60k Readers, Attention Dynamics Paper, and 18 links

January 2026: 60k Readers, Attention Dynamics Paper, and 18 links

Two HN front pages, one research paper, and a reading list across AI, markets, and quant

January started fast: the blog saw ~60,000 unique readers over the last 30 days, largely driven by two posts hitting the top of Hacker News and the downstream pickup that followed. If you subscribed because of that, Hi.

On the build side, I shipped a serverless RSS aggregator that bundles feeds for the Top 25/50/100 HN blog contributors. Two things I'm focusing on now: Gia Tecnica, an autonomous academic research pipeline we're rebuilding with LangGraph, and the first public version of my paper on Attention Dynamics in Online Communities SSRN preprint.

Below: what I've been writing, what I've been working on, and a reading list across AI/ML, markets, and quant.

What I've been writing

Preview image for: The Market Can Stay Irrational Longer Than You Can

The Market Can Stay Irrational Longer Than You Can Stay Solvent

Steve Eisman explains how U.S. equity markets have structurally decoupled from everyday economic reality through concentration and passive investing.

Preview image for: Beyond Vector Search: Why LLMs Need Episodic Memor

Beyond Vector Search: Why LLMs Need Episodic Memory

Explore cutting-edge approaches to LLM memory beyond context windows: from HeadKV attention optimization to EM-LLM's brain-inspired architecture.

Preview image for: Praise by Name, Criticize by Category: Warren Buff

Praise by Name, Criticize by Category: Warren Buffett Retires at 95

Buffett exits after paying $26.8B in federal taxes last year. What his letters reveal about admitting mistakes, insurance float, and why Abel inher...

Preview image for: Apple's AI Bet: Playing the Long Game or Missing t

Apple's AI Bet: Playing the Long Game or Missing the Moment?

Apple’s $157B cash pile and Gemini-powered Siri shift show a restrained AI strategy. Is Apple winning the war as AI models become low-cost commodit...

What I've been working on

GitHub - giatenica/gia-agentic-short-v2

Autonomous academic research pipeline; rebuilding the orchestration layer with LangGraph for better state, retries, and tool-using control flow.

Preview image for: Attention Dynamics in Online Communities: Power La

Attention Dynamics in Online Communities: Power Laws, Preferential Attachment, and Early Success Prediction on Hacker News

First public version: mapping how attention concentrates, decays, and reroutes. The HN sentiment post that hit ~500 points came from this research.

What I've been reading

■ The value of random zero-sum games: Expected payoff scales with √n strategies; connects game theory to option pricing. via arXiv

■ What an unprocessed photo looks like: Great posts that peels back the abstraction layer of something we believe to understand. via Maurycyz

■ Multiword matrix multiplication over large finite fields in floating-point arithmetic: Performance implications for cryptography; the floating-point trick is clever. via arXiv

■ Pokémon team optimization: Type coverage as a constraint satisfaction problem; includes working solver. You had me at optimization. via Nchagnet

■ 2025: The year in LLMs: Simon Willison's annual roundup; reminder that LLMs age like milk. via Simonwillison

■ Resistance training load does not determine resistance training-induced hypertrophy across upper and lower limbs: Volume matters more than weight; reinforcing my Push/Pull split. via Doi

■ Prompting People: On treating conversations like prompt engineering. Uncomfortably relatable. via Kuber

■ How Browsers Work: Interactive walkthrough of the rendering pipeline; the compositing section is worth the click. via Howbrowserswork

■ 21 Lessons From 14 Years at Google: Lesson 7 on "work on what matters" is the real insight. via Addyosmani

■ Why does a linear least squares fit appear to have a bias when applied to simple test data?: StackExchange remains the last bastion against AI slop; clean explanation of regression to the mean. via Stats

■ Try to Take My Position: The Best Promotion Advice Andrew Ever Got via Andrew

■ Comparing AI Agents to Cybersecurity Professionals in Real-World Penetration Testing: Key finding: the Triage Module matters more than the model. via arXiv

■ Top banking trends for 2026: Dumb money gets smarter. via Accenture

■ The Modern Peril of the Availability Heuristic: When Absence Becomes Evidence Everywhere. via Behavioraleconomics

■ What If Your Exhaustion Has Nothing to Do With Your Life?: The hidden war you've been waging since you woke up this morning via Thinkingrock

■ Ideas are cheap, execution is cheaper: The code was never the point. We just couldn’t see that until the code became free. via Davekiss

■ Fourier Space Time-Stepping for Option Pricing With Levy Models: Cleaner numerical approach to Lévy models than finite differences. via SSRN

■ Option Pricing as a Dynamic Nash Equilibrium: A Game-Theoretic Derivation of the Black-Scholes Model: Black-Scholes from game theory first principles; the arbitrage-free condition falls out naturally. via SSRN

And before we close this off, here’s a short brain teaser: Are two heads better than one? Thank you for reading; I hope you found something useful.

Have a wonderful week!
— Phil

Have feedback, comments, or ideas? I'd love to hear from you: [email protected]

Blog | Projects | Research | GitHub | Bluesky

Reply to unsubscribe