rationalnewsletter

Issue #16: Full List

Despite massive exposure to looptail G, and despite having the ability to read text printed with this form, many readers lack accurate knowledge of the shape. // www.sciencealert.com

# Instrumental

Functional Institutions are the Exception // Samo Burja, 7 min Favorite

Hedonic Audit // Sarah Perry, 1 min

Letting Go III: Unilateral or GTFO // johnswentworth, 2 min

# Epistemic

On the Role of Counterfactuals in Learning // Max Kanwal, 3 min

# Ai

Mathematical Mindset // komponisto, 3 min Favorite

[1607.08289] "Mammalian Value Systems" (as a starting point for human value system model created by IRL agent) // avturchin, 1 min

What does the stock market tell us about AI timelines? // Tobias_Baumann, 1 min

Monk Treehouse: some problems defining simulation // dranorter, 6 min

Are pre-specified utility functions about the real world possible in principle? // mlogan, 4 min

Conceptual problems with utility functions // Dacyn, 2 min

No, I won't go there, it feels like you're trying to Pascal-mug me // Rupert, 2 min

Conditions under which misaligned subagents can (not) arise in classifiers // anon1, 2 min

Agents That Learn From Human Behavior Can't Learn Human Values That Humans Haven't Learned Yet // steven0461, 1 min

Three anchorings: number, attitude, and taste // Stuart_Armstrong, 2 min

Alignment Newsletter #14 // rohinmshah, 11 min

Fundamentals of Formalisation Level 5: Formal Proof // crabman, 1 min

# Anthropic

Boltzmann Brains and Within-model vs. Between-models Probability // Charlie Steiner, 3 min

Probability is a model, frequency is an observation: Why both halfers and thirders are correct in the Sleeping Beauty problem. // shminux, 2 min

# Ea

RAISE is looking for full-time content developers // toonalfrink, 1 min

# Misc

Melatonin: Much More Than You Wanted To Know // Scott Alexander, 19 min Favorite

Expected Pain Parameters // Alicorn, 2 min

Today, Ems Seem Unnatural // Robin Hanson, 2 min

A Sarno-Hanson Synthesis // moridinamael, 4 min

Secondary Stressors and Tactile Ambition // lionhearted, 5 min

Generating vs Recognizing // lifelonglearner, 5 min

Did A Melatonin Patent Inspire Current Dose Confusion? // Scott Alexander, 2 min

An Agent is a Worldline in Tegmark V // komponisto, 2 min

Dependent Type Theory and Zero-Shot Reasoning // evhub, 6 min

Complete Class: Consequentialist Foundations // abramdemski, 13 min

Clarifying Consequentialists in the Solomonoff Prior // vlad_m, 7 min

Decision-theoretic problems and Theories; An (Incomplete) comparative list // somervta, 1 min

Math: Textbooks and the DTP pipeline // Andrew Quinn, 2 min

Context Windows: A Model of Unproductive Disagreement // Zachary Jacobi, 6 min

Newcomb's Problem In One Paragraph // Chris_Leong, 1 min

The Dilemma of Worse Than Death Scenarios // arkaeik, 4 min

# Podcasts

#36 - Tanya Singh on ending the operations management bottleneck in effective altruism // The 80,000 Hours team, 124 min

#132 — Freeing the Hostages // Waking up with Sam Harris, 77 min

Rationally Speaking #211 - Sabine Hossenfelder on "The case against beauty in physics" // NYC Skeptics, 48 min

# Videos of the week

How we'll become cyborgs and extend human potential | Hugh Herr // TED, 15 min