Essays
The Journal253 essays
Long-form writing on AI, philosophy, psychology, and systems thinking.
-
AI Ethics in Content Moderation: The Impossible Standard
AI content moderation is accurate for clear violations but falters on content that requires cultural context or nuanced judgment. The gap defines an impossible standard.
-
Vanity Metrics and the Theater of Data-Driven Decision Making
Few self-identified data-driven organizations could show a single decision that data changed last quarter. The rest perform data-drivenness without practicing it.
-
The Productivity Placebo: METR’s AI Coding Study
METR's randomized trial found AI tools made experienced developers slower while they believed the tools made them faster. The perception gap demands investigation.
-
The Art of the Escalation: When and How to Ask for Help
Escalating late is expensive. Delayed escalations added engineering days to each incident, and the typical gap between needing help and asking ran to days.
-
The paradox of AI-assisted creativity: can tools that compress knowledge also expand imagination?
-
Integration Architecture Is Where Good Systems Go to Die
Integration points cause a disproportionate share of production incidents despite making up a small part of the codebase. Integration is where architectural quality is tested most severely.
-
The difference between systems and processes (and why it matters for your work)
-
AI Ethics in the Supply Chain: Training Data Provenance Problem
None of the AI models traced could document full training data provenance; one drew on sources with no consent chain. AI has a supply chain problem.
-
Jira Has a Philosophy Problem, Not a UX Problem
Jira attempts to be both a coordination tool and a surveillance instrument. Separating these functions shortened cycle time, because engineers stopped performing for reports and used Jira to coordinate.
-
Estimation Is Epistemology: Confidence Intervals
Estimation is a question of what you know. Teams with calibrated confidence intervals land close to their estimates more reliably than gut-feel teams do. The difference was self-knowledge, not technical skill.