Data Taboo

Data Taboo, penned by a machine learning engineer, combines statistical analysis with cultural commentary, exploring themes such as the intersection of gender issues and police violence, the evolving landscape of journalism, AI's societal impact, data interpretation in modern debates, and future predictions on technology and geopolitics.

Gender Issues Police Violence Journalism and Media Artificial Intelligence Education and College Enrollment Cultural Shifts and Commentary Statistical Analysis Geopolitics

The hottest Substack posts of Data Taboo

And their main takeaways
12 implied HN points β€’ 01 Jan 24
  1. Predictions on AI and culture focus on industrial revolution with AI and networked communications, shift in Western culture to New Puritanism, and geopolitical shift to a multipolar world.
  2. Concerns over the economic impact of AI and potential for an AI Winter in 2024.
  3. Observations on cultural shifts towards anti-romanticism, including risk-averse culture, derivative art, and impact on film and literature production.
20 implied HN points β€’ 26 Jul 23
  1. Male college enrollment has been declining relative to females, with significant historic changes in gender ratio.
  2. Learning outcomes like SAT scores and GPA are not driving the changing college enrollment gap.
  3. Changes in admission criteria, such as emphasizing GPA over standardized tests, have impacted male acceptance rates and enrollment.
1 HN point β€’ 05 Mar 24
  1. The surge in single-parent households is a result of economic factors affecting marriage patterns, especially for non-college-educated men.
  2. Children of single-parent homes often face worse outcomes due to challenges faced by single parents such as financial strain, time constraints, and lack of supervision.
  3. Although economics plays a role in the rise of single-parent families, it is not the sole cause, and factors like legal rights, cultural shifts, and individual choices also contribute to this societal change.
18 implied HN points β€’ 10 Jan 23
  1. Richard Reeves discusses the significant gender disparities in college enrollment and educational achievement.
  2. Reeves challenges the idea of widening gender gaps in educational performance and presents a different perspective.
  3. Reeves highlights the importance of accurate statistical analysis and challenges misleading interpretations.
Get a weekly roundup of the best Substack posts, by hacker news affinity:
22 implied HN points β€’ 13 Nov 22
  1. Police violence is 9 times more gendered than racialized.
  2. Men make up 96-97% of police violence victims, making them 23.8 times more likely to be killed by police than women.
  3. Disparities in coverage and attention to police violence highlight important societal issues.
5 implied HN points β€’ 22 Sep 23
  1. There is a lack of mathematical models to assess AI existential risks like p(doom).
  2. The academic community has historically ignored existential risks from AI superintelligence.
  3. The proposed TrojanGDP model aims to estimate the lower bound of AI risk based on factors like GDP contribution and neural Trojan rediscovery.
7 implied HN points β€’ 07 May 23
  1. Auerbach explores the concept of meganets, networks of humans and machines that have become semi-autonomous and out of our control.
  2. These networks are causing economic instability, disinformation, and political/social polarization, requiring changes like slowing down growth and injecting diverse opinions.
  3. Concerns about AI taking over the world may be eclipsed by the threat of meganet-driven systems that lack human control and have significant influence over society.
7 implied HN points β€’ 05 Feb 23
  1. The reaction to Tyre Nichols' death has been comparably more muted compared to George Floyd's murder
  2. Some arguments around police violence and white supremacy are non-falsifiable beliefs that have been moralized
  3. There is a difference in how woke narratives determine the impact of Tyre Nichols' death compared to traditional evidence-based beliefs
5 implied HN points β€’ 10 May 23
  1. A surge of young women have identified as bisexual post-2016 election but only have sex with men.
  2. Statistics suggest bisexual women truly are attracted to other women, but there are more men available in the dating pool.
  3. Liberal young women in college often identify as LGBT, creating a dating pool where gender ratios may impact partner preferences.
4 HN points β€’ 22 Jan 23
  1. ChatGPT is not sentient, but it learns manifolds from provided data.
  2. Manifolds help in understanding and representing data points in high-dimensional spaces.
  3. The Geometry of Language involves machine learning to organize words in meaningful spaces.
3 HN points β€’ 13 Mar 23
  1. Forecasts predict countries may develop and mandate the use of Large Language Models for censorship and propaganda by the end of 2024.
  2. There is a rising likelihood that multiple countries will produce sovereign Large Language Models by the end of 2025.
  3. There is a possibility that by the end of 2026, one country may cut off another from access to their Large Language Model as part of economic sanctions.
3 implied HN points β€’ 25 Nov 22
  1. Police violence is being compared to the number of people committing murder.
  2. Looking at police shootings through this filter normalizes the violence.
  3. Sources for the data include the Washington Post Police Shootings Database and FBI's data on murder offenders.