99 — Pre-Mortem

“You model failure modes before deployment, not after.”

What This Section Is

This is not pessimism. This is engineering.

Before a system ships, a senior engineer runs a pre-mortem: they imagine the system has already failed, then work backward to identify the most likely cause. This is standard practice at every serious engineering organization. You would never skip it for production software. Do not skip it for a 13-month learning plan.

You are deploying yourself into an ambitious, self-directed curriculum alongside a full-time job. The failure modes are known. The probabilities are estimable. The mitigations are implementable. This section documents all three.

How to use this section:

  1. Read it before you start. Not to be scared — to be prepared.

  2. Return when you feel an early warning signal. Each file names them explicitly.

  3. Treat it as a circuit breaker, not a coffin. Recognizing a failure mode at Signal Level 2 is far better than recognizing it at Signal Level 5, when you’ve already quit.

The 8 Failure Modes — Overview Table

#

Failure Mode

Probability

Primary Mitigation

Escalation Trigger

01

Tutorial Hell & Paralysis

65%

1:1 consume:code rule

Sprint 3 with <3 committed projects

02

Math Phobia / Black-Box Usage

40%

One derivation per week

Can’t explain CE backward pass by Month 2

03

Scope Creep & Perfectionism

55%

72-hour ship rule

3 consecutive sprints with no shipped artifact

04

Comparison & Impostor Syndrome

70%

You-vs-you-2-months-ago metric

Paralysis lasting >1 full study session

05

Burnout & Overcommitment

50%

Buffer sprints + sleep floor

Sleep <6.5hrs for 5+ consecutive days

06

Job Pressure Derailment

35%

Protocol for both variants (offer/rejection)

Complete plan abandonment for >3 weeks

07

Hype Chasing

45%

2-hour frontier scan per sprint, nothing more

>20% of sprint hours on unplanned detours

08

Life & Family Risk

80%

Integration, not sacrifice

You stop showing up for people you love

Note on probabilities: These are estimated from patterns in self-taught ML learning, not clinical data. The numbers are calibrated to be honest, not alarming. A 70% probability does not mean you will fail — it means the failure mode is common enough that you need a named protocol to counter it.

A Note on the 80%

Failure Mode 08 has the highest probability because life is not a controlled experiment. Something will happen in 13 months. A parent’s health. A relationship that needs attention. A period of your own mental health that requires care. This is not a prediction of disaster — it is an acknowledgment of being human.

The pre-mortem for Failure Mode 08 is not a risk mitigation checklist. It is a permission slip.

Read it first if you need to. Read it last if you don’t. But read it.

How the Files Are Structured

Each failure mode file (01–07) follows this format:

  • What it is — precise definition, not vague label

  • Probability — honest estimate with reasoning

  • What it looks like — concrete symptoms at a specific milestone

  • Early warning signals — before it becomes critical

  • Mitigation — the specific protocol to counter it

  • Escalation trigger — the precise condition that means you invoke the reset

File 08 is different. It’s not a checklist. Read it as one.

File 09 is the reset protocol — the decision tree for when you’re off track and need to know what to do next.

The Pre-Mortem Is Your Circuit Breaker

A circuit breaker doesn’t prevent all failures. It prevents a small failure from becoming a catastrophic one. That’s the function of this section.

You don’t need to be afraid of these failure modes. You need to recognize them early, name them accurately, and respond proportionately. That’s all.

The plan is recoverable. You are recoverable. The pre-mortem exists to make that recovery faster.