How to explore Scott Alexander's work and his 1500+ blog posts? This unaffiliated fan website lets you sort and search through the whole codex. Enjoy!

See also Top Posts and All Tags.

Minutes:
Blog:
Year:
4622 tags
Show all filters
4 posts found
Jan 11, 2017
ssc

Heuristics Work Until They Don’t

8 min • 1,067 words • 350 comments
Scott Alexander warns against forming strong heuristics based on limited data, using examples from AI research, elections, and campaign strategies to illustrate the pitfalls of this approach. Longer summary
Scott Alexander discusses the dangers of forming strong heuristics based on limited data points. He presents three examples: AI research progress, election predictions, and campaign strategies. In each case, he shows how people formed confident heuristics after observing patterns in just one or two instances, only to be surprised when these heuristics failed. The post argues against treating life events as moral parables and instead advocates for viewing them as individual data points that may not necessarily generalize. Scott uses a mix of statistical reasoning, historical examples, and cultural references to illustrate his points. Shorter summary
Dec 12, 2016
ssc

Might People On The Internet Sometimes Lie?

12 min • 1,637 words • 317 comments
Scott Alexander examines why compelling but unlikely stories on large internet platforms are probably lies, despite our reluctance to believe so. Longer summary
Scott Alexander discusses the prevalence of seemingly incredible stories on large internet platforms like Reddit. He proposes that, given the massive user base, even a small percentage of trolls or liars can produce numerous convincing but false stories. This principle extends to viral news stories, blog posts, and even scientific research, where the most interesting or surprising results are disproportionately likely to be false. Despite understanding this logically, Scott notes that it's psychologically difficult to dismiss these stories as lies, and he explores possible reasons for this cognitive dissonance. Shorter summary
Feb 14, 2015
ssc

How Likely Are Multifactorial Trends?

5 min • 680 words • 381 comments
Scott Alexander questions the plausibility of multifactorial explanations for trends, using the US crime rate decline as an example, and seeks literature on comparing single-factor vs multi-factor explanations. Longer summary
Scott Alexander discusses the plausibility of multifactorial trends, using the decline in US crime rates as an example. He presents two perspectives: one arguing that complex phenomena like crime are likely caused by many small factors, and another suggesting that simultaneous changes in multiple factors is improbable. Scott leans towards the second perspective, questioning the likelihood of ten different factors each accounting for about 10% of the crime decline. He asks if there are ways to calculate the relative likelihood of single-factor versus multi-factor explanations, noting that he feels he's missing an existing body of literature on this topic. Shorter summary
Sep 03, 2014
ssc

The Guardian vs. Induction

10 min • 1,374 words • 82 comments
Scott Alexander criticizes a Guardian article's flawed reasoning in claiming 'Limits to Growth' accurately predicts economic collapse, comparing it to the philosophical 'grue-bleen induction problem'. Longer summary
Scott Alexander critiques a Guardian article claiming that the 1972 book 'Limits to Growth' accurately predicted economic collapse. He argues that the article's reasoning is flawed, comparing it to the philosophical 'grue-bleen induction problem'. Scott demonstrates how easy it is to make accurate predictions about the past and present, but how this doesn't necessarily translate to accurate future predictions. He uses humorous examples like a fictional book 'No Limits To Bears' and creates alternative models for the data presented. The post concludes that while economic collapse might still happen, the Guardian's argument for it is not scientifically sound. Shorter summary