Pigfox

Blog

Field notes on OSINT, security checks, SEO, and safer hiring and vendor due diligence.

• Kafka Lab and the Limits of Rate Limiting

  2026-09-15

  Kafka Lab shows a common systems mistake: slowing producers does not fix consumers whose steady-state processing rate is too low. The useful work is to measure service rate, inspect lag by cause, and separate burst control from throughput repair.

• Orbit Lab and the engineering of browser orbital simulation

  2026-09-14

  Orbit Lab shows what browser-based orbital simulation demands from engineering, stable time integration, clean state handling, and clear separation between physics and rendering. Its value is in how easily you can inspect those choices under real browser conditions.

• Reconcile Lab and the engineering of refused writes

  2026-09-13

  Reconcile Lab shows a core controller failure mode, what happens when a reconcile loop tries to write and gets refused. The useful signals are in idempotency, retry policy, state representation, and side effects around the error.

• pprof Lab and the performance story CPU profiles miss

  2026-09-12

  CPU profiles show where code spends cycles. Block profiles show where goroutines lose time waiting. pprof Lab demonstrates why that distinction matters and how to verify contention before you tune the wrong path.

• Aegean Placement and the value of a proven bound

  2026-09-11

  Aegean Placement shows why a feasible algorithmic answer means more when it sits next to a proven bound. The useful signal is the gap, how to verify it, and where placement systems often lose rigor.

• Backpressure Lab and the engineering of full-queue policy

  2026-09-10

  Backpressure Lab shows three live responses to the same failure mode, a full queue. The useful lesson is not which one wins, but which system property each policy protects, how to verify it, and where overload control often fails in practice.

• WASM Keygrind and the reality of browser-side compute

  2026-09-09

  WASM Keygrind shows what Go compiled to WebAssembly looks like when the browser tab does the work. The useful lesson is the execution model: local CPU-bound compute, static delivery, and the engineering checks you should apply to any browser-side worker.

• Agent Gateway and the engineering of policy at the action boundary

  2026-09-08

  Agent Gateway shows a practical control pattern for AI agents: put policy at the action boundary, edit it as a first-class artifact, and inspect diffs for behavior drift. The engineering value is in clear enforcement, repeatable verification, and reviewable change.

• Access Audit Lab and the engineering of repeatable access review

  2026-09-07

  Access Audit Lab shows a practical pattern for access review, a browser-based audit tied to a real control and built for repeatability. The core lesson is simple: an access check matters more when you can rerun it, inspect its evidence, and track drift over time.

• SBA Lab and the engineering of public risk grading

  2026-09-06

  SBA Lab shows a simple but useful pattern in risk modeling: make a concrete industry-default guess, then grade it against public federal data. The lesson is in the evaluation design, what to inspect, how to verify it, and where category-based credit models often fail.