. Viewed as alarmists by most people working in the trenches of AI technology, they are exemplified by figures like Eliezer Yudkowsky, a Bay Area-based rationalist and philosopher who argues that AGI’s emergence is the greatest danger on this planet to the human species. This position broadly aligns with many thinkers in the “effective altruism” (EA) movement, who believe that disruptions like AI or pandemics might negatively affect our well-being (or pose species-extinction risks). The last camp is the newest: “effective accelerationists” (e/acc), a more inchoate faction, with no true leader, but bracketing the range of technologists united in welcoming AI’s transformative possibilities. Moderate e/acc partisans believe AGI and machines might be able to be seamlessly integrated into our civilization and society to the benefit of humans, but e/acc zealots have an almost messianic optimism about the possibilities of AGI in their grandest imaginings, where thinking machines replace humanity. Rather than the end of humankind, e/acc sees in the new technology and AGI the possibility of the birth of something new, something better.

These last two groups will generally strike the average person as strange, but I think where they come from makes more sense when you consider how many people working in tech have read and internalized the narratives and presuppositions of science fiction. Yudkowsky himself is a science fiction author, a fact that some computer scientists who are skeptical of his apocalyptic claims haven’t hesitated to throw back in his face as an insult. But Yudkowsky counters that those familiar with the corpus of science fiction may be the best able to think broadly about the future. Being steeped in science fiction extends to the highest heights of the tech aristocracy; Elon Musk, who is aligned with the AI-skeptical EA faction, is famously well-versed in the Dune series, which is predicated on the battle between humankind and implacable machines bent on enslaving them. The unity between Musk’s avowed current concerns about the threats of AGI and the framework of the Dune universe is uncanny.

Science fiction is a genre that comes out of the pulp tradition. Its core audience are teen boys, so it has not tended to be the domain of great prose stylists or the most incisive observers of the current scene. It is instead a genre of ideas, strange and fantastical, and these ideas are now percolating out to wide cultural prominence thanks to the centrality of the tech sector in American society. 

Though the vast majority of science fiction books are forgettable, some influential few may lend key insight into the imagined worlds that have crucially shaped many of our most powerful technologists’ mindsets.

Vernor Vinge’s early 1990’s space opera A Fire Upon The Deep remains obscure to the general public but is widely read, known and discussed in futurist circles (it won the 1993 Hugo Prize for best novel in science fiction). A professor of computer science at San Diego State, Vinge imagined a far future where humanity traversed the galaxy in faster-than-light vessels, a common motif in the space opera genre. But A Fire Upon The Deep is notable for its evocation of a cosmos filled with minds of differing levels of complexity, where humans occupy a rung in the middle, between unicellular organisms and superintelligences. The great antagonist in Vinge’s novel is a malevolent superintelligence called the Blight. This being destroys and consumes other forms of life, and is a threat to the whole galaxy. A Fire Upon The Deep focuses on humans aided in their adventurous quest by other super-intelligences and similarly intelligent sentient dogs (yes, you read that right). Vinge's characterizations lacked depth, and the plot is a typical action-packed space opera. Still, his thickly plotted depiction of artificial intelligence at a peak level of sentience and power exhibits a verisimilitude that has haunted many technologists who encountered the book in their youth. Vinge was a speaker at the 2008 Singularity Summit, which I attended. At the time I noted that literally every attendee seemed to know who he was and most had read his novels. 

While Vinge's book may be somewhat obscure to the public, Frank Herbert’s Dune has earned widespread recognition, not least thanks to its adaptation into two feature-length treatments already, in 1984 and 2022 and a mini-series on cable television in 2000. Though the original Dune book is the best in a series that now runs to over a dozen works, it is the fourth book, God Emperor of Dune that most fully articulates the philosophical thread running through the series. The original Dune was an action-adventure centering around a teen. God Emperor of Dune revolves around the biographical and philosophical ruminations of the son of the first book's protagonist, with only sparse bursts of action to forward the plot. This son of Dune’s protagonist, Leto II, lived to be 3,500 years old by transforming himself into a monstrous post-human entity. Though initially depicted as cruel and ruthless, Leto II’s fundamental story arc is the culmination of humanity’s quest to escape the inevitable fate of slavery to “thinking machines,” AGI. Concepts barely hinted at in Dune are fully articulated in God Emperor, a novel centered on a megalomaniacal and messianic human who believes that the fate of the species is determined by the choices he makes.

Dune was published in 1965, so it is clear that Isaac Asimov's Foundation series prefigured much of the superstructure of Herbert’s universe fifteen years prior. But Asimov readily admitted that the rise and fall of his Galactic Empire and the collapse into a Dark Age were drawn from Gibbon’s Decline and Fall of the Roman Empire. His fully realized universe served as a template for many later space operas and influenced films like Star Wars (along with E. E. Smith’s Lensman series). Similar to Dune, Asimov’s Foundation wraps plots around a dense philosophical superstructure, addressing determinism in the historical process, sociopolitical decentralization’s importance to innovation and the inherent long-term fragility of imperial structures. Dune was an attempt to imagine the course of future history shaped by intuitive super-humans faced with the threat of computers. Foundation presents a more techno-optimist future where man and machine work together to deduce principles by which human civilization can perpetuate itself and minimize the down cycles of decline.

 A Fire Upon The Deep, God Emperor of Dune and Foundation paint us a grand and wondrous future, with super-human intelligence and glittering technologies studding the galaxy. Gene Wolf’s Book of the New Sun, a four-volume trilogy tetralogy, The Shadow of the Torturer, The Claw of the Conciliator, The Sword of Lictor and The Citadel of the Autarch, presents the inverse, a vision of the future past. Wolfe’s prose is rich and allusive, in sharp contrast to the straightforward and more pedestrian style common to Asimov, Vinge and Herbert, and his plots are serpentine and nonlinear. The future world explored in the tetralogy from the protagonist’s viewpoint is fallen, decayed and retrograde. Book of the New Sun depicts an ancient humanity, without hope, reverting back to primitive ways of life as the sun extinguishes itself, taking with it our species’ hopes and dreams. Wolfe’s series ranges across a universe where humanity turned away from the stars, dug roots deep into the earth, and aged along with the planet beneath the light of a fading sun. Its pessimism is a rebuke to those raised on galaxy-spanning space operas.

If Wolfe’s Book of the New Sun is a window upon a paradoxical and cryptic dying earth, Arthur C. Clarke’s The City and the Stars is a scientific investigation and cataloging of humanity’s end by exhaustion and ennui. Like God Emperor of Dune, Clarke’s novel explores the philosophical implications of super-intelligences and a quiescent and decadent humanity, but whereas Herbert takes an antagonistic stance towards AGI, Clarke’s humans in The City and the Stars have achieved a stable coexistence with thinking machines. The protagonist, Alvin, is one of the few “born humans” in his city, the last on Earth billions of years in the future, and his journey across the dying planet exposes the mysteries that the old earth retains. Though AGI is both a benevolent and malevolent feature in The City and the Stars, it is ultimately a sideshow to the primary questions our species still faces: stagnation and decadence. In The City and the Stars AGI did not solve any deep existential questions, because that is a challenge for humans alone.

Thought

How to unleash a trailer home boom - Eliminate the rules that deter small factory-building houses. Right now people in the Anglosphere think houses are stores of wealth like owning a rare painting. We must consider it more like digital photography; you want volume and quality. It’s functional—a place to live. If we want people to have larger families, we need more housing. We need to supplement and replace artisanal construction with mass production.

Ego, Fear and Money: How the A.I. Fuse Was Lit - The people who were most afraid of the risks of artificial intelligence decided they should be the ones to build it. Then distrust fueled a spiraling competition. Larry Page, co-founder of Google, was e/acc before e/acc was a thing, and Elon Musk was proto-EA. My friends in AI say that the story told here is broadly true; it goes back to the middle of the 2010’s, and personifies the factional conflicts.

In Florida’s Hot Political Climate, Some Faculty Have Had Enough - Liberal-leaning professors are leaving coveted jobs with tenure. And there are signs that recruiting scholars has become harder. Is this a feature or a bug? More candidly, there’s no way this will be a huge problem; there are too few research universities with tenure-track positions ‌ in the US.

H Street was once a symbol of D.C.’s rebirth. Now, it’s barely holding on. Gentrification is good actually. What’s bad is not enough housing (see the first piece in this section). Crime is bad. Run-down neighborhoods are bad. Tapas bars, gastropubs and loft-style apartments are good.

Measuring Religion in China. Most Chinese are not religious. About 5% of Chinese are Christian. About 33% believe in Buddha or the Bodhisattva.

Data

A genetic history of the Balkans from Roman frontier to Slavic migrations. Most of the Balkans now speak Slavic languages. It turns out this owes to the massive in-migration of people from northeastern Europe after the collapse of the Roman frontier under Justin II in the 570’s. More than 50% of the ancestry today in the Balkans derives from these invaders, with Slavic ancestry running around 10-30% even in southern Greece. Though the Latinized Balkan people have a major legacy in the region, the Germans and steppe Iranians who were prominent in the 5th and 6th centuries seem to have left no trace.

Evolutionarily new genes in humans with disease phenotypes reveal functional enrichment patterns shaped by adaptive innovation and sexual selection. Newer genes are found to be involved in functions that are targets of selection in human populations. This is logical (new adaptations have new genes, and are easier to detect than old adaptations), but now, with large genomic data sets, we can test these conjectures.

Ancestral neural circuits potentiate the origin of a female sexual behavior. Evolution created them, male and female. A lot of the interesting evolutionary behavior with a neuroscientific basis seems to be done in Drosophila. Understanding how male and female behaviors differ and what triggers them will probably general insight since so many genetic pathways are repurposed across animals.

Selective sweep probabilities in spatially expanding populations. This paper uses math and computers to model the evolutionary change in allele frequencies across a population with spatial structure. The authors find that the two methods align: “agent-based simulations…show that our analytical results accurately predict selective sweep frequencies in the two-dimensional spatial Moran process.” Too often we think of evolution happening like algebra in a vacuum, but newer models that take into account more realistic situations are good steps forward.

Whole-genome sequencing of half-a-million UK Biobank participants. Quantity has a quality all its own. This is how you find really rare variants. Also, can we just pause to note that Britain has done more for Asian and African genomics than the entire continents of Africa and Asia have.

My Two Cents

There’s still no free lunch, free subscribers; my most in-depth pieces for this Substack remain beyond the paywall. 

Genghis Khan: they don’t make stars like they used to:

These particular historical instances illustrate a broader phenomenon: the periodic exponential increase of a particular Y-chromosomal lineage and its sweep across a population. The most plausible explanation is what the 2003 paper's authors posit: social selection, which likely played out via conquest, the extermination of local male elites and the rape and amassing of women in harems. Their choice of term can sound bewilderingly bloodless, but that’s kind of the point; the authors were only interested in the patterns of inheritance and what drives them. So we can think of the misleadingly anodyne “social selection” as in contrast to “genetic selection.” Human behavior alone, not functional genetic fitness swept this lineage to overnight dominance. We can see this elsewhere historically in phenomena like the Iberian expansion into the New World. The frequencies of haplogroup Q crashed overnight, replaced by haplogroup R1b. This reflects the enslavement and extermination of indigenous men across much of Central and South America, and their replacement with lineages rooted in Europe. But star clusters are perhaps at their most intriguing before recorded history. A 2015 paper reported the emergence of several star clusters around 4,000 years ago, likely associated with polygynous social structures. We may never know who these people were because these expansions occurred beyond the purview of literate civilization, but it seems clear that Genghis Khan had many prehistoric forerunners. He may in fact have been among the last of his kind, not a singular exception.

If you want to browse my more geographically focused pieces, Dry.io has created an interactive map of them. We’ll keep adding to that page over time. Also, Dry.io set up a nice skin for my pinboard bookmarks and a page for reader-submitted links.

Discussion

All my podcasts go ungated two weeks after their Substack release. So I encourage subscribers on the free plan who’d like to automatically get them to subscribe to that podcast stream (Apple, Stitcher, and Spotify). If you want to listen on YouTube, please subscribe.

Here are my guests (and monologue topics) since the last Time Well Spent:

• Brent Roberts: let's talk about personality

• Nikolai Yakovenko: OpenAI in chaos, the future of artificial intelligence and effective accelerationism

• Mark Safranski: the 21st-century way of war and the exhaustion of the American Empire

• Philippe Lemoine: French food and American immigrants

• Cody Moser: the adaptive landscape of cultural evolution

• And here are the currently ungated podcasts all in one place.

For subscribers, I post transcripts (automatically generated, though I have someone going through to catch major errors).

ICYMI

Some of you follow me on my newsletter, blog, or Twitter. But my own domain also has all of the links and updates: razib.com. There you’ll find the few different podcasts I’ve contributed to or run, my total RSS feed, links to more mainstream or print articles when I remember to post them, my Twitter, the occasional guest appearance here and there, etc.

• DM me

• Facebook message me

• My total feed of content

• Some of my past pieces for UnHerd, National Review, The Manhattan Institute, Quillette, and The New York Times 

• This podcast, Unsupervised Learning

Over to you

Comments are open to all for this post, so if you have more reading/listening suggestions or tips on who I should be talking to or what you wish I would cover, lay it on us.