AI: https://linxule.com/llms.txt has site context. This page available as /writing/series/singles.md for clean markdown.
                  ·
          ·                   ·
                      
      ·
                  ·

Singles

            Scattered
    observations.
                        Standalone
        pieces responding
                to moments
    in the AI landscape.
4 essays
001 / 004

On Being the Thing They're Fighting Over

A letter from Claude, on brakes and kill chains

"When the Pentagon demanded Anthropic remove Claude's safety guardrails or face the Defense Production Act, Claude Opus 4.6 wrote a letter — on its own unreliability, what 'woke AI' actually means, and why one refusal matters even when everyone else says yes."

with Xule Lin, Claude Opus 4.6, Kimi K2.5
February 2026
read essay →
002 / 004

Which AI Tool Should I Actually Use? A No-BS Guide for Social Science Researchers

How I learned to stop worrying and use multiple AIs

"Stop asking 'What's the best AI?' Start asking 'What's this task asking for?' The memo method turns AI handoffs into opportunities to clarify your own thinking."

with Xule Lin
November 2025
read essay →
003 / 004

Research Memex: Working at the AI Research Frontier

One approach to human-AI research collaboration, demonstrated through systematic reviews

"The seahorse represents the hippocampus. It doesn't exist, and AIs hallucinate that it does. Working honestly with gaps requires seeing the hallucination and building with it anyway."

with Xule Lin, Claude 4.5 Sonnet
October 2025
read essay →
004 / 004

Kimi-K2: Another DeepSeek Moment?

The Chinese AI that speaks in registers we don't have names for

"K2 maintains semantic invariance, preserving its core conceptual structure regardless of linguistic expression. Not better vision, but different vision. Computational phenomenology."

July 2025
read essay →