Razib Khan's Unsupervised Learning

How do we make science in the 21st century better? Stuart Buck, Executive Director of the Good Science Project has some ideas. More concretely, Buck is part of a broader movement of researchers, activists and philanthropists reimagining how science can be done in the wake of the replication crisis. Between 2010 and 2015 many fields of science relying on statistical methods from the 20th century were found to be plagued by methodological errors that produced the ‘sexy’ results the breathless media loves, but that turned out to be totally unfounded. Not replicable. Though the problem was especially rife in psychology, it applied to many fields that use statistics, even including biomedical science. On this episode of the Unsupervised Learning podcast, Razib talks to Buck about the problems that plagued science in the last few decades, and how science has become an overly professionalized quest for publications, rather than a method to uncover the truth. They explore the various planks of the Good Science Funding Manifesto. How do you get rid of the bureaucracy that interferes with scientists in their day to day? How do you get scientists to think boldly and take risks? And how do you enforce standards of open data access? Buck believes that with 21st-century information technology there are wide open possibilities to speed up knowledge production and dissemination.

On this episode of Unsupervised Learning Razib talks to his friend Claire Lehmann, founder and editor-in-chief of Quillette magazine, and columnist for The Australian. Though Lehmann’s initial public prominence involved her key role in the “intellectual dark web,” publishing thinkers critical of identity politics like Coleman Hughes, John Wood Jr. and John McWhorter, Razib was especially interested in the fact that over the last few years she has gotten involved in various online discussions centered around cultural differences between her home nation of Australia, and the US, where the majority of Quillette’s readers live. Razib draws Lehmann out about the fact that few Americans grasp what different views  Australians usually hold about the balance between values of liberty and equality. This gap has resulted in several clashes online fueled more by ignorance than disagreement. Lehmann also discusses the disintegration of the intellectual dark web itself, and the future directions that Quillette will take. She notes that one of the major fissures between her own views and those of many intellectual dark web luminaries is that many of them, like Bret Weinstein, are more reflexive dissenters. Outside of a few topics, like racial essentialism, Lehmann observes that the intellectual dark web was never a coherent movement. Finally, she reflects on the positives and negatives of social media, and how it has changed over the last five years of her editorship of Quillette.

A bit over one percent of Americans are of Filipino ancestry, making them one of the largest Asian American subgroups. Unlike Chinese, Mexicans or Europeans, Filipino immigrants are unique in that their homeland, the Philippines, was actually an American colony for five decades, between 1898 and 1946. This is one reason that the level of English fluency in the Phillippines is very high, a factor in very strong economic integration with the US through outsourcing. And yet despite the historically close ties between the US and the Philippines, most Americans are unaware that as many as one million Filipinos died in a rebellion against the US army 120 years ago. From the perspective of many Americans, the Philippines is just another Pacific nation with more American entanglements than most. Today on the Unsupervised Learning podcast, Razib talks to Manuel L. Quezon III, columnist for the Philippine Daily Inquirer, and grandson of Philippine President Manuel L. Quezon. Razib tells Quezon that his first awareness of the Philippines came with the overthrow of Ferdinand Marcos in 1986, and they discuss the fact that the late dictator’s son is now President-Elect and what means for politics in the island nation. Quezon addresses how Filipinos view themselves, whether as Southeast Asian, Pacific or Latin? He also notes that the global rise of populism has arrived in the Philippines, and predicts that Americans may not recognize much of its politics in the near future.

In 1973 the eminent evolutionary geneticist Theodosius Dobzhansky wrote an essay  entitled “Nothing in Biology Makes Sense Except in the Light of Evolution.” Presumably, that would include molecular biology, and as Dobzhanksy was writing, the field of molecular evolution was bearing fruit that would revolutionize our understanding of Darwinian evolutionary biology. Or, perhaps more precisely, it would extend and move beyond a purely Darwinian understanding of changes in the DNA sequence on the molecular level. In the 1970’s, the idea that evolution at the scale of DNA and proteins was “neutral” in relation to adaptive fitness came to the fore through the work of both population geneticists and molecular biologists. This is in contrast to the emphasis placed on natural selection and adaptation in Darwin’s original theory, and pushed forward by Dobzhansky and his colleagues in the mid-20th century with the Neo-Darwinian Synthesis. Today on the Unsupervised Learning podcast Razib talks to Alex Palazzo, a molecular biologist who has also thought deeply about the relationship between his field and evolution, and where we are 40 years after the neutralist revolution. The conversation covers the issues brought up in Palazzo’s paper Non-Darwinian Molecular Biology. Was Charles Darwin wrong? Well, his ideas and theory were clearly incomplete in various ways. Palazzo argues for the importance of the mechanistic and structural details of genes and DNA that go into explaining why evolution produces the diverse traits and characteristics we see all around us. He also discusses why complex lifeforms exist due to the built-in tolerance of sloppiness in DNA replication, and addresses questions such as why genomes vary in size so greatly (did you know that the wheat genome is forty times larger than the rice genome?).

Who was the smartest human of the 20th century? Though intellectual celebrity probably dictates that the majority would answer Albert Einstein, another candidate is the mathematician John von Neumann. Today on Unsupervised Learning Razib talks to science journalist Ananyo Bhattacharya, author of The Man from the Future: The Visionary Life of John von Neumann, and erstwhile physicist and editor at Nature. They discuss the life and science of a scholar whose mental acuity was so preternatural that he was affectionately labeled a “Martian” by his colleagues. Razib and Bhattacharya discuss the social context of von Neumann’s upbringing in the haute bourgeoisie of the late Austro-Hungarian Empire (his family was elevated to the nobility when von Neumann was ten), a milieu that facilitated his insatiable intellectual appetites and provided him an incomparable set of peers that would ensure he never became complacent. Then, Bhattacharya notes that Von Neumann was not exceptional at every intellectual endeavor. He may have made original contributions to mathematics, physics, economics, statistics and computing, but non-polymath mortals may take comfort that he was known to be a mediocre chess player and a life-threatening driver. To sum up, they consider some of the aforementioned contributions that the “Martian” made to human knowledge before dying prematurely from cancer at the age of 53.

In this episode of Unsupervised Learning Stuart Ritchie joins Razib., Ritchie is the author of Science Fictions: How Fraud, Bias, Negligence, and Hype Undermine the Search for Truth and Intelligence: All that Matters. Ritchie is also a lecturer at King’s College London and the author of the new Substack Science Fictions. Razib and Ritchie first discuss why he has a Substack considering all the different projects he’s already juggling, and what value he sees coming out of it (beyond the remunerative one). They also rewind the clock and discuss Ritchie’s involvement in the replication crisis a decade ago, where he judges we are today in terms of the awareness of pitfalls in science and best practices, and the path forward. Razib also wonders how debunked findings like “implicit bias” still continue to percolate through the popular culture and policy forums despite scant scientific support for their validity. Eventually Razib and Ritchie pivot to his professional bread and butter, behavior genetics, and the social and scientific debates around its relevance and abuses. Razib wonders if the cultural chasm between behavior genetics and other genetics fields can ever be bridged, while Ritchie makes the case for his chosen field as an important human endeavor. Finally, they discuss the controversy around Kathryn Paige Harden’s book The Genetic Lottery and his defense of her work at his Substack. 

Last month Razib talked to Alex Nowrestah of the Cato Institute about the state of migration and policy in the US in 2022. An enthusiast for immigration, Nowrestah expressed some chagrin that the issue has fallen off the American public’s radar, at least judging by the sharp dropoff in media inquiries to his office. And yet there remains a whole policy class in Washington D.C. that is still attending to the complex and fraught topics in and around migration that shape the future trajectory of American demographics. While Nowrestah definitely leans toward opening up the borders, today on the Unsupervised Learning podcast Razib talks to Jason Richwine, a resident scholar at the Center for Immigration Studies, who comes at the question from the opposite viewpoint. While much of the American elite, from the conservative pro-business chamber of commerce to liberal human-rights activists, aim to increase the stream of migration, Richwine and his colleagues at CIS argue that not only should America take back control of its borders, a serious case can be made for restricting and reducing immigration. Over the course of their discussion, they touch on the relevance of economics, culture and politics to the immigration question. In a 2022 America of full employment, Richwine is not persuaded of the economic case for more labor supply.  He makes the argument that full employment is a feature, not a bug, of the low-migration scenario imposed by the recent pandemic. And full employment is good for America’s working class in particular. Rather than economics, Richwine seems more concerned about the cultural impact of immigration, as newcomers transform the character and values of American society bit by bit. Where many analysts are focused on the macroeconomic impacts of migration, Razib and Richwine probe the deep roots of 17th-century American culture in different streams of Anglo migration, and consider what that can tell us about cultural differences in the 21st-century.

Three of R.A. Fisher’s  Ph.D. students remain active today, C. R Rao at age 101 and A. W. F. Edwards, and W. F. Bodmer, both 86. Bodmer was not only a student of Fisher, the cofounder of both population genetics and modern statistics, he was also mentored by Joshua Lederberg, the 1958 winner of the Nobel Prize in Medicine for his work in bacterial genetics. With more than 60 years in science, Bodmer joins Razib on this episode of Unsupervised Learning to discuss everything from his recollections of Fisher, Lederberg and Cavalli-Sforza, to the recent cancellation controversy around his Ph.D. advisor. Over the course of the hour, they go on to discuss what has surprised Bodmer about the trajectory of genetics over the past few decades (he thinks the recent “completion of the human genome” is a bit overhyped), his continuing passion for the HLA loci (which are notably difficult to map genomically), the People of the British Isles Project, as well as his current interest in cancer genomics. Bodmer’s massive public record spans the history of much of modern genomics, from work on linkage and recombination in the 1960s to being part of the 1000 Genomes Project in the 2010’s.

The official conversion of the nation of Lithuania to Christianity was in 1387. This means officially Lithuanians have been Christian for 635 years, and did not adopt the religion until more than 1,000 years after Constantine the Great accepted Christianity and set the Roman Empire on its way to becoming synonymous with the faith. But Francis Young, a historian of religion, is here to tell you there’s more to this story. His new book, Pagans in the Early Modern Baltic: Sixteenth-Century Ethnographic Accounts of Baltic Paganism, is an account of the practices and persistence of Baltic paganism down to the 16th-century, the age of the Renaissance and Reformation. Over the course of their conversation, Razib asks Young the reasons Lithuania came to Christianity so late (in the 1500’s, 30-40% of Lithuanians were pagan in their practice and belief), and how late did Lithuanian folk paganism persist? Debates still rage in the history of religion about the persistence of heterodox religious views and practices in Europe after Christianization, but Young makes a convincing case that in the instance of Lithuania there were historical and cultural reasons why a critical mass of the rural peasantry remained staunchly pagan down to early modernity, in contrast to the case in Western and Southern Europe, where Christianity’s roots ran deeper.

Rand Simberg is the author of 2014's Safe Is Not An Option: Overcoming The Futile Obsession With Getting Everyone Back Alive That Is Killing Our Expansion Into Space, and a space business consultant, as well as a longtime blogger and commentator. Today, on Unsupervised Learning Razib talks with Simberg about SpaceX’s ambitiously named vessel, Starship, and what it means for the space business. In the process, Simberg outlines just how much of a lead SpaceX has over its competitors, and how it has transformed the game over the last decade, lapping the private-sector competition and putting pressure on national space programs. Razib asks Simberg about the long-term prospects for manned spaceflight, and what’s stopping us from reaching Mars and beyond. Simberg revisits some of the ideas that he presented almost a decade ago in Safe Is Not An Option, and argues that many of the hurdles are cultural and regulatory, not technological.