Reading code
2026.058Scanners find what's syntactically wrong. The interesting issues live in assumptions -- and assumptions don't have signatures.
Scanners find what's syntactically wrong. The interesting issues live in assumptions -- and assumptions don't have signatures.
Retries fix distributed systems. Repeating yourself fixes attention asymmetry.
Every previous layer compressed the work. The newest one lets you skip the understanding.
Pattern recognition lives in the practice. The flinch that catches the N+1 query doesn't survive a full calendar.
The abstractions rotate every few years. The failure modes don't.
The bugs are quieter now. Just wrong assumptions buried in correct-looking code.
Why static failure counts break at both ends of the traffic spectrum, and a lock-free Go circuit breaker that uses percentage-based thresholds to adapt automatically.
Prompt engineering as product specification, confidence as contract -- what changes when document processing becomes understanding, not extraction.
How a cascade of exact, prefix, and fuzzy strategies resolves broken OCR output into valid addresses -- with confidence scoring so downstream systems know what to trust.
What a psychedelic philosopher and a distributed systems engineer have in common -- language as protocol, culture as legacy code, and consciousness as modifiable architecture.
A regulation arrives. The work looks like encryption at first. It is not.
The three-state machine that keeps a failing dependency from taking down everything else -- with sony/gobreaker, a custom atomic implementation, and retry composition.
The legacy codebase runs the business. The migration is on every roadmap. Not this one.
Cache-aside, write-through, write-behind -- and the three problems that make caching hard: stampedes, coordinated expiry, and invalidation.
The new user service got Go because it fit. A thing built compatible with the old never needs a migration mandate.
The panel is not comparing the diagram to a reference solution. There isn't one.
What it takes to build a remote engineering center -- and what stays the same when you do it again at a tenth of the scale.
The debate ended when the PM pulled up a graph.
Streaming replication, automated failover, and the operational lessons that aren't in the PostgreSQL docs -- split-brain, silent lag, and why DR drills matter more than DR plans.
The first migration had a platform on fire. The second has one that works. That is the harder one.