#41 | Sunday reads for EMs

My favourite reads of the week to make your Sunday a little more inspiring.

Stephane Moreau

Apr 19, 2026

👋 Hey, it’s Stephane. Every Sunday I share with you my favourite reads of the week.

To accelerate your growth see:

Paid subscribers get 50 Notion Templates, The EM’s Field Guide, and access to the complete archive. Subscribe now.

What Agent Orchestration Actually Changes: The Hidden Costs Behind Shared Understanding (Viktor Cessan)

tl;dr: Most of what managers do is “negentropy” - preventing shared understanding from degrading in three specific ways (state, semantic, verification). Agent orchestration is a phase shift that decouples output from coordination cost, not just a productivity bump.

The Economics of Software Teams: Why Most Engineering Organizations Are Flying Blind (Viktor Cessan)

tl;dr: A team of eight costs ~€87k/month. Break-even isn’t the right bar - a 50% initiative success rate means you need 3–5x cost in value to be financially viable, and almost nobody on the team knows these numbers.

15 Principles for Managing Up (Wes Kao)

tl;dr: A dense, tactical playbook with concrete before/after scripts for each principle. Lead with the punchline, show your thought process with conviction levels, bring solutions not complaints, proactively assert what to do rather than ask.

Mechanical Sympathy (Vicki Boykis)

tl;dr: The best engineers are also good product designers because both skills come from “mechanical sympathy” - an intuition for the grain of the system, built up over a human lifetime of context, that coding agents fundamentally lack. Agents fix code that already works, rewrite tests to assert wrong outputs instead of fixing the underlying bug, skip modern tooling defaults (uv, vectorized ops), and erode under iterative refinement - none of which are fatal alone, but collectively produce software that feels “hard” to develop.

Developer Ramp-Up Time Continues to Accelerate with AI (Justin Reock)

tl;dr: DX’s quarterly data shows Time to 10th PR dropped to 33 days in the latest cut (from 39 in Q4 2025, and 91 pre-AI), with the biggest driver being orgs now guiding new hires to use AI for codebase Q&A and draft changes rather than leaving it optional. Justin flags the honest limitation himself that 10th PR tells you nothing about code quality, rework rate, or whether you’ve just shifted the burden onto senior reviewers.