Tech

Private complaints:"Ralph smells bad" Placating Tell the complainers to put up with it→ someone quits Blaming "You stink — deal with it"→ Ralph hurt, still doesn't know why Super-reasonable All-hands email: "mind your hygiene"→ everyone feels accused — except Ralph Irrelevant Do nothing, wait for it to pass→ it gets worse Congruent Talk to Ralph — privately, directly→ Ralph sees a doctor, and says thanks Dots = self (top) / other (left) / context (right); filled = attended to, hollow = ignored

Leadership — Power and Congruence

· tech · 21 min read · 📚 Becoming a Tech Leader — Reading Notes #11

The last two chapters of the motivation part (Ch14–15) are about power — but the power Weinberg mean…

#leadership

Each comparison can only split the remaining permutations in two a < b ? b < c ? a < c ? < > a b c a < c ? b a c b < c ? < > < > a c b c a b b c a c b a < > < > n! leaves → height ≥ log₂(n!) ≈ n log n → at least n log n comparisons in the worst case

Sorting Is the Cheapest Structure: The Comparison Lower Bound, Counting Sort, and Binary Search

· tech · 22 min read · 📚 Competitive Programmer's Handbook — Reading Notes #2

Part one was about reading $n$ and knowing which complexity to write. This part covers CPHB chapter …

#algorithms#cphb#sorting#binary-search#book-notes

1-second limit ≈ a few hundred million ops: read n, infer the complexity input size n acceptable complexity typical approach operations n ≤ 10 O(n!) try every permutation 10! ≈ 3.6×10⁶ n ≤ 20 O(2ⁿ) try every subset 2²⁰ ≈ 10⁶ n ≤ 500 O(n³) three nested loops 500³ ≈ 1.3×10⁸ n ≤ 5000 O(n²) two loops, every pair 5000² = 2.5×10⁷ n ≤ 10⁶ O(n log n) or O(n) sort, or one pass 10⁶×20 = 2×10⁷ n is large O(log n) or O(1) binary search, a formula — e.g. n = 10⁵ with O(n²) → 10¹⁰ ops, tens of seconds — known before writing a line

Read n, and You Know What to Write: Time Complexity and the Maximum Subarray Sum

· tech · 25 min read · 📚 Competitive Programmer's Handbook — Reading Notes #1

A new series: reading Antti Laaksonen's Competitive Programmer's Handbook (CPHB). It's the companion…

#algorithms#cphb#complexity#book-notes

231 CI runs Green 194 Cancelled 22 Red 15 The 15 red runs, by cause Tests 8 Lint 4 Infrastructure 3 mypy 0 What the 8 test failures caught Silent bug ×3 state reads CLEAN One OS ×2 confirmed One OS ×2 log expired flaky All 15 on PR branches Reached master: 0 "Log expired": we know which OS went red, not which test. The flaky one is the only PR merged on red. 22 cancelled: 21 master runs superseded by a newer merge, 1 PR run superseded by a new push.

Guardrail Engineering: The New Role of Tests and Specs on an AI Production Line

· tech · 26 min read · 📚 The Craft of Working with AI (2026) #7

The first post defined guardrails as "tools for widening the neck" — hand the checks a person would …

#ai#testing#side-project

Author, blind Bare AI +CLAUDE.md +review skill B1 reversed index logic boundary △ "no test pins it" not verified 3/3 3/3 3/3 B2 Apply → Pop silent regression ✓ 3/3 3/3 3/3 B3 layer bypass architecture call ✓ 1/3 2/3 3/3 Unplanted real bug half-done, says "error" ✗ 1/3 1/3 3/3 Cost 20 minutes 0.4–2.2 min, $0.09–0.19 per run How B1 was checked Read the diff 8 of 9 tested pygit2 with a script

The Craft of Acceptance: How to Review AI's Code

· tech · 29 min read · 📚 The Craft of Working with AI (2026) #6

The first post said the width of the neck is the bandwidth you have for accepting work responsibly; …

#ai#code-review#leadership

No spec With spec Same six source files, same layers, nearly the same line count The existing codebase is an implicit spec — both runs grew along it Code Responsibility layer: every difference is here Shipping path Commits straight to master Opens a branch, waits for a PR Publish gate draft: false — ships as written draft: true — waits for the author Experience Writes "in my experience…" Refuses; flags it for the author Acceptance Unit tests; app never launched Actually launches the app (e2e) Splitting One commit Two commits, split by layer A spec changes where responsibility lands, not what the code looks like

Spec as Code: CLAUDE.md Is the New Onboarding Doc

· tech · 37 min read · 📚 The Craft of Working with AI (2026) #5

The first four posts laid down axioms: conservation of responsibility, a clock on the neck, how tool…

#ai#leadership#side-project

What you think it is (people as computers) Instruction = Action What actually happens (people) What you saidyou think it's clear ≠ What they heardhalf of it ≠ How they read itin their frame ≠ How they feelthreatened? trusted? ≠ What they doall you see Only the two ends are visible — the three steps between happen inside their head

Leadership — The Two Great Obstacles to Motivating Others

· tech · 14 min read · 📚 Becoming a Tech Leader — Reading Notes #9

The series moves into part three: motivation — the M in MOI. The innovation part was about ideas and…

#leadership

How many reports you have decides what you actually do Time left to invest (coaching · strategy · driving change) peak 1–3 reports You're a TLM still designing and coding 4–5 Barely a team cover is thin 6–8 Sweet spot: investor coach · strategy · change 9 and up Coach + safety net all catching, no investing Number of direct reports → One level up, same axis: a manager of managers supports 4–6 managers (past that, even "coordinating managers" turns into pure forwarding)

How Big Should a Team Be: What the Number 6-to-8 Is Buying You

· tech · 25 min read · 📚 An Elegant Puzzle — Reading Notes #2

"How many people should a team have" sounds like an HR question. It's actually the book's first and …

#engineering-management

Working the events Catch them faster, get more of them System (org · process · unwritten rules) You Every one of them handled ✓ Same load next week — the source never moved Changing the system Third time? Stop catching it System + one policy / one mechanism You Only these need your judgment ✓ The cost: weeks to months before it shows

Engineering Management Is an Elegant Puzzle: Fix the System, Not the Events

· tech · 26 min read · 📚 An Elegant Puzzle — Reading Notes #1

There's already a leadership series on this site (Becoming a Tech Leader), so why start another one?…

#engineering-management

1 Offer ideasyours — one of five 2 Steal ideasborrow and adapt 3 Work on theirsfinish the half-baked 4 Let yours gomake room for better Ideafragile the team's idea flow Innovationshipped 5 the umbrella — don't let ideas get killed early "we tried that" "that won't work" "not how we do it"

Leadership — Idea Power

· tech · 19 min read · 📚 Becoming a Tech Leader — Reading Notes #7

Earlier we covered the three walls that block innovation, and using a journal to take down the first…

#leadership#innovation

A team of people: a reinsurance net CTO: external cover the face; reputation claims PM: covers scope wrong scope is their claim Peer: bidding fully retained Peer: frontend fully retained Me: backend the poison parser retained by layer, excess sent up — nobody faces all of it alone One person + agent fleet: a waterfall One person sole underwriter, all policies agent agent agent agent 100% reflects back, nobody absorbs — output of a team, cover of one

The Responsibility Premium: AI Doesn't Collect It, and Doesn't Pay Out

· tech · 29 min read · 📚 The Craft of Working with AI (2026) #4

This post starts from an observation so plain it's nearly a tautology: hire an employee and they are…

#ai#leadership

Multica: neck downstream (dashed) AI agent output claim issues · write code · comment in_review: human check a convention — not enforced done agent → done (no alert fires) the auth layer can tell human from machine — but the lock is only on the billing API Superpowers: neck upstream (solid) Human: sign the spec HARD-GATE: no code before approval AI implements (TDD · subagents) plan written for a zero-judgment executor deliver after verifying no evidence, no completion claim the signature is a committed specs/*.md — physical responsibility; weakness: sign-off fatigue

The Neck on the Board: How Tools Design Human Responsibility — Auditing Two Open-Source Projects

· tech · 30 min read · 📚 The Craft of Working with AI (2026) #3

Part one set up the model: AI eats the wide layers, a human holds the neck. Part two threw the model…

#ai#leadership

⏱ a clock on the neck: the host is shouting, time is running Symptoms / evidence metrics · logs · what you checked AI: narrative + handle hypotheses · checks · draft comms Human: run the check sample · align timestamps Human: decide and act restart · hotfix · backfill results feed back once it converges believing the narrative = the authority shortcut

A Clock on the Neck: AI During an Incident — Replaying a Poison-Message Outage

· tech · 38 min read · 📚 The Craft of Working with AI (2026) #2

Last time was the funnel: AI eats the volume in the wide layers, a person holds the responsibility a…

#ai#incident#sre

The cost of interrupts: not time, but fragmentation work ramp-up interrupt ① a fragmented day complete deep work ≈ none (all fragments) ② interrupts batched into one block uninterrupted deep work (one whole block) interrupts same total interruption → deep work = one whole block An interrupt's cost isn't the time it takes; it's fragmenting what's left until deep work is impossible

Operational Interrupts: What Kills Productivity Isn't the Workload, It's Fragmented Time

· tech · 10 min read · 📚 Google SRE — Reading Notes #17

The on-call post covered "how to design alerts and who carries the pager"; the toil post covered "th…

#sre#reliability

PRR: a gate before launch Dev builds the service features OK ≠ launch-ready PRR · Production Readiness Review ① SLO defined? ② monitoring + alerts (golden signals) ③ load test / capacity / load shedding ④ can a release roll back fast? ⑤ a dependency dies → degrade? ⑥ runbook: usable at 3am pass fail SRE takes over shares the pager sent back to fix pager stays with Dev SRE won't take an "unoperable" service — PRR makes reliability a hard launch gate

Production Readiness Review (PRR): What Makes a Service Worth SRE Taking Over

· tech · 13 min read · 📚 Google SRE — Reading Notes #16

The previous fifteen posts almost all talked about "the service is already live; how do we make it m…

#sre#reliability

Wouldn't change (the core) the FSM · lookup-as-validation append facts + derive on read batches that silt but don't fall 3NF · three-number stock ledger per-provider payment fact tables five boring components · one VM order the queue, don't schedule · merge = release transcoding-as-validation · generic FKs Would add (protection) four golden signals + a batch lag gauge a language for incidents: severity · runbooks a dead-letter queue for poison pills Cloudflare on day one scheduled invariant queries (self-reconciliation) deleted_at · a reverse lookup before sweeping load testing · the other half of the checklist — all protection, not one of them a feature The core was right; everything owed was protection Which also explains why I later moved towards SRE

Re: If I Really Started Over

· tech · 26 min read · 📚 Re:Building a Live-Commerce Platform from Zero #21

This series started with the journey of one comment and ran for twenty chapters. The finale adds no …

#war-story#live-commerce#retrospective

Station 1: one instance + tenant_id tenant_id on every table · leads every composite index the terminus for most SaaS trigger to move on: connections, data volume, noisy neighbours — not "it feels time to distribute" Station 2: read/write separation replicas absorb reporting and storefront reads writes still single-point; this station buys time trigger: reads overwhelm writes, reports disturb transactions Station 3: pool + silo hybrid small merchants share a pool shard · big hosts get their own DB the DB-layer answer to noisy neighbours = enterprise tiering per-tenant backup/restore = the silo's killer advantage upgrading = a move: one merchant down for export/import, affects only them, done late at night Station 4: real sharding Citus-style shard by tenant_id / NewSQL needed only when the pool itself must scale out honestly: most live-commerce SaaS never reach this station Every station has a trigger — untriggered, you stay put; evolution is a response, not a hobby

A Parallel World: What If It Had Become a SaaS

· tech · 33 min read · 📚 Re:Building a Live-Commerce Platform from Zero #20

First, something nineteen chapters never mentioned: back then, every one of us took a pay cut to joi…

#war-story#live-commerce#system-design