09 — Summary & Reset Protocol

This file is the one you open when something has broken and you need to know what to do next.

The Reset Decision Tree

The Reset Protocol (Red-Flag Response)

When two sprints in a row miss their core target, or when a life event forces a pause, run this protocol. Do not improvise. Follow the steps in order.

Step 1 — The Sunday Audit (30 minutes)

Sit somewhere quiet. Phone in another room. Paper notebook open. Answer, in writing:

  1. What happened? One paragraph. No blame, no excuses. Just events.

  2. Which failure mode fired? Match against the 8 modes in this directory. Usually it is one of two: Zoho creep (04) or AI rebound (02). Sometimes it is life (08).

  3. What did I feel this week? One sentence. Name the feeling. “Tired.” “Ashamed.” “Behind.” “Confused.” Name it.

  4. What’s the smallest possible next move? Not “get back on track.” The literal smallest step. A 15-minute session tomorrow. One commit. One cppreference page.

Step 2 — Read the Corresponding File

Open the failure mode file matching Step 1.2. Read it fully. Do not skim. It was written for exactly this moment.

If the failure mode is life (08) — read that file twice. Take a walk between readings.

Step 3 — Cut Scope by 50%

Whatever the next sprint had planned, halve it. This is not negotiable and it is not proportional to how you feel. The rule is 50%. If you feel fine, still 50%. If you feel awful, still 50%. Consistency beats calibration in the moment right after a break.

Concretely:

  • If the next sprint had 3 targets, drop to 1–2.

  • If the target was “finish P3.2 concurrency chapter + write benchmark + blog post” — drop to “finish P3.2 concurrency chapter.”

  • Move the dropped work to the following sprint, not this one. Do not double up.

Step 4 — Ship Something Tiny Within 7 Days

This is the most important step. Momentum returns through shipping, not through planning.

A “tiny ship” example:

  • One .cpp file that compiles and runs and does one small thing.

  • One commit to any repo, however small.

  • One blog post, even 200 words on “what I learned this month.”

  • One benchmark run committed with a note.

The ship does not need to be impressive. It needs to be public and recent. The repo commit graph shows a green square this week. That is the whole ask.

Step 5 — Return to Normal Cadence

The sprint after the tiny ship runs at the reduced (50%-cut) scope. If that succeeds, the following sprint returns to full scope. Do not skip the intermediate reduced-scope sprint. The step-down / step-up is the rehab.

Step 6 — Log the Reset in the Lab Notebook

Write one entry:

RESET LOG — [date]
Trigger: [which failure mode fired]
Scope cut: [what was dropped]
Tiny ship: [what was shipped in the 7-day window]
Lesson: [one sentence]

This is not for accountability theatre. It is data. After 13 months you may run 2–4 resets. Reading them back reveals patterns — which modes fire most often for you, which mitigations actually work, which weeks of the year are risk zones. That data is worth writing down.

The Escalation Ladder

Flag

Trigger

Response

Green

Sprint complete

Continue

Yellow

1 sprint core target missed

Cut next sprint’s stretch

Yellow-Red

2 core targets missed in a sprint

Long-form Sunday audit

Red

2 sprints in a row missed

Full reset protocol (this file)

Deep-Red

3 sprints in a row missed

1–2 week pause + reset

Re-plan

Pause exceeds 4 weeks

13-month plan becomes 15–18 months

Do not skip flags upward. Yellow does not require a full reset. Red does. Deep-red requires the pause. Escalating too aggressively causes plan whiplash. Escalating too slowly causes silent collapse.

The Six-Month Reduced Plan (Emergency Only)

If a Deep-Red escalation happens after M9, and the M9 hard gate (rung 6) is not shipped, the roadmap collapses to a six-month reduced arc. This is described in 05_the_hard_gate_M9_miss.md but restated here for the file you’ll actually open in a crisis:

  • Month 1 of reduced plan: ship rung 6 (Eigen MLP + pybind11), reduced scope (2-layer, no CUDA, no PyPI).

  • Month 2: polish rung 6, write one blog post, add benchmarks.

  • Month 3: ship rung 7 — one small applied inference tool.

  • Month 4: capstone scope drops to a single small end-to-end project.

  • Month 5: capstone ships as MVP. LinkedIn optimization. First applications.

  • Month 6: study conversion begins.

The pitch changes from “13-month applied C++ + ML” to “focused 6-month applied ML integration story.” Still a real pitch. Still lands. Just narrower.

Things That Are Not Failure

Restating, because the emotional weight of a reset lies to you:

  • Missing a sprint to a family emergency is not failure.

  • Missing a sprint to a wedding is not failure.

  • Missing a sprint to fever, migraine, or a bad-sleep week is not failure.

  • Cutting scope is not failure.

  • Stretching the plan from 13 to 15 months is not failure.

  • Not getting an offer by M13 is not failure — that is the study conversion phase running on its own timeline.

The only actual failure is stopping and not coming back. Everything else is pacing.

The Closing Line

Every file in this pre-mortem directory ends the same way. Every crisis, every reset, every sprint retro can return to this one line and find the ground under it.

The roadmap serves you. You do not serve the roadmap.

Come back on Monday. Type two lines. Then two more. That is how it rebuilds.

Nav: ← 08_the_life_risk.md · README · ← 13_discipline · Roadmap root