newsletter/2019/02 (Link Bibliography)

“newsletter/​2019/​02” links:

  1. 02

  2. https://gwern.substack.com

  3. 01

  4. newsletter

  5. Changelog

  6. https://www.patreon.com/gwern

  7. ⁠, Gwern Branwen (2019-02-19):

    (TWDNE) is a static website which uses JS to display random anime faces generated by StyleGAN neural networks, along with GPT-3-generated anime plot summaries. Followups: ⁠/​​​​⁠/​​​​⁠.

    A screenshot of “This Waifu Does Not Exist” (TWDNE) showing a random StyleGAN-generated anime face and a random GPT-3 text sample conditioned on anime keywords/​​​​phrases.
  8. TWDNE

  9. Faces

  10. Littlewood-origin

  11. https://edwardtufte.github.io/tufte-css/#epigraphs

  12. http://share.obormot.net/misc/gwern/image-focus.js

  13. http://share.obormot.net/misc/gwern/image-focus.css

  14. https://journals.sagepub.com/doi/pdf/10.1177/0956797617744542

  15. ⁠, M-R. Rautiainen, T. Paunio, E. Repo-Tiihonen, M. Virkkunen, H. M. Ollila, S. Sulkava, O. Jolanki, A. Palotie, J. Tiihonen (2016-09-06):

    The pathophysiology of antisocial personality disorder (ASPD) remains unclear. Although the most consistent biological finding is reduced grey matter volume in the frontal cortex, about 50% of the total liability to developing ASPD has been attributed to genetic factors. The contributing genes remain largely unknown. Therefore, we sought to study the genetic background of ASPD. We conducted a (GWAS) and a replication analysis of Finnish criminal offenders fulfilling DSM-IV criteria for ASPD (n = 370, n = 5850 for controls, GWAS; n = 173, n = 3766 for controls and replication sample). The GWAS resulted in suggestive associations of two clusters of single-nucleotide polymorphisms at 6p21.2 and at 6p21.32 at the human leukocyte antigen (HLA) region. Imputation of HLA alleles revealed an independent association with DRB1*01:01 (odds ratio (OR) = 2.19 (1.53–3.14), p = 1.9 × 10-5). Two polymorphisms at 6p21.2 LINC00951LRFN2 gene region were replicated in a separate data set, and rs4714329 reached genome-wide statistical-significance (OR = 1.59 (1.37–1.85), p = 1.6 × 10−9) in the ⁠. The risk allele also associated with antisocial features in the general population conditioned for severe problems in childhood family (β = 0.68, p = 0.012). Functional analysis in brain tissue in open access GTEx and Braineac databases revealed eQTL associations of rs4714329 with LINC00951 and LRFN2 in cerebellum. In humans, LINC00951 and LRFN2 are both expressed in the brain, especially in the frontal cortex, which is intriguing considering the role of the frontal cortex in behavior and the neuroanatomical findings of reduced gray matter volume in ASPD. To our knowledge, this is the first study showing genome-wide statistically-significant and replicable findings on genetic variants associated with any personality disorder.

  16. https://pdfs.semanticscholar.org/f116/27888a7e0b1cb7077dff91c9d97df1cad062.pdf

  17. ⁠, Jorim J. Tielbeek, J. C. Barnes, Arne Popma, Tinca J. C. Polderman, James J. Lee, John R. B. Perry, Danielle Posthuma, Brian B. Boutwell (2018-08-23):

    Prior evolutionary theory provided reason to suspect that measures of development and reproduction would be correlated with antisocial behaviors in human and non-human species. Behavioral genetics has revealed that most quantitative traits are heritable, suggesting that these phenotypic correlations may share genetic etiologies. We use GWAS data to estimate the genetic correlations between various measures of reproductive development (n = 52,776 – 318,863) and antisocial behavior (n = 31,968). Our genetic correlation analyses demonstrate that alleles associated with higher reproductive output (number of children ever born, r g = 0.50, p = 0.0065) were positively correlated with alleles associated with antisocial behavior, whereas alleles associated with more delayed reproductive onset (age of first birth, r g = -0.64, p = 0.0008) were negatively associated with alleles linked to antisocial behavior. Ultimately, these findings coalesce with evolutionary theories suggesting that increased antisocial behaviors may partly represent a faster life history approach, which may be significantly calibrated by genes.

  18. 2019-odintsova.pdf: “Genomics of human aggression: current state of genome-wide studies and an automated systematic review tool”⁠, Veronika V. Odintsova, Peter J. Roetman, Hill F. Ip, René Pool, Camiel M. Van der Laan, Klodiana-Daphne Tona, Robert R. J. M. Vermeiren, Dorret I. Boomsma

  19. 2005-baud.pdf: “Personality traits as intermediary phenotypes in suicidal behavior: Genetic issues”⁠, Patrick Baud

  20. 2005-moffitt.pdf: “Genetic and environmental influences on antisocial behaviors: evidence from behavioral–genetic research”⁠, Terrie E. Moffitt

  21. 2012-craig.pdf: “Genetics of human aggressive behaviour”⁠, Ian W. Craig, Kelly E. Halton

  22. ⁠, Tuvblad, Catherine Baker, Laura A (2011; genetics  /​ ​​ ​heritable  /​ ​​ ​adoption):

    This chapter reviews the recent evidence of genetic and environmental influences on human aggression. Findings from a large selection of the twin and adoption studies that have investigated the genetic and environmental architecture of aggressive behavior are summarized. These studies together show that about half (50%) of the in aggressive behavior is explained by genetic influences in both males and females, with the remaining 50% of the variance being explained by environmental factors not shared by family members. Form of aggression (reactive, proactive, direct/​​​​physical, indirect/​​​​relational), method of assessment (laboratory observation, self-report, ratings by parents and teachers), and age of the subjects-all seem to be significant moderators of the magnitude of genetic and environmental influences on aggressive behavior. Neither study design (twin vs. sibling adoption design) nor sex (male vs. female) seems to impact the magnitude of the genetic and environmental influences on aggression. There is also some evidence of gene-environment interaction (G × E) from both twin/​​​​adoption studies and molecular genetic studies. Various measures of family adversity and social disadvantage have been found to moderate genetic influences on aggressive behavior. Findings from these G × E studies suggest that not all individuals will be affected to the same degree by experiences and exposures, and that genetic predispositions may have different effects depending on the environment.

  23. 2012-anholt.pdf: “Genetics of Aggression”⁠, Robert R. H. Anholt, Trudy F. C. Mackay

  24. https://www.nature.com/articles/mp201331

  25. https://jeb.biologists.org/content/218/1/123

  26. ⁠, Jorim J. Tielbeek, Sarah E. Medland, Beben Benyamin, Enda M. Byrne, Andrew C. Heath, Pamela A. F. Madden, Nicholas G. Martin, Naomi R. Wray, Karin J. H. Verweij (2012-08-17):

    Crime poses a major burden for society. The heterogeneous nature of criminal behavior makes it difficult to unravel its causes. Relatively little research has been conducted on the genetic influences of criminal behavior. The few twin and adoption studies that have been undertaken suggest that about half of the variance in antisocial behavior can be explained by genetic factors. In order to identify the specific common genetic variants underlying this behavior, we conduct the first genome-wide association study (GWAS) on adult antisocial behavior. Our sample comprised a community sample of 4816 individuals who had completed a self-report questionnaire. No genetic polymorphisms reached genome-wide for association with adult antisocial behavior. In addition, none of the traditional candidate genes can be confirmed in our study. While not genome-wide statistically-significant, the gene with the strongest association (p = 8.7×10−5) was DYRK1A, a gene previously related to abnormal brain development and mental retardation. Future studies should use larger, more homogeneous samples to disentangle the etiology of antisocial behavior. Biosocial criminological research allows a more empirically grounded understanding of criminal behavior, which could ultimately inform and improve current treatment strategies.

  27. ⁠, McGue, Matt Zhang, Yiwei Miller, Michael B. Basu, Saonli Vrieze, Scott Hicks, Brian Malone, Steve Oetting, William S. Iacono, William G (2013):

    We report results from a genome wide association study (GWAS) of five quantitative indicators of behavioral disinhibition: nicotine, alcohol consumption, alcohol dependence, illicit drugs, and non-substance related behavioral disinhibition. The sample, consisting of 7,188 Caucasian individuals clustered in 2,300 nuclear families, was genotyped on over 520,000 markers from Illumina’s Human 660W-Quad Array. Analysis of individual SNP associations revealed only one marker-component phenotype association, between rs1868152 and illicit drugs, with a p value below the standard genome-wide threshold of 5 × 10−8. Because we had analyzed five separate phenotypes, we do not consider this single association to be significant. However, we report 13 SNPs that were associated at p < 10−5 for one phenotype and p < 10−3 for at least two other phenotypes, which are potential candidates for future investigations of variants associated with general behavioral disinhibition. Biometric analysis of the twin and family data yielded estimates of additive heritability for the component phenotypes ranging from 49 to 70%, estimates of heritability for the same phenotypes ranged from 8 to 37%. Consequently, even though the common variants genotyped on the GWAS array appear in aggregate to account for a sizable proportion of heritable effects in multiple indicators of behavioral disinhibition, our data suggest that most of the additive heritability remains “missing”.

  28. ⁠, Tiihonen, J. Rautiainen, M-R Ollila, H. M Repo-Tiihonen, E. Virkkunen, M. Palotie, A. Pietiläinen, O. Kristiansson, K. Joukamaa, M. Lauerma, H. Saarela, J. Tyni, S. Vartiainen, H. Paananen, J. Goldman, D. Paunio, T (2015):

    In developed countries, the majority of all violent crime is committed by a small group of antisocial recidivistic offenders, but no genes have been shown to contribute to recidivistic violent offending or severe violent behavior, such as homicide. Our results, from two independent cohorts of Finnish prisoners, revealed that a monoamine oxidase A (MAOA) low-activity genotype (contributing to low turnover rate) as well as the CDH13 gene (coding for neuronal membrane adhesion protein) are associated with extremely violent behavior (at least 10 committed homicides, attempted homicides or batteries). No substantial signal was observed for either MAOA or CDH13 among non-violent offenders, indicating that findings were specific for violent offending, and not largely attributable to substance abuse or antisocial personality disorder. These results indicate both low monoamine metabolism and neuronal membrane dysfunction as plausible factors in the etiology of extreme criminal violent behavior, and imply that at least about 5–10% of all severe violent crime in Finland is attributable to the aforementioned MAOA and CDH13 genotypes.

  29. ⁠, Eric Mick, James McGough, Curtis K. Deutsch, Jean A. Frazier, David Kennedy, Robert J. Goldberg (2013-12-27):

    Background: Community samples suggest that approximately 1 in 20 children and adults exhibit clinically-significant anger, hostility, and aggression. Individuals with dysregulated emotional control have a greater lifetime burden of psychiatric morbidity, severe impairment in role functioning, and premature mortality due to cardiovascular disease.

    Methods: With publicly available data secured from dbGaP, we conducted a genome-wide association study of proneness to anger using the Spielberger State-Trait Anger Scale in the Atherosclerosis Risk in Communities (ARIC) study (n = 8,747).

    Results: Subjects were, on average, 54 (range 45–64) years old at baseline enrollment, 47% (n = 4,117) were male, and all were of European descent by self-report. The mean Angry Temperament and Angry Reaction scores were 5.8±1.8 and 7.6±2.2. We observed a nominally statistically-significant finding (p = 2.9E-08, λ = 1.027—corrected pgc = 2.2E-07, λ = 1.0015) on chromosome 6q21 in the gene coding for the non-receptor protein-tyrosine kinase, Fyn.

    Conclusions: Fyn interacts with NDMA receptors and inositol-1,4,5-trisphosphate (IP3)-gated channels to regulate calcium influx and intracellular release in the post-synaptic density. These results suggest that signaling pathways regulating intracellular calcium homeostasis, which are relevant to memory, learning, and neuronal survival, may in part underlie the expression of Angry Temperament.

  30. https://www.nature.com/articles/tp201536

  31. https://onlinelibrary.wiley.com/doi/pdf/10.1002/ajmg.b.32363

  32. https://onlinelibrary.wiley.com/doi/pdf/10.1002/ajmg.b.32419

  33. http://biology-web.nmsu.edu/~houde/genetics%20of%20aggression.pdf

  34. https://onlinelibrary.wiley.com/doi/pdf/10.1002/ajmg.b.32388

  35. 2017-manchia.pdf: “Targeting aggression in severe mental illness_ The predictive role of genetic, epigenetic, and metabolomic markers”⁠, Mirko Manchia, Vassilios Fanos

  36. 2018-beaver.pdf: ⁠, Kevin M. Beaver, Eric J. Connolly, Joseph L. Nedelec, Joseph A. Schwartz (2018-04-01; genetics  /​ ​​ ​heritable):

    There is a great deal of interest in examining the genetic and environmental architecture to aggression, violence, and antisocial behaviors. This interest has resulted in hundreds of studies being published that estimate genetic and environmental effects on antisocial phenotypes. The results generated from these studies have been remarkably consistent and have contributed greatly to the knowledge base on the etiology of antisocial behavior. This chapter reviews the research on the genetic basis to antisocial phenotypes by presenting the results related to the heritability of antisocial phenotypes. It also discusses some of the molecular genetic association studies as well as genome-wide association studies that focus on the development of antisocial behaviors. In doing so, it also reviews findings related to gene-environment interactions. The chapter concludes by discussing some of the ways in which these findings could be used for intervention and prevention programs.

  37. 2018-tremblay.pdf: “Developmental Origins of Chronic Physical Aggression: A Bio-Psycho-Social Model for the Next Generation of Preventive Interventions”⁠, Richard E. Tremblay, Frank Vitaro, Sylvana M. Côté

  38. https://pdfs.semanticscholar.org/b15f/026aadf93744168fe9992cb86002a85647d2.pdf

  39. 2019-gard.pdf: “Genetic influences on antisocial behavior: recent advances and future directions”⁠, Arianna M. Gard, Hailey L. Dotterer, Luke W. Hyde

  40. https://pdfs.semanticscholar.org/2cb9/e5fe8009b934faa127778bf6aae028e30b62.pdf

  41. https://duepublico.uni-duisburg-essen.de/servlets/DerivateServlet/Derivate-28060/Oades2008Does.pdf

  42. https://duepublico.uni-duisburg-essen.de/servlets/DerivateServlet/Derivate-28837/Anney_ea_2008.pdf

  43. 2010-viding.pdf: “In search of genes associated with risk for psychopathic tendencies in children: a two‐stage genome‐wide association study of pooled DNA⁠, Essi Viding, Ken B. Hanscombe, Charles J. C. Curtis, Oliver S. P. Davis, Emma L. Meaburn, Robert Plomin

  44. ⁠, Dick, D. M Aliev, F. Krueger, R. F Edwards, A. Agrawal, A. Lynskey, M. Lin, P. Schuckit, M. Hesselbrock, V. Nurnberger, J. Almasy, L. Porjesz, B. Edenberg, H. J Bucholz, K. Kramer, J. Kuperman, S. Bierut, L (2011):

    Conduct disorder (CD) is one of the most prevalent childhood psychiatric conditions, and is associated with a number of serious concomitant and future problems. CD symptomatology is known to have a considerable genetic component, with heritability estimates in the range of 50%. Despite this, there is a relative paucity of studies aimed at identifying genes involved in the susceptibility to CD. In this study, we report results from a genome-wide association study of CD symptoms. CD symptoms were retrospectively reported by a psychiatric interview among a sample of cases and controls, in which cases met the criteria for alcohol dependence. Our primary phenotype was the natural log transformation of the number of CD symptoms that were endorsed, with data available for 3963 individuals who were genotyped on the Illumina Human 1M beadchip array. Secondary analyses are presented for case versus control status, in which caseness was established as endorsing three or more CD symptoms (n = 872 with CD and n = 3091 without CD). We find four markers that meet the criteria for genome-wide statistical-significance (p < 5 × 10−8) with the CD symptom count, two of which are located in the gene C1QTNF7 (C1q and tumor necrosis factor-related protein 7). There were six additional SNPs in the gene that yielded converging evidence of association. These data provide the first evidence of a specific gene that is associated with CD symptomatology. None of the top signals resided in traditional candidate genes, underscoring the importance of a genome-wide approach for identifying novel variants involved in this serious childhood disorder.

  45. https://www.nature.com/articles/tp201113

  46. ⁠, Mick, Eric McGough, James Loo, Sandra Doyle, Alysa E. Wozniak, Janet Wilens, Timothy E. Smalley, Susan McCracken, James Biederman, Joseph Faraone, Stephen V (2011):

    Objective: A potentially useful tool for understanding the distribution and determinants of emotional dysregulation in children is a Child Behavior Checklist profile, comprising the Attention Problems, Anxious/​​​​Depressed, and Aggressive Behavior clinical subscales (CBCL-DP). The CBCL-DP indexes a heritable trait that increases susceptibility for later psychopathology, including severe mood problems and aggressive behavior. We have conducted a genome-wide association study of the CBCL-DP in children with attention-deficit/​​​​hyperactivity disorder ().

    Method: Families were ascertained at Massachusetts General Hospital and University of California, Los Angeles. Genotyping was conducted with the Illumina Human1M or Human1M-Duo BeadChip platforms. Genome-wide association analyses were conducted with the MQFAM multivariate extension of PLINK.

    Results: CBCL data were available for 341 ADHD offspring from 339 ADHD affected trio families from the UCLA (n = 128) and the MGH (n = 213) sites. We found no genome-wide statistically-significant associations but identified several plausible candidate genes among findings at p < 5E-05: TMEM132D, LRRC7, SEMA3A, ALK, and STIP1.

    Conclusions: We found suggestive evidence for developmentally expressed genes operant in hippocampal dependent memory and learning with the CBCL-DP.

  47. https://onlinelibrary.wiley.com/doi/pdf/10.1002/ajmg.b.32333

  48. https://moffittcaspi.trinity.duke.edu/sites/moffittcaspi.trinity.duke.edu/files/Agression_ViolentBehavior.pdf

  49. ⁠, Sophie von Stumm, Emily Smith-Woolley, Ziada Ayorech, Andrew McMillan, Kaili Rimfeld, Philip S. Dale, Robert Plomin (2019-02-04):

    The two best predictors of children’s educational achievement available from birth are parents’ (SES) and, recently, children’s inherited DNA differences that can be aggregated in genome-wide polygenic scores (GPS). Here we chart for the first time the developmental interplay between these two predictors of educational achievement at ages 7, 11, 14 and 16 in a sample of almost 5,000 UK school children. We show that the prediction of educational achievement from both GPS and SES increases steadily throughout the school years. Using latent growth curve models, we find that GPS and SES not only predict educational achievement in the first grade but they also account for systematic changes in achievement across the school years. At the end of compulsory education at age 16, GPS and SES respectively predict 14% and 23% of the variance of educational achievement; controlling for genetic influence on SES reduces its predictive power to 16%. Analyses of the extremes of GPS and SES highlight their influence and interplay: In children who have high GPS and come from high SES families, 77% go to university, whereas 21% of children with low GPS and from low SES backgrounds attend university. We find that the effects of GPS and SES are primarily additive, suggesting that their joint impact is particularly dramatic for children at the extreme ends of the distribution.

  50. https://www.nature.com/articles/d41586-019-00395-4

  51. https://quillette.com/2019/01/30/the-death-of-a-dreamer/

  52. https://www.technologyreview.com/s/612838/the-transhumanist-diy-designer-baby-funded-with-bitcoin/

  53. ⁠, Alec Radford, Jeffrey Wu, Dario Amodei, Daniela Amodei, Jack Clark, Miles Brundage, Ilya Sutskever () (2019-02-14):

    Our model, called GPT-2 (a successor to GPT), was trained simply to predict the next word in 40GB of Internet text. Due to our concerns about malicious applications of the technology, we are not releasing the trained model. As an experiment in responsible disclosure, we are instead releasing a much smaller model for researchers to experiment with, as well as a ⁠.

    GPT-2 is a large -based language model with 1.5 billion parameters, trained on a dataset of 8 million web pages. GPT-2 is trained with a simple objective: predict the next word, given all of the previous words within some text. The diversity of the dataset causes this simple goal to contain naturally occurring demonstrations of many tasks across diverse domains. GPT-2 is a direct scale-up of GPT, with more than 10× the parameters and trained on more than 10× the amount of data.

    GPT-2 displays a broad set of capabilities, including the ability to generate conditional synthetic text samples of unprecedented quality, where we prime the model with an input and have it generate a lengthy continuation. In addition, GPT-2 outperforms other language models trained on specific domains (like Wikipedia, news, or books) without needing to use these domain-specific training datasets. On language tasks like question answering, reading comprehension, summarization, and translation, GPT-2 begins to learn these tasks from the raw text, using no task-specific training data. While scores on these downstream tasks are far from state-of-the-art, they suggest that the tasks can benefit from unsupervised techniques, given sufficient (unlabeled) data and compute.

  54. ⁠, OpenAI (2018-06-11):

    We’ve obtained state-of-the-art results on a suite of diverse language tasks with a scalable, task-agnostic system, which we’re also releasing. Our approach is a combination of two existing ideas: and ⁠. These results provide a convincing example that pairing supervised learning methods with unsupervised pre-training works very well; this is an idea that many have explored in the past, and we hope our result motivates further research into applying this idea on larger and more diverse datasets.

  55. 2019-radford.pdf#openai: ⁠, Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever (2019-02-14; ai):

    Natural language processing tasks, such as question answering, machine translation, reading comprehension, and summarization, are typically approached with supervised learning on task-specific datasets.

    We demonstrate that language models begin to learn these tasks without any explicit supervision when trained on a new dataset of millions of webpages called WebText. When conditioned on a document plus questions, the answers generated by the language model reach 55 F1 on the CoQA dataset—matching or exceeding the performance of 3 out of 4 baseline systems without using the 127,000+ training examples.

    The capacity of the language model is essential to the success of zero-shot task transfer and increasing it improves performance in a log-linear fashion across tasks. Our largest model, GPT-2, is a 1.5B parameter that achieves state of the art results on 7 out of 8 tested language modeling datasets in a zero-shot setting but still underfits WebText. Samples from the model reflect these improvements and contain coherent paragraphs of text.

    These findings suggest a promising path towards building language processing systems which learn to perform tasks from their naturally occurring demonstrations.

  56. ⁠, Sagar Savla (2019-02-04):

    The World Health Organization (WHO) estimates that there are 466 million people globally that are deaf and hard of hearing. A crucial technology in empowering communication and inclusive access to the world’s information to this population is automatic speech recognition (ASR), which enables computers to detect audible languages and transcribe them into text for reading. Google’s ASR is behind automated captions in Youtube, presentations in Slides and also phone calls…Today, we’re announcing Live Transcribe, a free Android service that makes real-world conversations more accessible by bringing the power of automatic captioning into everyday, conversational use. Powered by Google Cloud, Live Transcribe captions conversations in real-time, supporting over 70 languages and more than 80% of the world’s population. You can launch it with a single tap from within any app, directly from the accessibility icon on the system tray.

    …Relying on cloud ASR provides us greater accuracy, but we wanted to reduce the network data consumption that Live Transcribe requires. To do this, we implemented an on-device neural network-based speech detector, built on our previous work with AudioSet. This network is an image-like model, similar to our published VGGish model, which detects speech and automatically manages network connections to the cloud ASR engine, minimizing data usage over long periods of use.

    …Known as the cocktail party problem, understanding a speaker in a noisy room is a major challenge for computers. To address this, we built an indicator that visualizes the volume of user speech relative to background noise. This also gives users instant feedback on how well the microphone is receiving the incoming speech from the speaker, allowing them to adjust the placement of the phone…Potential future improvements in mobile-based automatic speech transcription include on-device recognition, speaker-separation, and speech enhancement. Relying solely on transcription can have pitfalls that can lead to miscommunication. Our research with Gallaudet University shows that combining it with other auditory signals like speech detection and a loudness indicator, makes a tangibly meaningful change in communication options for our users.

  57. https://www.blog.google/outreach-initiatives/accessibility/making-audio-more-accessible-two-new-apps/

  58. https://github.com/NVlabs/stylegan

  59. #gwern-twdne

  60. https://www.punchlinedesign.net/pun_generator

  61. https://nips2018creativity.github.io/doc/entendrepreneur.pdf

  62. https://slatestarcodex.com/2019/02/24/ot122-openne-thread/#comment-724333

  63. https://news.ycombinator.com/item?id=19241236

  64. https://llamasandmystegosaurus.blogspot.com/2017/05/alpha.html

  65. Faces-graveyard#progan

  66. 2019-lortieforgues.pdf: ⁠, Hugues Lortie-Forgues, Matthew Inglis (2019-03-11; sociology):

    There are a growing number of large-scale educational randomized controlled trials (). Considering their expense, it is important to reflect on the effectiveness of this approach. We assessed the magnitude and precision of effects found in those large-scale RCTs commissioned by the UK-based Education Endowment Foundation and the U.S.-based National Center for Educational Evaluation and Regional Assistance, which evaluated interventions aimed at improving academic achievement in K–12 (141 RCTs; 1,222,024 students). The mean was 0.06 standard deviations. These sat within relatively large confidence intervals (mean width = 0.30 SDs), which meant that the results were often uninformative (the median Bayes factor was 0.56). We argue that our field needs, as a priority, to understand why educational RCTs often find small and uninformative effects.

    [Keywords: educational policy, evaluation, meta-analysis, program evaluation.]

  67. 1987-rossi

  68. 2018-hanley.pdf: “Review of Scientific Self-experimentation: ethics history, regulation, scenarios, and views among ethics committees and prominent scientists”⁠, Brian Hanley, William Bains, George Church

  69. ⁠, Paul Fussell (1989-08):

    [Hard-hitting longform piece on WWII about demystifying the ‘good war’ and bringing home the chaos, folly, incompetence, suffering, death, destruction visited on soldiers, and propaganda or silence which covered it all up.]

    On its fiftieth anniversary, how should we think of the Second World War? What is its contemporary meaning? One possible meaning, reflected in every line of what follows, is obscured by that oddly minimizing term “conventional war.” With our fears focused on nuclear destruction, we tend to be less mindful of just what conventional war between modern industrial powers is like. This article describes such war, in a stark, unromantic manner

  70. 2010-dobelli.pdf: ⁠, Rolf Dobelli (2010; culture):

    This article is the antidote to news. It is long, and you probably won’t be able to skim it. Thanks to heavy news consumption, many people have lost the reading habit and struggle to absorb more than four pages straight. This article will show you how to get out of this trap—if you are not already too deeply in it.

  71. 2019-poulos.pdf: “Land lotteries, long-term wealth, and political selection”⁠, Jason Poulos

  72. ⁠, William Buckner (Traditions of Conflict) (2019-02-23):

    During an undetermined time period preceding European contact, a gargantuan, humanoid spirit-God conquered parts of the Sepik region of Papua New Guinea. With a voracious appetite for pork and yams—and occasional demands of ritual murder—Nggwal was the tutelary spirit for a number of Sepik horticulturalist societies…what specific demands does Nggwal make? The first is for food. Nggwal must be fed, and while it is the men who are his most devoted servants and the keepers of his great secrets, it is often the responsibility of the women to provide for his subsistence, “Women are well aware of Nggwal’s hunger, for to them falls much of the gardening, hauling and cooking needed to feed him”, Tuzin writes. But how does Nggwal consume the food offered to him? “Needless to say, it is not the Tambaran [Nggwal himself] which eats the pork but the men themselves, in secret conclaves”, and Tuzin continues describing the “feasts among Tambaran Cult members in secret seclusion, during which non-members are under the impression that the food is being given directly to the spirits.”

    …Despite the playful, Halloween-like aspects of this practice, the hangahiwa wandafunei [violent spirits] were a much more serious matter. 10% of the male masks portrayed hangahiwa wandafunei, and they were associated with the commission of ritually sanctioned murder. These murders committed by the violent spirits were always attributed to Nggwal.

    …Traditionally, hangahiwa wandafunei sought out victims who were alone in their garden or on the forest paths at dusk. Pigs, dogs and chickens were also fair game. After spearing the victim, the offending hangamu’w would escape back to its spirit house. The wearer would replace it with the other costumes and emerge without fear of detection—in time to join the general alarm aroused by the discovery of the body.

    Sometimes the wearer would not put the mask away, however, and instead he would take it to a nearby enemy village, where a relative or other acquaintance of his would take the mask and keep it in their own community’s spirit house, until it was time to be used and transferred once more. Through these ritual killings and the passage of costumes between communities, Nggwal impels cooperation between men of even hostile villages, and unites them in cult secrecy.

    Nggwal, who travels in structures of fiber and bone atop rivers of blood.

  73. ⁠, Noel D. Johnson, Mark Koyama (2014-02-01):

    How is rule of law established? We address this question by exploring the effect of increases in fiscal capacity on the establishment of well-enforced, formal, legal standards in a preindustrial economy. Between 1550 and 1700, there were over 2,000 witch trials in France. Prosecuting a witch required local judges to substantially deviate from formal rules of evidence. Hence, we exploit the substantially variation across time and space in witch trials and fiscal capacity across French regions between 1550 and 1700 to show that increases in fiscal capacity were associated with increased adherence to the formal rule of law. As fiscal capacity increased, local judges increasingly upheld de jure rules, and the frequency of witch trials declined.

  74. Abandoned-Footnotes

  75. ⁠, Sy Montgomery (Orion Magazine) (2011-10-25):

    [Discussion of the remarkable abilities & intelligence of octopuses, despite being small, fragile, asocial beings. With hundreds of millions of neurons (most in its arms, which appear to be able to think and act independently, coordinating with the other arms/​​​​mouth, with their immensely-strong suckers), octopus are able to recognize individuals and bear grudges (squirting water at the foe), somehow imitate color despite being color-blind, use tools, solve puzzles, and manipulate rocks to create shelters, they are noted escape artists: one octopus was found breaking out of its aquarium at night to feast in other tanks, sneaking back before humans returned.]

  76. http://learning.mpi-sws.org/memorize/

  77. https://www.newyorker.com/magazine/2019/02/11/a-suspense-novelists-trail-of-deceptions

  78. 2018-banziger.pdf: ⁠, Hans Bänziger (2018-05-04; biology):

    Wild Lisotrigona cacciae (Nurse) and L. furva Engel were studied in their natural forest habitat at three sites in northern Thailand, May 2013–November 2014. The author, both experimenter and tear source, marked the minute bees while they drank from his eyes viewed in a mirror. All marked workers, 34 L. cacciae and 23 L. furva, came repeatedly to engorge, 34 and 27 times on average, respectively. The maximum number of times the same L. cacciae and L. furva came was 78 and 144 visits in one day, respectively; the maximum over two days was 145 visits by one L. cacciae; the maximum number of visiting days by the same bee was four over seven days by one L. furva which made 65 visits totally. The same forager may collect tears for more than 10h in a day, on average for 3h15min and 2h14min for L. cacciae and L. furva, respectively. Engorging from the inner eye corner averaged 3.1 and 2.2 min, respectively, but only 1.3 and 0.9 min when settled on the lower eye lid/​​​​ciliae. The interval between consecutive visits averaged 3.3 min and 3.8 min, respectively. Lachryphagy occurred during all months of the year, with 91–320 foragers a day during the hot season and 6–280 foragers during the rainy season; tear collecting resumed after a downpour. During the cold season eye visitation was reduced to 3–64 foragers, but none left her nest when the temperature was below 22°C. Flying ranges were greater than in comparable non-lachryphagous meliponines. It is proposed that Lisotrigona colonies have workers that are, besides nectar and pollen foragers, specialized tear collectors. Tears are 200 times richer in proteins than sweat, a secretion well-known to be imbibed by many meliponines. Digestion of proteins dissolved in tears is not hampered by an exine wall as in pollen, and they have bactericidal properties. These data corroborate the inference that Lisotrigona, which also visit other mammals, birds and reptiles, harvest lachrymation mainly for its content of proteins rather than only for salt and water.

  79. ⁠, Matthew Skala (2004-06-10):

    [Philosophy piece attempting to explain, via an amusing analogy to classic RPG game Paranoia, to programmers how the rest of the world sees information: as tainted, in a dualist immaterial sense, by their history. Two bits are not identical even if they are identical, because they may have different histories; these are recorded and enforced by consensual society-wide hallucinations, such as intellectual property law. This may be insane, like in Paranoia, but that is how the human world works, and why many clever copyright hacks will fail.]

  80. https://www.lowtechmagazine.com/2015/12/fruit-walls-urban-farming.html

  81. http://code4k.blogspot.com/2010/12/crinkler-secrets-4k-intro-executable.html

  82. ⁠, Dave Addey (Typeset In The Future) (2014-12-01):

    [Discussion with screenshots of the classic Ridley Scott SF movie Alien, which makes extensive use of Helvetica, Futura, Eurostile Bold Extended, and other “modern” fonts to give a futuristic industrial feel to all of the (multilingual) spaceship/​​​​computer displays, controls, and credits; Alien also makes intriguing use of many logos, icons, and symbols for quick communication, tracing back to the Semiotic Standard.]

  83. ⁠, Dave Addey (2016-06-19; technology⁠, design⁠, fiction):

    [Discussion with screenshots of the classic Ridley Scott SF movie Blade Runner, which employs typography to disconcert the viewer, with unexpected choices, random capitalization and small caps, corporate branding/​​​​advertising, and the mashed-up creole multilingual landscape of noir cyberpunk LA (plus discussion of the buildings and sets, and details such as call costs being correctly inflation-adjusted).]

  84. http://blog.archive.org/2011/08/17/scanning-a-braille-playboy/

  85. 1957-nutter.pdf: “The true story of Russia's weakness [stifled by bad planning, bureaucratic inefficiency, and lack of any real incentive]”⁠, G. Warren Nutter

  86. http://neojaponisme.com/2011/11/28/the-great-shift-in-japanese-pop-culture-part-one/

  87. http://neojaponisme.com/2011/11/29/the-great-shift-in-japanese-pop-culture-part-two/

  88. http://neojaponisme.com/2011/11/30/the-great-shift-in-japanese-pop-culture-part-three/

  89. http://neojaponisme.com/2011/12/01/the-great-shift-in-japanese-pop-culture-part-four/

  90. http://neojaponisme.com/2011/12/02/the-great-shift-in-japanese-pop-culture-part-five/

  91. https://www.vanityfair.com/style/2012/01/prisoners-of-style-201201

  92. https://steve-yegge.blogspot.com/2012/03/borderlands-gun-collectors-club.html

  93. https://www.theatlantic.com/magazine/archive/1973/11/the-force-that-drives-the-flower/308963/

  94. http://www.infinityplus.co.uk/stories/colderwar.htm

  95. https://slatestarcodex.com/2019/02/13/the-proverbial-murder-mystery/

  96. https://slatestarcodex.com/2014/04/03/the-study-of-anglophysics/

  97. https://slate.com/news-and-politics/2003/04/the-poetry-of-donald-rumsfeld.html

  98. https://fivebooks.com/best-books/favourite-art-books-andrew-graham-dixon/

  99. https://practicaltypography.com/

  100. Movies#carmen