Halfhaven halftime

[Viliam]

Halfhaven is a virtual blogger camp, an online alternative to Inkhaven Residency.

The rules are simple:

• every day post max 1 article with min 500 words (or equivalent effort)

• try to get 30 by the end of November (but there are no hard lines)

The invitation links keep expiring, the current one is: https://​​discord.gg/​​rZv3a2Jt

If you wanted to join during November, but you couldn’t join because you didn’t have the link, I apologize; I was on a vacation without internet. If you published your articles online anyway, feel free to add them retroactively in Halfhaven with their actual publication date. In general, post a link in Halfhaven the same day you publish the post.

Here are the posts published during October:

a11ce

• Qamar

Adam Shai

• Don’t give up on ambitious interpretability

• Physical Computation and GPTs (Part 1: The Framework)

Alex Kurilin

• Hiring—The Big Picture

• Hiring—Telling Your Company’s Story

• The CTO Fork in the Road

• Which Game Are You Playing?

Algon

• Why’s equality in logic less flexible than in category theory?

• In which the author is struck by an electric couplet

• Do One New Thing A Day To Solve Your Problems

• What I’ve Learnt About How to Sleep

• Maybe Social Media Algorithm’s Don’t Suck

• Notes on the need to lose

• Chaos Alone is No Bar to Superintelligence

• What shapes does reasoning take but circular?

• What criticisms have been made of Bostrom’s Open Global Investment Model?

• Stars are a rounding error

• What does it feel like to understand?

• You’re not confused by Thermodynamics

• Don’t mock yourself

• Predictability is Underrated

• What is Lesswrong good for?

• Some astral energy extraction methods

• Book Review: To Explain the World

• Meditation is Dangerous

• Libraries need more books

• Why do leaves fall in autumn?

• Discovery Fiction: Max Flow Algorithm

• The Doomers Were Right

• Kleinian view of geometry

• Remembrancy

• Credit goes to the presenter, not the inventor

• All the lab’s AI safety Plans: 2025 Edition

• Genius is Not About Genius

• Centralization begets stagnation

• Supervillain Monologues Are Unrealistic

• Random things I learnt about ASML after wondering how critical they were to GPU progress

Aq

• No, That’s Not What the Flight Costs

Ari Zerner

• Treading Water

• Introducing Halfhaven

• Some Nonsense For My Third Halfhaven Post

• You Are Good

• SlutCon Vignettes: Day 1

• More About SlutCon

• Good Grief

• Letter To My Past

• Free Writing

duck master

• How could criminal justice work in Ducktopia?

• my retrospective of the october 8 board games meetup

Gyrodiot

• Off start

• Indulge

ironlordbyron

• Aspiring Phoenix

• Rumination Is A Functional Goal-Achievement Subsystem

• Parasite Apostle

• SlutCon Diary

• Slutcon: Confidence Engineering Postmortem

• You Can Just Give Teenagers Social Anxiety!

• You Want Exposure Therapy To Be Ego-Syntonic

keltan

• Why did OpenAI make a Sora App?

• ENIGMA

• Bring Honor to Us All

• Starts to Stories I did not Finish

• Sure, I can help you with Human flourishing

• Response to a Reddit Post

• The Rationalist PR Department

• Minds as Art/​Odd Prompt

• Why My Classes Like Me

• A Superbly Crafted Sculpture of a Naked Obese Woman from the Stone Age

• Hunger Pains—Part 1

• Limitless Box & Random Patterns

• Making Films Quick Start − 1 Audio

• Making Films Quick Start − 2 Visuals

• Film Making Quick Start 3 - Editing

Logan Riggs

• How to Feel More Alive

• Making Your Pain Worse can Get You What You Want

• Telling the difference between memories and logical guesses

• Replacing RL w/​ Parameter-based Evolutionary Strategies

• Hospitalization: a Review

• Your First Love Song is Perfect

• How to Solve Racism (in voice chat games)

• C is for Cat. B is for Boundaries

• I Didn’t Sign Up For These Standards

• My comment on the bayes influence function paper

• Temporarily Losing My Ego

lsusr (videos)

• Why Meditation Causes Ego Death

• stories about skinwalkers

• Sometimes, the Best Criticism is an Homage

• How to Cultivate Spikier Preferences

• Why Jhānas are Non-Addictive

• You can just Choose your Social Reality

• The Direct Perception of the Simulation in Your Brain

mishka

• Meta: a problem of choice

• Sequence: non-standard approaches to AI existential safety

• Some of the promising non-anthropocentric invariants

• Interlude: non-anthropocentric vs anthropocentric approaches

• Invariants of self-modifying systems and self-governance

• Methods for handling invariants of self-modifying systems

• Ways to maintain continuous and gradual nature of changes

• How to include the changing world in our modeling

• Different levels of including a model of the world

• Recap 1: “Modest alignment” vs “MIRI alignment”

• Recap 2: Approximate invariants of self-modifying systems

• Summary: non-anthropocentric approaches to AI existential safety

• Diary: hyperstitionai, agentic engineering, recent papers, NVIDIA DGX Spark

• Follow-up: HyperstitionAI, agentic engineering

• Diary: agentic stuff and other things

• Diary: overview of recent “zzznah” (neural cellular automata, etc.)

• Qualitative transitions: from AlexNet to competent agents

niplav

• Shake Brains First

• I Believe the Value Misspecification Argument

• Some Thoughts on the Stupid Successionism Debate

• Waiting Outside a Shop Is Rarely Worth It; Public Transport Has Bad Vibes; Doing Cold Approach Every Day Is Pretty Powerful

• Quantum Computing is about Atoms not Bits

• The Need for Speed

• Humanity Learned Almost Nothing From COVID19

• Edits of Pergraphs, Pascal’s Mugging and Anti-Superpersuasion interventions

• “Solving” Pascal’s mugging by making some very strange modifications to the prior

ParrotRobot

• Freewriting in my head, and overcoming the “twinge of starting”

Philip

• Do Things for as Many Reasons as Possible

• Journal Update (June-Sep)

• Lessons from the Mountains

Tassilo

• Some Biology Related Things I Found Interesting

• Consider Doing Small Walks at Work

• Notes on “Homology, Genes and Evolutionary Innovation”

• Patience and Willingness to Be Slow

• My Software Setup

• Cheap Labour everywhere

• Learnable Skills

• Brainstorming Food on the Cheap+Healthy+Convenient+Edible Frontier

Taylor G. Lunt

• Fool Heart Joins Halfhaven

• Claude Imagine Bypassing its Security Sandbox Using Screenshot APIs

• Trial of Claude 4.5 on Chess and Fantastical Chess

• One Does Not Simply Walk Away from Omelas

• 10 Ways to Waste a Decade

• Lift ’n’ Peel

• The Liberty Tractor

• If a Lioness Could Speak

• The Mom Test for AI Extinction Scenarios

• The Three Levels of Agency

• Give Me Your Data: The Rationalist Mind Meld

• Cover Your Cough

• The Real Cost of a Peanut Allergy

• Guys I might be an e/​acc

• Unsureism: The Rational Approach to Religious Uncertainty

• Asking Paul Fussell for Writing Advice

Viliam

• Animating images in Java (1)

• Paint.NET

• The Quest

• Animating images in Java (2)

• Animating images in Java (3)

Feel free to join anytime. If you join later than November 1st, you won’t be able to achieve the soft goal of 30 total articles before the end of November, but a few articles is still more than nothing, if you need the nudge.

Congratulations to: Algon, keltan, mishka, and Taylor G. Lunt—guessing by your productivity so far, you are most likely to have the 30 posts before the end of November. But nothing is certain until you actually do it, and the rest of us may still have a chance.