> ✅ **CORRECTED 2026-08-25** by decoding the IXUD containers directly — see
> [structures/ixud-localised-text](structures/ixud-localised-text.md).
> Two errors here: `SUBOBJ_009` **does** have text (`'You destroyed all enemy
> units!'` — all 91 of 91 sub-objectives have an `_Mes_L1` string), and
> `SUBOBJ_005` is **two** fields (`'You rescued'` + `'the ally destroyer!'`),
> which a single-line rendering drops.

# The mission script in plain English — and why phase 1 never completed

Status: ✅ the localised string tables are decoded and readable; ✅ the phase
structure is confirmed narratively *and* by roster composition; 🟡 a concrete,
testable reason the phase never advanced in any run.

## ✅ `IXUD` — the localised string container

`language\*_local_string.tbl` (the prefix is `language\`, a third convention
after `stage\` and `message\`) is **not** IDXD. Magic is `IXUD`, and its strings
are **UTF-16 big-endian**. That is why every earlier ASCII-oriented dump of these
files produced garbage.

Reading them is immediate once the encoding is known, and it makes the mission
design directly legible instead of inferred.

## ✅ Stage 02's nine sub-objectives, in words

| id | text |
|---|---|
| `SUBOBJ_005` | You rescued the ally destroyer! |
| `SUBOBJ_006` | You destroyed all enemy fighters! |
| `SUBOBJ_007` | You sunk all enemy warships! |
| `SUBOBJ_008` | You rescued the ally cruiser, Amalthea! |
| `SUBOBJ_010` | The ally cruiser CALIBAN took no damage! |
| `SUBOBJ_011` | You sunk the enemy destroyer targeting the ACROPOLIS! |
| `SUBOBJ_013` | You destroyed all enemy cruise missiles! |
| `SUBOBJ_014` | You destroyed all enemy missiles and units! |

(`SUBOBJ_009` has no text.) 91 objectives are defined across the game.

## ✅ The three phases, confirmed twice over

The guide script reads as three distinct acts, and each matches the per-phase
roster composition found independently in
[mission-phase-deployment.md](mission-phase-deployment.md):

| phase | script | roster |
|---|---|---|
| 1 | *"The attackers with the orange markers, right?"*, *"Protect the ACROPOLIS!"* | turrets, attackers, fighters |
| 2 | *"destroy the ship's engine"*, *"destroy that ship's weapons"*, *"take out the enemy's shield projector"* | destroyers, frigates, cruisers |
| 3 | *"start by taking out that missile"*, *"Concentrate on destroying the missiles one by one"* | **`UN_e201_ADAN_ISCMissile` ×9** |

Phase 3's nine cruise missiles against `SUBOBJ_013` *"You destroyed all enemy
cruise missiles!"* is an exact match between two files decoded on different days
by different routes.

## 🟡 Why phase 1 never advanced: the pilot killed the wrong things

The phase-1 objective is stated outright — **destroy the marked attackers**:

> *"Rhinos, prepare for combat. Are you all clear on your targets?"*
> *"The attackers with the orange markers, right? Roger that!"*
> *"Katana, those attackers are our targets!"*

Those are `UN_e010_ADAN_Attacker_S`, of which phase 1 fields four squadrons.

**The hunting pilot targeted turrets almost exclusively.** Every loss line in the
run logs reads `UN_e007_ADAN_Turret`, with `UN_e010_ADAN_Attacker_S` appearing
only twice across all runs — unsurprising, since `SYLPH_HUNT` was built to make
turrets targets and `SYLPH_KEEPOUT` was tuned for them. So 42 kills were
overwhelmingly the *wrong* 42.

That gives a mundane explanation for the whole run of null results, and it does
not require the frames reading or event-gating to be settled first.

**Prediction:** destroying the phase-1 `UN_e010_ADAN_Attacker_S` squadrons should
advance the phase, which is directly observable as `deployed` jumping from 41
toward the phase-2 roster.

**Test:** a pilot that prioritises `e010` over `e007` — a target-preference knob,
not new decoding — and the existing probe watching `deployed`.

This is 🟡, not ✅: the objective text is certain, but that *this particular
objective* gates the phase advance is inference from the script's structure.

---

# The target-preference knob works, and the blocker is now combat, not RE (2026-08-24)

## ✅ `SYLPH_PREFER` added and measured

`pilot.py` gained a target-preference knob: a unit-name substring whose matches
get their target score multiplied by 0.05 and everything else by 4.0. With
`SYLPH_PREFER=e010`, one clean 320 s run (witness: **0 stalls**):

| killed | count |
|---|---|
| `UN_e007_ADAN_Turret` | 8 |
| `UN_e010_ADAN_Attacker_S` | **2** |

The preference is real — `e010` kills went from about one *across all previous
runs* to two in a single run — but it is **weak**: turrets still outnumber
attackers 4:1 in the kill log. Target commitment (`COMMIT_MAX`, `COMMIT_DROP`)
and simple proximity keep pulling the nose back to turrets, which are far more
numerous.

`deployed` stayed at **41** throughout. No phase advance.

## 🚧 The blocker, quantified

Phase 1's roster, by craft rather than by squadron:

| unit | members | craft (`sum(n)`) |
|---|---|---|
| `UN_e007_ADAN_Turret` | 12 | **108** |
| `UN_e106_ADAN_Destroyer` | 7 | 7 |
| `UN_f001_TCAF_DeltaSaber_T` | 7 | 7 |
| **`UN_e010_ADAN_Attacker_S`** | **4** | **16** |

Clearing the marked attackers means destroying **16 craft**. At the measured rate
of 2 per 320 s that is **~2560 s — roughly 43 minutes** of continuous
verified-live flight, across many chained attaches, with a freeze rate of about
two runs in five.

**This is no longer a reverse-engineering problem.** Everything needed to observe
the phase advance is built and validated: the roster link, the liveness read, the
stall witness, chained attaches, the discard rule. What is missing is a pilot
good enough to actually complete the mission objective, and that is game-playing
work with an uncertain payoff.

## The choice, which is the user's to make

1. **Invest in the pilot** — better target selection, missiles, using the
   marker/lock the script keeps referring to. Uncertain, potentially several
   iterations, and drifts away from RE.
2. **Accept the static answer.** The phase structure, per-phase rosters, route
   paths and objective texts are all decoded and cross-confirmed. Only the
   *trigger* is inferred rather than observed, and it is inferred from the game's
   own guide script naming the objective outright.
3. **One very long run** — 40+ minutes of chained attaches — betting against the
   freeze rate.

Recorded rather than chosen, because it is a question about how much effort this
one confirmation is worth, not a technical unknown.