Essays

Integration patterns: What enterprise architecture teaches about organizing a life

The Thundering Herd of 2026: SRE for AI Agents

AI agents generated 847 million daily API calls in Q4 2025. Traditional capacity planning cannot handle their correlated burst patterns. New SRE approaches are required.

On the Relationship Between Speed and Wisdom

Development cycles compressed 39x since 2005. Speed and wisdom operate on different timescales. Structured slowness, deliberate reflection built into velocity, bridges the gap.

Zero-Trust Architecture as Epistemic Humility

Organizations implementing zero-trust architecture reduced attack surfaces by 68% and eliminated lateral movement in 94% of simulated breaches. Zero trust is epistemic humility, not paranoia.

Simone Weil on Attention and Code Review Practice

Simone Weil called attention the rarest form of generosity. Applied to code review, attention means reading code to understand the mind that wrote it, not just to find bugs.

AI Ethics in Content Moderation: The Impossible Standard

AI content moderation achieves 92-96% accuracy for clear violations but drops to 54-68% for content requiring cultural context or nuanced judgment. The gap defines an impossible standard.

Bayesian Reasoning as Engineering Philosophy of Judgment

Bayesian reasoning is the discipline of holding beliefs as probabilities and updating as evidence arrives. Google teams using explicit probabilistic reasoning experienced 34% fewer costly reversals.

When Best Practices Become Worst Practices

Tracking 12 industry-standard practices across 8 organizations found 5 actively harmed the adopter. Practices are context-dependent tools, not universal rules.

Alienation in the Age of Automation: Marx Was Partly Right

Marx described alienation as separation from the products of labor. The $395 billion automation industry has scaled that separation. Designing automation that preserves meaning is an engineering responsibility.

Retrieval-Augmented Generation and the 89% Problem

RAG systems achieving 89% retrieval accuracy mean 1 in 9 queries produce responses built on incorrect context. These errors are harder to detect than hallucination because every verification signal confirms the response.