π How to Fix Your Context (Drew Breunig; via Simon Willison). Drew is coining lots of little phrases around the problems of LLM context (even in huge 1M context windows). Better yet, he's describing how to fix those problems.
π Your Review: Alpha School (Redacted / ACX). This is the most detail I've gotten about Alpha School since I heard about them a few years ago. This review actually immediately changed my perspective on what I could have my son accomplish over the summer and we started implementing it yesterday. Let's see how this goes.
π¦ Sam Altman on Founder Secrecy: No One Cares (Sam Altman). "No matter how great your idea is, no one cares." I see this a lot.
π Maybe youβre not Actually Trying (Cate Hall / Useful Fictions; via Thinking About Things). But actual reasons why you might not be trying.
π AI to the Rescue (Beth McMurtrie / The Chronicle of Higher Education; via Jeremy Wertheimer). Strong students ask questions; weaker students try to avoid asking questions.
π Face it: you're a crazy person (Adam Mastroianni / Experimental History). Unpacking a job into what it actually entails is a thing I actually enjoy doing.
π The Real Reasons Your Appliances Die Young (Rachel Wharton / Wirecutter). Did you have "It's Complicated!" on your bingo card?
π How not to lose your job to AI (Benjamin Todd / 80,000 Hours). We're all managing someone's context now.
π Writing Toy Software Is A Joy (Joshua Barretto). Nice eclectic set of examples.
π Running "Running a Bootcamp" Bootcamp (Ricki Heicklen). Teaching teacher-teachers how to teach teachers to teach.
π Going to an office and pretending to work: A business thatβs booming in China (Inma Bonet / El Pais; via Tyler Cowen). It's a role-playing co-working space.
π How I Built A Task Management Tool For Almost Nothing (Mike Masnick / TechDirt). We're entering an age of bespoke tools. Mike is a power user who had very little programming experience, but a clear vision for what he wants.
π Consider Knitting (Bob Nystrom). Not sure I'm going to start knitting, but Bob makes a pretty good case for it.
π Holacracy: The New Management System for a Rapidly Changing World by Brian J. Robertson (2015). If you wait a decade to read a book, you'll discover that all the major early adopters (Medium, Zappos) have already abandoned this newfangled philosophy.
π The Great Z'manim Debate: The History, The Science, And The Lomdus by Rabbi Ahron Notis (2022; via Harold Zazula). Solid introduction to the relevant astronomical terms and older and more modern astronomical models. Good discussion of the relevant halachot.
π Halakhic Positions of Rabbi Joseph B. Soloveitchik: Volume 8 by Aharon Ziegler (2020). Only two more to go until I finish the series.
π Outlive: The Science and Art of Longevity by Peter Attia & Bill Gifford (2023; via Isaac Selya). Assuming you want to have certain kinds of mobility when you're older, this book helps you work backwards to where you want to be today.
π The Education of an Idealist: A Memoir by Samantha Power (2019; via Tyler Cowen).
Some may interpret this bookβs title as suggesting that I began with lofty dreams about how one person could make a difference, only to be βeducatedβ by the brutish forces that I encountered. That is not the story that follows.
I mean, that's kinda what happens. She learns very quickly how to say the right things to get the things she wants.
π The E-Myth Revisited: Why Most Small Businesses Don't Work and What to Do About It by Michael E. Gerber (2009; via Shalev NessAiver). Inside you there are three wolves. The Technician just wants to do good work and thinks bottom up. The Entrepreneur looks ahead and thinks top down. The manager wants everything in a little box and hates change.
π Inside the AI boom that's transforming how consultants work at McKinsey, BCG, and Deloitte (Lakshmi Varanasi / Business Insider; via Moshe Kinderlehrer). I wonder how long this policy will last:
McKinsey consultants have access to ChatGPT with guardrails and cannot input client data into it.
π DOGE Days (Sahil Lavingia; via Tyler Cowen). In which Sahil discovers, like many before him, that you may have aspirations to write wonderful code to help the government, but you will be thwarted every step of the way.
π Saying Bye to Glitch (Pirijan Keth; via Simon Willison). I never used Glitch, but I remember when it was announced and thinking that it was very cool. Pirijan also describes how various FogCreek software was ahead of its time.
π Thoughts on thinking (Dustin Curtis). As the nature of work transitions, I expect to see lots of posts of this nature.
π How I taught my 3-year-old to read like a 9-year-old (Erik Hoel / The Intrinsic Perspective).
Itβs all the advantages of an iPad, none of the guilt. Youβve unlocked infinite self-entertainment. Long drive? Bring a book. Or five. Roman toddles into restaurants clutching a book as a backup activity, and reads while waiting in boring lines. Itβs also calming, and so helps with emotional regulation. Toddler energy descending rapidly into deviance? Go read a book! Itβs a parenting cheat code. I donβt know if this alone justifies the hours spent, but it sure is one heck of a benefit.
π I really donβt like ChatGPTβs new memory dossier (Simon Willison). It's pretty detailed.
π LLM Memory (Grant Slatton). I like mini taxonomy of approaches to memory. I think this shows us that we don't yet have a really good model for our own memory.
π The Price of Remission (David Armstrong / ProPublica).
When I started taking the drug, Iβd look at the smooth, cylindrical capsule in my hand and consider the fact I was about to swallow something that costs about the same as a new iPhone. A monthβs supply, which arrives in an ordinary, orange-tinged plastic bottle, is the same price as a new Nissan Versa.
π Book Review: Selfish Reasons To Have More Kids (Scott Alexander / Astral Codex Ten). "Have you tried doing less?" is often a good question. I don't think I've been able to implement the advice (at all?).
π AI-Generated Law (Bruce Schneier). Soon: AI laws on the limits of AI laws.
π I don't like NumPy (Dynomight). This one is for everyone who has ever had to try axis=0
when doing an operation in numpy
. The points about indexing and broadcasting hit home hard. I'm looking forward to the API proposals.