Links

Genetics:

• Everything Is Heritable:

  • “The genetic architecture of the human cerebral cortex”, Grasby et al 2020

• Engineering:

  • “Genetic Architecture of Complex Traits and Disease Risk Predictors”, Yong et al 2020 (joint optimization of human traits is easy because the genetic correlations cooperate & in high-dimensional space there’s always an alternative)

AI:

• “Agent57: Outperforming the Atari Human Benchmark”, Badia et al 2020 (blog; Agent57 reaches the median human level across ALE—including Pitfall!/Montezuma’s Revenge. It is impressive but still sample-inefficient & uncomfortably baroque in combining what seems like every DM model-free DRL technique in one place: DDQN, Impala, R2D2, Memory Networks, Transformers, Neural Episodic Control, RND, NGU, PBT, MABs… Is model-free DRL a dead end if this is what it takes? I would have preferred to see ALE solved by better exploration in the enormously simpler MuZero.)

• “AutoML-Zero: Evolving Machine Learning Algorithms From Scratch”, Real et al 2020 (Github; blog; using evolutionary search to bootstrap SGD/regularization; presumably the evolved SGD can be used to meta-learn, and so on—it’s optimizers all the way up)

• 15.ai: NN TTS service for generating natural high-quality voices of characters with minimal data (especially MLP:FiM voices; demos: 0/1/2/3/4; EQ:D contest results)

• Matters Of Scale: “Train Large, Then Compress: Rethinking Model Size for Efficient Training and Inference of Transformers”, Li et al 2020 (previously: Kaplan et al 2020); “Rethinking Parameter Counting in Deep Models: Effective Dimensionality Revisited”, Maddox et al 2020; “Bayesian Deep Learning and a Probabilistic Perspective of Generalization”, Wilson & Izmailov2020; “On Linear Identifiability of Learned Representations”, Roeder et al 2020 (why do NNs work so well, don’t overfit, and train both faster & better the more you overparameterize? Perhaps bigger = better because you’re searching over more classes of models which all have a bias towards simplicity, so you have a better chance of finding a simple sub-network which performs well.)

Statistics/Meta-Science:

• “A controlled trial for reproducibility: For three years, part of DARPA has funded two teams for each project: one for research and one for reproducibility. The investment is paying off”, Raphael et al 2020 (the difficulties of creating reproducible biology research; the projects: 1/2/3/4/5; see also Lithgow et al 2017/Lucanic et al 2017)

• “The Ph.D. Grind: A Ph.D. Student Memoir”, Philip J. Guo2012/2015 (on the grim reality of grad school/academia; cf. “What Does Any of This Have To Do with Physics?”)

Politics/religion:

• “The most predictable disaster in the history of the human race: This is what Bill Gates is afraid of”, Vox2015; “‘It’s Just Everywhere Already’: How Delays in Testing Set Back the U.S. Coronavirus Response” (the Noble Lie & Tragedy Of The Anticommons: how regulators, paperwork, IRBs, privacy, and bioethics controlled, censored, and shut down US & global pandemic prevention efforts—extra Ecclesiam nulla salus)

• SSC Book Review: Hoover: An Extraordinary Life in Extraordinary Times, Whyte2018 (how Herbert Hoover became one of history’s greatest philanthropists, then one of the most hated presidents—‘Quakers (and Ashkenazi Jews) are the skeleton key to history’ etc; apropos of recent events, “Mister, we could use a man like Herbert Hoover again”?)

• “Myopic Voters and Natural Disaster Policy”, Healy & Malhotra2009; “How Near-Miss Events Amplify or Attenuate Risky Decision Making”, Tinsley et al 2012

• “Reddit and the Struggle to Detoxify the Internet: How do we fix life online without limiting free speech?” (chilling account of how Reddit employees actually moderate)

• “Inside the Soviet Union’s secret pornography collection: Off limits to the public but enjoyed by Soviet-era leaders, the Lenin Library collection grew out of erotica confiscated from aristocrats after the revolution”; “Russia’s House of Shadows: My apartment building was made to house the first generation of Soviet élite. Instead, it was where the revolution went to die” (“rank hath its privileges”, but “La révolution, comme Saturne, dévore ses propres enfants”…)

• “In the 1980s, a Far-Left, Female-Led Domestic Terrorism Group Bombed the U.S. Capitol” (WP)

• The Burned House Horizon

• “Radical Solutions: French mathematician Évariste Galois lived a full life. When he wasn’t trying to overthrow the government, he was reinventing algebra” (the short & unlucky life of a genius)

Psychology/biology:

• “Going Critical”, Kevin Simler (An explorable interactive discussion of how stuff spreads in networks)

• “Public health interventions and epidemic intensity during the 1918 influenza pandemic”, Hatchett et al 2007 (the key finding: “cities with ≤3 nonpharmaceutical interventions (NPIs)…experienced a median total excess death rate of 551/100,000, compared with a median rate of 405/100,000 in cities with four or more NPIs (p = 0.03)”; media)

• “Is the 1918 Influenza Pandemic Over? Long-Term Effects of In Utero Influenza Exposure in the Post-1940 U.S. Population”, Almond2006 (MR)

• “Why do dozens of diseases wax and wane with the seasons—and will COVID-19?” (on Martinez2018 & Price et al 2019)

• “Tripping on nothing: placebo psychedelics and contextual factors”, Olson et al 2020 (“12% were certain that they had taken a psychedelic…Before leaving, she…asked where she could get the placebo drug again”)

• “Beliefs About Human Intelligence in a Sample of Teachers and Nonteachers”, Warne & Burton2020 (radical overestimation of the power of IQ-boosting interventions—even made-up ones)

• “Orangutans, Resistance and the Zoo” (orangutans like Ken “the hairy Houdini” Allen are brilliant escape artists)

Technology:

• “The History of the URL”, Zack Bloom (the winding path that gave us the confusing modern DNS/URL system; similar to Eevee’s history of CSS)

• “Zip Files: History, Explanation and Implementation”, Hans Wennborg2020 (compression/decompression accounts for 5% of Google computer cycles)

• “Seeing the World in a Bag of Chips”, Park et al 2020 (media)

Economics:

• “How Cameo Turned D-List Celebs Into a Monetization Machine” (WP; see also “How Cameo Blew Up During Quarantine” & “Status as a Service”)

• The unsolved Chicago Tylenol murders

Fiction:

• “The Master of the Curators”, The Book Of The New Sun: The Shadow of the Torturer, Wolfe 1980 (homage to Jorge Luis Borges)

Film/TV

Animated:

• Golden Kamuy (review)