Current plan, superseded plan, and the gap between plan and reality

What this plan believes

The plan is a control system, not a mileage prescription. Every day it answers one question — proceed, hold, reduce, substitute, or seek review — from a fixed set of rules, and every week it changes at most one thing: volume, long-run duration, or quality. I follow the rules; the rules follow the data; and when my body and the dashboard disagree, my body wins in exactly one direction — symptoms can stop a green day, but no wearable score can clear a red one.

Its priorities are ordered, and the order is the plan: arrive healthy → durability → aerobic fitness → race-specific systems → time goal. The five-hour target is real, but it is fifth. That's why the mileage is deliberately modest next to conventional plans, why the long run is capped by time on feet (3:15–3:30, whichever comes before 28 km — in Singapore heat, pace is an output, not a target), and why a midweek medium-long run exists purely to keep any single run from carrying too much of the week.

Adaptation cuts both ways. The amber rules shorten bad days without drama — but arriving under-trained is also a failure, so quality sessions are the default on green days rather than optional extras, a minimum viable peak (one fueled 26 km run) must exist before the taper, and a Week 13 go/no-go picks the race goal from evidence — a half-marathon result and marathon-pace rehearsals — never from hope. Three amber days in a week triggers a structured review instead of another silent reduction.

And the record stays honest: the plan below is frozen as written, deviations are logged as deviations, device estimates are labelled as estimates, and anything my own controller refuses to act on — single-day heart-rate readings, wearable readiness scores — doesn't get published as if it meant something.

Current plan — v4

Status: current · Effective from: 19 Aug 2026 (mid-Week 2) · Decision owner: Nishanth Sudharsanam, from Dr. Varun Reddy's 19 Aug review · 17 weeks, race day Fri 4 Dec in Week 17.

v4 is Dr. Reddy's ramp: fall back to 20–30 km now, then 15% per week from 20 km to a 50 km peak held for two weeks (Weeks 9–10), then a descent through the half-marathon tune-up (Week 11), the go/no-go (Week 13), and a reduced-volume dress rehearsal (Week 14) to race day. v3's adaptive gates, time caps, and fueling ladder carry over unchanged. No strides while the heel issue is active; strength narrows to hip, core, calf, and knee; a daily stretching protocol runs alongside.

One shared scale: every track runs from zero to the peak week. The tick is the week's target — watch it march right through the build and retreat through the taper — and the dark bar is what was actually run, with the percentage as the completion signal.

Open the full v4 snapshot and week-by-week table

Weekly decisions

The plan is a control loop: every week ends with one recorded decision — progress, hold, reduce, substitute, or review — one lever, and a one-sentence rationale. The published entries appear here; the raw daily capture behind them stays private.

• 19 Aug 2026
  Week 2, Stabilize phase: reduce

  Dr. Varun Reddy's review: two idle weeks then ~40 km in seven days is too sharp a ramp even at easy pace; biomechanical risk outranks cardiovascular risk. Fall back to 20-30 km this week, daily stretching protocol, hip/core/calf/knee strength work, podiatrist assessment to be booked. Flat-route advice partially adopted: slopes walked, not run. Rebuild plan from the same review: 15% per week from 20 km toward a 50 km/week peak (about 20, 23, 26, 30, 35, 40, 46, 50), reaching peak around mid-October. The plan's written 54 km peak is not a target.

  Lever: weekly volume

  Went right: Easy-pace discipline improved (16 Aug run mostly zone 2); symptom notes now flow into the gate engine; three AI expert reviews plus Dr. Reddy converged on the same diagnosis.

  Data confidence: medium.

Superseded plan — v3

Status: superseded 19 Aug 2026 · Effective: 10–18 Aug 2026 · the adaptive control system whose gates, time caps, and fueling ladder v4 keeps; its volume schedule lasted nine days before Dr. Reddy's review and the June–August training history replaced it. Open the full v3 snapshot.

Superseded plan — v2

Status: superseded 16 Aug 2026 · Effective: 1–9 Aug 2026 · the project's starting plan, superseded before its first training week finished; that week (10–16 Aug) is now plan v3's retroactive Week 1.

Open the full v2 snapshot and week-by-week table

Doctor review

Plan vs. actual — this week

Week 8, Build phase. Sessions come from the day-level plan in Supabase, not a hand-edited table; run status reflects approved, published runs only.

Change history

• 19 Aug 2026
  v3 → v4: Dr. Reddy's ramp — rebase to reality, 15% per week, peak 50

  What changed: the weekly volume schedule was rebuilt from Dr. Varun Reddy's 19 Aug review. Fall back to 20–30 km immediately; then 15% per week from 20 km — 20, 23, 26, 30, 35, 40, 46 — to a 50 km peak held Weeks 9–10 (5–18 Oct); then ramp down through the fixed landmarks — half-marathon tune-up (Week 11), go/no-go (Week 13), dress rehearsal at reduced volume (Week 14) — to race day. The minimum viable peak (one fueled ≥26 km long run) moves to the Week 9–10 peak, the only weeks where it fits the ≤55% long-run share rule, with its deadline unified at Week 13. Strides are removed while the heel is active. Strength narrows to hip, core, calf, and knee — drop heel raises are, in Dr. Reddy's words, “specifically a loading stretch” — and a daily stretching protocol (towel stretch, toe extension, calf on a step, calf against a wall) starts. Flat routes, with one documented deviation: park connectors stay, slopes are walked.

  Peak week: 54 km in Week 14 → 50 km held Weeks 9–10

  Why: Garmin history back to mid-June showed the base v3 assumed never existed — one run a week at 7–14 km, mostly above easy intensity, with two multi-week gaps — and Week 1 then tripled the rolling volume; the heel, knee, and calf symptoms followed within days. Dr. Reddy reviewed the history independently of the AI critiques and prescribed the rebase: biomechanical risk outranks cardiovascular risk, so volume is rebuilt on demonstrated tolerance and speed is deferred.

  Evidence/experience informing the change: Dr. Reddy's 19 Aug review; the June–August training backfill and time-in-zone data on the dashboard; three AI-assisted critiques, recorded with their disagreements in the journal.

  Dr. Reddy's view: this change is his view — v4 implements his 19 Aug prescription.

  Affected weeks: 2–17. Week 2 splits mid-week: the first half is recorded as run, the remainder follows v4.

  Outcome: awaiting — the weekly decision log above will record how it plays out.

• 16 Aug 2026
  v2 → v3: the adaptive plan — long run capped, quality committed, gates quantified

  What changed: the whole plan was rebuilt as a decision-led control system from a written framework document, then revised against two structured critiques — an exercise-physiology pass and a coaching pass, both run as AI-assisted reviews. The headline corrections: the long run is capped at 28 km or 3:15–3:30 (v2 peaked at 32 km inside a 42 km week — roughly two-thirds of the week in one session, the classic recreational-marathoner injury signature); a midweek medium-long run from Week 6 lifts peak weeks to ~54 km so the long run stays ≤55% of weekly volume; quality sessions stop being “optional” and become the default when green; marathon-pace blocks and a Week 11 half-marathon tune-up calibrate race pacing; fueling targets move to the current consensus (≥60 g carbs/hour trained across Weeks 9–13, carb load 10–12 g/kg/day); and the green/amber/red rules get numbers — wearable ambers need 3+ days off baseline plus a symptom, and wearable-only signals without symptoms default to an easy run, not a reduction.

  Peak long-run share: 67% of the week (32 of 48 km) → 52% (28 of 54 km)

  Why: the physiology critique showed v2's biggest injury risk was structural — a long run growing 107% while weekly volume grew 35% — and that the adaptive rules as written were so conservative that arriving under-trained was the most likely failure mode of following them faithfully. The coaching critique added what the spreadsheet missed: a tune-up race, a missed-week contingency table, a Week 12 go/no-go, and the fact that “optional” quality means “never” for a tired working adult.

  Evidence/experience informing the change: training-load and injury systematic reviews (Damsted 2018, Fredette 2022), current carbohydrate consensus (Jeukendrup 2014; ACSM 2016), HRV-guided-training literature on rolling baselines, and the strength meta-analysis already cited in v2 — full references in the v3 snapshot.

  Dr. Reddy's view: not yet recorded — his review of v3 is the outstanding milestone.

  Affected weeks: all 17; Week 1 (10–16 Aug) was run before the v3 document existed and is adopted retroactively, recorded as run rather than as planned.

  Outcome: awaiting — the weekly decision log above will record how it plays out.

Plan states used on this site

Each session can be: planned, completed as planned, completed with modification, moved, skipped, stopped, or awaiting outcome. Minor copy fixes don't create a new plan version; anything that changes training does.