Skip to stories

16–22 Nov 2026 3 articles

Reading digest.

0/3
read this
week
This week’s reading notesMany of this week's best reads ask how to live, judge, or act well when the old frameworks are shaky. In philosophy, Aeon's "Equality talk" uses the history of one word to show why it now divides people, and ends on juries and shared institutions. Asterisk Magazine's "We Are All Playable Characters" offers a fresh account of why character types like the tradwife and the NPC have taken over how people describe themselves. Scott Alexander's "Escaping The 'Outside Criticism' Trap" (Astral Codex Ten) adds a practical heuristic for deciding which critics to read. In AI, the standout is Ethan Mollick's "Fable 5.2 and Jagged AGI" (One Useful Thing). It finds expert-level work across many tasks, and it finds that the failures have changed: bloat, and no sense of when work isn't good enough. LessWrong's "Alignment 'train/test splits' should be sacred" makes a clear case for holding some alignment techniques out for testing. In economics and society, Experimental History's "We need more goat sacrifices" argues that most social interventions fail because we never check whether they work. SemiAnalysis's "The Datacenter Backlash Goes Bipartisan" is solid analysis, but it is tied to the moment. In science, Quanta Magazine's "The Evolving Story of How Chemistry Became Biology" is the strongest of several origins-of-life pieces, and Nautilus covers plate tectonics and a white dwarf pulsar more briefly. In history and culture, 3 Quarks Daily's "Architecture's Glass Worship is Decimating Bird Populations" links modernist design to an ecological cost. The Public Domain Review's "Soldiers of Misfortune: The Drawings of Urs Graf the Elder" reads art against mercenary culture. The Paris Review's "Belgrade Diary: Election Day in Serbia" is an eyewitness account. For a short break, try Charlotte Mew's "From a Window" at Dead Poets Daily, and Psyche's "My friend, my witness" for something personal.

Many of this week's best reads ask how to live, judge, or act well when the old frameworks are shaky. In philosophy, Aeon's "Equality talk" uses the history of one word to show why it now divides people, and ends on juries and shared institutions. Asterisk Magazine's "We Are All Playable Characters" offers a fresh account of why character types like the tradwife and the NPC have taken over how people describe themselves. Scott Alexander's "Escaping The 'Outside Criticism' Trap" (Astral Codex Ten) adds a practical heuristic for deciding which critics to read. In AI, the standout is Ethan Mollick's "Fable 5.2 and Jagged AGI" (One Useful Thing). It finds expert-level work across many tasks, and it finds that the failures have changed: bloat, and no sense of when work isn't good enough. LessWrong's "Alignment 'train/test splits' should be sacred" makes a clear case for holding some alignment techniques out for testing. In economics and society, Experimental History's "We need more goat sacrifices" argues that most social interventions fail because we never check whether they work. SemiAnalysis's "The Datacenter Backlash Goes Bipartisan" is solid analysis, but it is tied to the moment. In science, Quanta Magazine's "The Evolving Story of How Chemistry Became Biology" is the strongest of several origins-of-life pieces, and Nautilus covers plate tectonics and a white dwarf pulsar more briefly. In history and culture, 3 Quarks Daily's "Architecture's Glass Worship is Decimating Bird Populations" links modernist design to an ecological cost. The Public Domain Review's "Soldiers of Misfortune: The Drawings of Urs Graf the Elder" reads art against mercenary culture. The Paris Review's "Belgrade Diary: Election Day in Serbia" is an eyewitness account. For a short break, try Charlotte Mew's "From a Window" at Dead Poets Daily, and Psyche's "My friend, my witness" for something personal.

3 articles · by category

AI2

6.5

Why this one

It sits squarely on the reader's interest in AI alignment and model character, with concrete data on surprising dispositions, though it is more a lab report than a polished essay.

Fit
8
Subst.
6
Orig.
6
Impact
6
Writing
5

What Futures Do AI Models Want? An Exploratory Character Evaluation of AI for AI Strategy

LessWrong

An interesting, transparent exploratory eval of how frontier models judge long-term AI-strategy tradeoffs, though the samples are thin and the author admits the results are fragile.

+1

Epistemic Status: Evaluation represents approximately 2 weeks of work. This is mostly exploratory evaluation, and the code for evaluation was made using LLMs.

Instapaper
6.7

Why this one

It gives a simple, memorable framing of a core problem in alignment, how we can tell whether a model is actually aligned, and it doesn't need specialist knowledge to follow.

Fit
7
Subst.
6
Orig.
7
Impact
7
Writing
6

Alignment 'train/test splits' should be sacred

LessWrong

A short, clear argument that gives a useful name to a real problem in alignment practice, though it uses one example and offers little evidence.

+1

Tl;dr: Alignment techniques can be used for training alignment or for testing it, but they cannot be used for both! As we build out the set of alignment techniques available to us, we should enforce that some are judiciously reserved only for evaluating alignment.

Instapaper

Science1

6.3

Why this one

It's an accessible piece of science writing that walks through how astronomers reasoned from evidence to a conclusion, which suits the reader's interest in physics and nature.

Fit
7
Subst.
6
Orig.
6
Impact
5
Writing
7

A Dead Star Is Flashing Radio Waves Every 8 Minutes

Nautilus

A clear, short first-hand account of a new astronomical discovery that explains how the researchers reasoned it out, but it doesn't go into much depth.

+2

Since 2022, astronomers have been puzzling over a new kind of cosmic object called long-period radio transients.These are objects that emit bright flashes of radio waves into space every few minutes to every few hours.Some have been traced back to binary systems of a small, dim cool star called a red dwarf in orbit…

Instapaper

Keyboard shortcuts

j / k
Next / previous story
o / Enter
Open story
x
Mark read / unread
i
Save to Instapaper
w
Show / hide why
c / C
Toggle focused / all overview context
v
Cards / List
t
System / Light / Dark
h / ←
Previous week
l / →
Next week
/
Filter by tag
?
This shortcut sheet