99 — Pre-Mortem: The Honest Version of the Plan¶

Axiom: Every plan is a hypothesis. This document is the null hypothesis.

Framing¶

It is Aug 5, 2027.

The 13-month roadmap ended. It failed.

Not in the melodramatic sense of “you gave up in month 2” — the plan was too well-scaffolded for that. Failed in the quieter, more common sense: the artifacts got built but the outcome (the inference role, the sustained OSS presence, the internal Zoho pivot) did not materialize. Or: three of the eight phases got done at merit-badge depth and the other five got done at wikipedia depth. Or: you shipped the mini-engine but nobody read it. Or: month 8 was a shell of a month and you never quite recovered the tempo.

Your job now, sitting in that August 2027, is to work backward. What actually killed it? What did we not see coming when we wrote the beautiful 366-line roadmap in July 2026? What was the plan wrong about?

This document does that work now, before the failure, so it can be prevented — or so that when it starts to happen, you recognize it fast enough to parry.

The Method (Klein / Kahneman)¶

Gary Klein’s pre-mortem is a decision-quality tool. Kahneman calls it one of the highest-leverage cognitive interventions available to anyone making a large committed bet. The procedure is dead simple:

Assume the plan has already failed.
Generate reasons why — as many as possible, adversarially, no soft-pedaling.
For each reason, define the early warning signals and the parry.
Fold the parries back into the plan.

The value comes from prospective hindsight — imagining a specific bad future outcome makes the mind reach for concrete causes rather than vague concerns. “The plan is risky” is useless. “In month 6, Zoho launched a big on-prem product push, you were on-call, you missed three sprints, you never recovered the SGEMM ladder momentum, and by month 9 you’d stopped opening the roadmap” is actionable.

That’s what the eight vector files do.

The Eight Failure Vectors¶

Each file examines one dominant way this plan dies. They are not independent — a real failure will be a chord of two or three struck at once — but decomposing them lets you build specific parries.

#	Vector	Class	Modal timing
01	Industry moves faster than the plan	Technology	Months 4–10
02	Zoho starves the roadmap of time	Employer	Months 4–8
03	You burn out in the Valley of Despair	Human	Months 6–8
04	Shiny object syndrome — never finishing	Attention	Continuous
05	OSS community rejection or picking the wrong horse	Political	Months 8–12
06	Hardware access chokes at the wrong time	Infrastructure	Months 3–7
07	Geography / visa / offer market doesn’t cooperate	Career-outcome	Months 12–18
08	Family, health, life	The unspeakable	Any time

Then 09_summary_and_reset_protocol.md — what to actually do when one of these hits. Because one of them will hit. That is not pessimism; it is arithmetic on an 18-month wall-clock with a human being in the loop.

How To Use This Folder¶

This is not a document you read once and file away. It is an operating manual.

At the start of each phase: re-read the two or three vectors most active in that phase (see the modal-timing column above). File 06 matters most at Phase 2. File 03 matters most between Phase 3 and Phase 4. File 07 matters most in Phase 7 and after.
In every 2-week sprint retro: run a 60-second mini-pre-mortem — “what could kill the next sprint?” — and check it against these vectors. If the answer starts to sound like one of these files, you have advance warning that most people never get.
When something feels wrong: open 09_summary_and_reset_protocol.md first, before opening the roadmap. The reset protocol is designed to be run under duress; the roadmap is not.
When a vector hits: log the incident (date, vector #, what you did, what worked, what didn’t). Over 13 months this becomes the most honest artifact you produce, and — not incidentally — it is a phenomenal portfolio piece for a staff-level study. “Here’s what tried to kill my learning plan and how I handled it” tells hiring managers something no benchmark repo can.

The Tone of This Folder¶

Two things are true at once:

The plan is a Viking commitment. You signed up for 13 months of hard, unforgiving, mostly-solo work. This folder does not undermine that. It hardens it.
The plan is not you. The roadmap serves you; you do not serve the roadmap. When the plan and reality diverge, reality is right. This folder is the permission slip to adjust without shame.

The voice in these files is adversarial about the failure modes and warm about the human. That is the correct combination. A mentor who only cheers you on is useless when the wheels start coming off. A mentor who only warns you is exhausting. The pre-mortem tries to be the friend who does both — the one who says “this specific thing will try to hit you in month 7, here’s the parry, and by the way, if it hits harder than you can parry, call me, we’ll figure it out.”

The Meta-Rule¶

If, at any point in the next 13 months, you find yourself in a situation not addressed by these files — a failure mode we didn’t see coming — write it down here. Extend the pre-mortem in real time. The point is not that these eight vectors are exhaustive; the point is that pre-mortem thinking is a discipline you carry forward.

Every failure that ships you a lesson you paid for in months of your life is a failure worth naming.

The One-Sentence Version¶

You have not committed to the roadmap. You have committed to the underlying skill. The roadmap is one route; if the route fails, take another route to the same summit.

Now go read file 01.