Category
Product
- ·3 min read
The Deadline That Moved Three Times: Cyberpunk 2077
Cyberpunk 2077's release date moved three times in a single year — April to September to November to December 2020 — and each move was the plan absorbing information the organization refused to respond to. Scope didn't move. Marketing promises didn't move. The gap was paid first in mandatory overtime, then in a launch so broken Sony pulled the game from its store. The closing post of this series, on what 'responding to change' costs when you decline to pay for it.
- ·3 min read
The Spotify Model Spotify Doesn't Use
The most copied agile framework of the 2010s was a 13-page snapshot of one company at one moment — a company that kept changing, whose authors spent years telling everyone not to copy it, and whose own employees later reported it didn't work as marketed. The story isn't that Spotify lied. It's that thousands of organizations bought the org chart and skipped the reasoning.
- ·3 min read
FBI's Sentinel: Green Status Reports vs. Working Software
The FBI spent roughly a decade and over half a billion dollars on two failed attempts to replace its paper case-management system — with reportedly 800-page requirements documents and status reports that stayed green right up until the products turned out not to exist. Then a small in-house team in the basement of the Hoover Building shipped it in about a year for roughly $30 million. The interesting question isn't which methodology won. It's why documentation can lie for years and a sprint demo can't.
- ·3 min read
Backlog Cleanup: How to Actually Do It
A 900-item backlog feels like diligence and functions as denial. Here's the working playbook for rescuing a rotten one: the deletion pass, the now/next/never triage, the hard cap, and the standing hygiene rules that stop it from growing back — plus why the fear of losing important ideas is the least justified fear in product work.
- ·3 min read
Running Fully Agile Without Scrum or Kanban
Scrum and Kanban aren't agile — they're vendors of defaults. Defaults are great when they match your constraints and cargo cult when they don't. Here's how to derive the minimum process your team actually needs from the principles directly, five minimal shapes real teams run, and the honest cases where the frameworks are still the right answer.
- ·3 min read
The Agile Manifesto, Reread: Four Values, Twelve Principles
The Agile Manifesto contains no sprints, no standups, no story points, no backlogs — not one of the ceremonies teams point to when they say 'we're agile.' Reread as engineering advice instead of poster material, it's four priority calls under uncertainty and twelve principles about feedback, people, and simplicity. Everything else is derived — and derived practices decay into ritual the moment they're severed from the principle that produced them.
- ·3 min read
Practicing the Mom Test: The Same Interview, Twice
Reading The Mom Test takes two hours. Actually asking good questions under pressure takes practice nobody schedules. So here's the practice: one customer interview run twice — first the version that produced nothing but polite encouragement, then the rerun that killed the feature in twenty minutes — annotated line by line, with the drills I now use to train the reflex.
- ·3 min read
The Issues List for Innovation: Surfacing Bad News at Line Speed
The V/TO ends with its humblest artifact — the issues list, a running inventory of everything wrong, unresolved, or unsaid, worked weekly through Identify-Discuss-Solve. For innovation it's the load-bearing page: new bets die of suppressed bad news, from the Challenger's normalized O-ring anomalies to every zombie project whose doubts never made it onto an agenda. Toyota's andon cord, Grove's revolving-door question, and Google's Project Aristotle are all versions of the same machine — one that makes bad news cheap to say and expensive to sit on.
- ·3 min read
Rocks for Innovation: The Ninety-Day Experiment Contract
Rocks are the V/TO's smallest unit — three to seven quarterly priorities, each owned by one person, each simply on-track or off-track at the weekly meeting. The one-year plan post covered how rocks get funded; this one covers how they get written and run when the work is an experiment: the rock as a contract about evidence rather than delivery, why the binary status check beats status theater, what OKR grading gets right that rocks should steal, and Parkinson's law as the real argument for ninety days.
- ·3 min read
Marketing Strategy for Innovation: Beachheads and The List
The V/TO's oddest section is a marketing strategy inside a vision document — a target-market List, three uniques, a proven process, a guarantee. For innovation it's the section that prevents the most expensive mistake there is: building for 'the market' instead of a named beachhead. Moore's chasm, Tesla's rich-people-first sequencing, Facebook's campus-by-campus rollout, and Superhuman's 40% test are all the same discipline — and Startup Genome's data says its absence is the leading cause of death.