Defence Dry-Run Log¶
Rehearsal tracker for the dissertation defence (target date 2026-05-19). One row per practice run: date, duration, which slides ran long, which Q&A came up, which answer did not land cleanly, what to fix before the next run.
The point of this log is not bookkeeping — it is committee-proximity is the single biggest predictor of defence outcome, and practice is only useful if it's deliberate. A rehearsal without written diagnostics is a rehearsal wasted.
Protocol for each dry run¶
- Time-box. 40-minute talk target; allow 45 minutes; stop at 50 regardless.
- Audience. Solo first (T-14 through T-28 days), then 1-on-1 with advisor (T-7 through T-14), then committee-like panel (T-3 through T-7).
- Record. Video yourself if solo; audio if panel. Review the recording within 24 h.
- Diagnostics written within 30 minutes of the rehearsal ending. Memory fades fast.
- Acceptance criterion. The rehearsal is successful if you identified at least two concrete interventions for the next run. If none, the practice was performative, not diagnostic.
Slide-level timing target¶
The 24-content-slide target at 40 minutes = ~100 seconds per slide average. Some slides take 30 s (a section divider); some take 3 min (Act 2 evidence with visual). Track per-slide timing on at least two runs to identify which slides consistently run long.
Q&A preparation¶
Each dry-run should include 3–5 anticipated questions drawn from the Q&A backup slides and the Reviews / J2, Reviews / J3, and Reviews / V-MSSP trackers. If a question from a reviewer tracker comes up in the rehearsal, log which reviewer identifier (R1.1, R2.a, R3.8, etc.) and whether the answer came smoothly.
Log entries¶
2026-04-18 — Rehearsal 0 · Agent diagnostic read-through¶
This is not a speaking rehearsal. It is a cold-read diagnostic of the deck's .qmd source, scored by speaker-note word count, body density, and argumentative coherence. Purpose: identify which slides need script expansion or compression before the first real spoken rehearsal. Automated via a small script that parses the qmd for ::: {.notes} blocks and slide bodies.
Corpus: presentation-source/defence_v3.qmd, 24 content slides + 3 Q&A backup + title. Analysis assumes a speaking pace of ~150 words/min (conservative — typical is 130–160).
Headline metrics¶
| Measure | Value | Interpretation |
|---|---|---|
| Total speaker-notes speaking time (as-written) | 12.0 min | ~3× short of a 40-min defence target |
| Slides running long (> 120 s of notes) | 0 | No single slide is over-scripted |
| Slides running short (< 40 s of notes) | 13 of 24 | Speaker notes are a skeleton, not a script |
| Slides with dense body content (> 200 words) | 2 of 24 (both Q&A backup) | Reading-heavy — appropriate for Q&A reference, not for presentation |
Primary diagnosis. The speaker notes are a floor, not a ceiling. If the speaker reads the notes verbatim, the talk lands around 12 minutes — way below target. For a 40-min defence, each slide needs roughly +50–60 s of extemporaneous commentary beyond the written notes. For a 30-min defence, roughly +30 s per slide. Either is feasible with practice; without practice, the talk under-fills.
Per-slide attention list¶
Slides needing script expansion (< 40 s notes):
| # | Slide | Notes (s) | Recommendation |
|---|---|---|---|
| 1 | 국문 초록 | 16 | Korean abstract; short speaker notes are fine because the slide body is dense. OK as-is. |
| 4 | One Question, Four Sub-Questions, Ten Papers | 36 | Critical slide — this sets up the entire argument spine. Needs +40 s explaining why these four sub-questions partition the problem naturally. Expand. |
| 8 | Evidence 4 · Probabilistic Capacity | 32 | Headline is "30 % below deterministic." Needs +30 s on why this matters for asset owners. Expand. |
| 9 | Evidence 5 · 32 Months of Field Data | 39 | Longest published record — worth pausing for effect. Needs +30 s on what "22,617 windows" represents operationally. Expand. |
| 14 | What Changes for Engineers | 34 | Photograph-worthy slide; the "before vs after" table speaks for itself but the speaker needs +30 s to walk through the specific cost-reduction claim. Expand. |
| 15 | Limitations | 29 | Skeletal. Needs +30 s — committee probes limitations hard; honest expansion here is defensive gold. Expand. |
| 17 | Future Work | 29 | Three directions listed; needs +30 s to tie each to the portfolio arc. Expand. |
| 18 | One Thing to Remember | 26 | The callback closer. Must feel like landing, not restating. Needs +30 s of pause-and-delivery. Expand with deliberate pacing. |
| 19 | 감사합니다 | 28 | Thanks slide; 30 s is correct. OK as-is. |
| 20–22 | Q&A backup | 1–2 | Intentionally minimal — slides are reference, not scripted. OK as-is. |
| 23 | Questions? | 2 | Closing CTA. OK as-is. |
Slides running within target (40–80 s notes): 2, 3, 5, 6, 7, 10, 11, 12, 13, 16. Eleven slides on target.
Q&A backup density¶
Slides 20 and 21 are dense (>200 body words each). They are reference cards, not presentation slides — the speaker is meant to read the anticipated question quickly and deliver the answer from memory. Verdict: density is appropriate for reference but problematic if the speaker tries to read them live on screen. Cross-reference against the J2 R2 tracker reviewer items (R1.1–R1.2, R2.1–R2.6, R3.1–R3.8) to confirm every reviewer question has a corresponding Q&A answer in the backup.
Argumentative coherence checks (content, not timing)¶
- Act 1 → Act 2 transition (slides 5 → 6): The portfolio arc is introduced on slide 4 and the first evidence block on slide 5. The transition works; speaker should pause briefly at slide 4 to let the spine land before diving into centrifuge.
- Mid-deck energy dip risk (slides 10–12): Three consecutive "evidence" slides (cointegration failure, feature selection, cross-soil transfer) without visual variety. Speaker notes do not suggest a pacing shift. Recommendation: consider a brief physical cue — stepping back from the podium, pausing, or the chalkboard shortcut B — to break the pattern.
- Slide 18 callback: One Thing to Remember quotes the teaser on slide 2 verbatim. Speaker notes on slide 18 say "I opened with the destination. I close with it." This is the single highest-leverage delivery moment in the entire talk. Recommendation: rehearse this transition specifically; the pause before reading the callback matters more than the words themselves.
Q&A readiness against reviewer trackers¶
Cross-checked Q&A backup slides 20–22 against the reviewer items in J2 R2, J3 R1, and V-MSSP trackers. Coverage:
| Tracker | Items | Covered in Q&A backup |
|---|---|---|
| J2 R2 | 19 | ~8 (methodology + scope probes overlap heavily with R1.1–R3.8) |
| J3 R1 | 18 | all closed, low Q&A probability |
| V-MSSP | 15 | ~4 (scope + downstream probes on slide 21) |
Gap: no explicit Q&A answer prepared for R2.5 (tripod load-sharing quantification) or R3.1 (stiffness-scaling sensitivity). These are the two items the J2 methodology and results prototypes independently flagged as needing new figures. Recommendation: once those figures are generated (R2.4 / R2.5 / R2.2 figure batch), add a one-line answer to Q&A slide 20 referencing them.
Actionable items before Rehearsal 1¶
| # | Item | Owner | Cost |
|---|---|---|---|
| 1 | Expand speaker notes on slides 4, 8, 9, 14, 15, 17, 18 — add +30 s each | KSK | 2 h total |
| 2 | Practice the slide-18 callback pause — rehearse the 5 s before reading | KSK | 30 min |
| 3 | Add Q&A answer stubs for R2.5 and R3.1 on slide 20 | KSK | 30 min |
| 4 | Do Rehearsal 1 (solo, video-recorded, full talk) once #1 is done | KSK | 1 h |
When Rehearsal 1 (first speaking rehearsal) happens, log it as a NEW entry below with per-slide timing from the recording. The word-count estimate above is a lower bound; the actual timing under speaking conditions will likely run 1.2–1.5× longer due to pauses, emphasis, and transitions.
Next rehearsal — proposed¶
Rehearsal 1 — Solo video-recorded walkthrough.
- Target: after Item 1 above is complete (speaker-note expansion on 7 slides).
- Duration target: 30–35 min (conservative first pass; improving toward 40–42 on subsequent rehearsals).
- Record: video + screen; review within 24 h.
- Diagnostics to capture: per-slide stopwatch (the Speaker View timer in reveal.js at S is enough), each slide's actual duration, any slide where the speaker improvised > 30 s beyond the written notes, any slide where the speaker ran out of words in < 20 s.
- Q&A practice: ~10 min at the end, drawing 3 questions from the backup slides and 2 from the reviewer trackers.
Running tallies¶
Fill in as rehearsals accumulate:
| Metric | Target | Current | Trend |
|---|---|---|---|
| Best-case duration | 38–42 min | — | — |
| Worst-case duration | ≤ 50 min | — | — |
| Slides consistently running long | ≤ 3 | — | — |
| Q&A unanswered smoothly | 0 | — | — |
| Days of slide content last edited | — | 2026-04-18 | — |
Links into the rehearsal material¶
- Deck:
defence_v3/index.html— the renovated three-act arc. - Q&A backup slides: slides 25–27 of the deck (Methodology · Scope · Downstream).
- Reviewer tracker (for Q&A content): J2 R2, J3 R1, V-MSSP.
- Gap framing: Dissertation gap analysis — defensive / offensive / constructive layer per paper.
- Claim-driven workflow: Claim-driven workflow — the writing methodology referenced in the deck's Act 2 framing.
Meta-reflection¶
After every third rehearsal, answer these three questions in writing:
- Is the deck as it stands the right structural shape, or does the three-act arc need to change? (If same answer three rehearsals in a row: restructure.)
- Is the teaser on slide 3 earning its slot? (If the answer "they understood the destination immediately" comes up twice, keep it. If not, either the teaser needs a different visual or it needs to move.)
- Are the limitations (slide 19) landing honestly or defensively? The committee respects honest limitations; they probe defensive ones. If the speaker notes feel hedged, rewrite.