Newsdesk
Engineering
Data & AI
Industries
Enterprise Systems
Go-to-Market
Longform
DesignIndiaAll stories

Craft & Leadership

17 October 2026

42 new posts across 8 practice and leadership sources — 5 worth your time.

One week of Craft & Leadership, 5 stories, as published.

Testing & Quality

Are Harnesses Worth the Cost?

An empirical experiment evaluated coding agents building an API service with and without behavioral testing harnesses across different LLMs. A frontier model succeeded either way, but a cost-optimized model improved from a 30% success rate without a harness to 100% success with one. The harnessed lower-cost model matched frontier accuracy at the trade-off of higher token usage and execution time.

Why it matters — Behavioral testing harnesses can elevate lower-cost LLMs to frontier-model reliability levels in exchange for higher token consumption and runtime.

Martin Fowler · 17 October 2026 · Read the original →

Incidents & Ops

Uber Automates Dependency Analysis Across Its Service Mesh

Uber built an automated dependency analysis system that leverages RPC middleware and production metrics to classify service dependencies as fail-close, fail-open, or unknown. By correlating caller and downstream failures at the middleware layer, this approach overcomes the sampling limitations of distributed tracing when diagnosing rare failures across a large service mesh.

Why it matters — RPC middleware metrics can automatically classify service dependencies as fail-open or fail-close to diagnose rare production outages without relying on sampled tracing.

InfoQ Architecture · 17 October 2026 · Read the original →

Incidents & Ops

Presentation: How Eve Online Leverages Head Based Sampling to Observe "Fun"

EVE Online manages distributed tracing across a 20-year-old MMO architecture using head-based, dynamic exponential moving average, and deterministic sampling strategies. This telemetry architecture controls data volume and costs while maintaining system fidelity during high-throughput real-time game simulations.

Why it matters — Combining head-based, dynamic EMA, and deterministic sampling allows high-throughput real-time systems to retain critical telemetry fidelity without runaway observability costs.

InfoQ Architecture · 17 October 2026 · Read the original →

Practice

When Software Meets Hardware

Modernizing a large fleet of battery-powered IoT sensors required adapting traditional software engineering practices to severe hardware constraints. Operational experience and failures demonstrated that while core architectural principles hold true, sensible defaults around execution frequency, network use, and power consumption must change.

Why it matters — Adapting software practices to hardware or IoT fleets requires altering operational defaults like power usage and execution frequency while preserving core architectural principles.

Martin Fowler · 17 October 2026 · Read the original →

Leading Teams

Your Team is a System ⚙️ — with Francisco Trindade

Francisco Trindade, VP of Engineering, explains how to think about an engineering team as a complex system. The discussion covers strategies for leading effective software teams by focusing on systemic improvements and organizational structure.

Why it matters — Treat team performance as a systemic outcome to identify structural improvements that go beyond individual performance management.

Refactoring · 17 October 2026 · Read the original →

Another collection of sophisticated solutions for problems we created ourselves.

5 stories, every Saturday

Published here every week. Follow by RSS to get it as it lands.

← Previous issue Next issue →