briantball.org/blog

What I'm Reading & Thinking

Books, articles, and occasional observations. Curated, not comprehensive. Paywalled links are marked.

The Antidote to Social Media Is Already Invented

Nicky Case — creator of The Evolution of Trust, Parable of the Polygons — just posted a Patreon update with rough illustrations for Audrey Tang's upcoming book on democratic tech and AI. That's what sent me down this rabbit hole. Audrey Tang is the Digital Minister of Taiwan. Built something called vTaiwan — a platform that uses software called Polis to find genuine consensus online. Not compromise. Actual agreement, surfaced from noise. The opposite of what Twitter does. Here's what bugs me about the social media problem. It's not just misinformation. It's structural. Every position you take online is a performance. You're playing to followers, to opponents, to the algorithm. Nuance gets punished. Outrage gets rewarded. The whole system is engineered against the kind of thinking democracy actually requires. A citizens' assembly does the opposite. Random sample of citizens — stratified to actually reflect the population. No cameras. No party whips. No Twitter. Experts present, small groups deliberate, and then people reason their way to a recommendation without an audience to perform for. Ireland used them twice. Abortion rights and same-sex marriage — two issues their legislature couldn't touch for decades. The assemblies moved. Both went to referendum. Both passed. Ordinary citizens got there before their politicians did. We need this in the United States. Now. Not as a replacement for elections — as a circuit-breaker for when the legislature is captured, the courts are politicized, and the media environment is broken. Which is now. Madison worried about factions. He'd recognize exactly what we've built. — James Fishkin, Democracy When the People Are Thinking. Hélène Landemore, Open Democracy. Nicky Case at ncase.me.

The Engram Remembers

Mike Bee (MikeyBeez) — my former IBM Informix colleague, now one of the most interesting AI writers on Medium — solving memory architecture for small local models using entropy as a single control signal. PAYWALLED. The core insight: perplexity can lie, MAUVE catches what perplexity misses. Directly relevant to the Groundhog Day problem I've been wrestling with in my own workflow. Code at github.com/MikeyBeez/HRS.

Read →

Anthropic Founders' 80% Pledge

All seven Anthropic co-founders pledging 80% of roughly .7B each. Dario Amodei's accompanying essay — The Adolescence of Technology, 38 pages — is the serious version of this argument. Hanauer said it pre-AI. Amodei is saying it as AI threatens to make the inequality problem exponentially worse. Pair with Federalist 10.

Read →

The Pitchforks Are Coming — For Us Plutocrats

Nick Hanauer wrote this in 2014. A Seattle billionaire, first non-family Amazon investor, making the enlightened self-interest argument: a healthy middle class isn't charity, it's the precondition for everything else. Twelve years later it reads like prophecy. He also has a TED talk by the same name.

Read →

Math Is a Mess: A Manifesto for Mathematical Cleansing

Tomio Kobayashi argues that mathematics is inherently simple and beautiful — we just buried it under centuries of notational noise. His two-root thesis (additive structure + true/false segregation) resonates with everything I've been trying to do at algorisme.org. Worth reading slowly.

Read →