A transport is worth what only it can carry (T-710, 5 Oct)

Jordan's idea, from his Votann notes: the Hekaton is the only transport for Votann's slow exosuit infantry, the army's killiest units. A transport is worth its own guns and toughness (already in the ledger) plus the delivery of riders that are slow, valuable and have no other carrier. Built as a post step for the exam (tools/exclusivecarry.js), a game-wide dial exclusiveCarry at 0 (0 = today, checked to the bit on all 16 armies). Track A (step 34b, wCarry 0), snapshot c5e7d85b43f8, BSData 374f505. Discovery only; nothing adopted, no ledger file changed.

The short answer

• Not above its placebo game-wide. Held out by army (choose on 10, score on 5, 3 seeds) it gains +1.33 ± 1.80 on consensus pairs, its placebo +0.54. Every fold picks w 1; the held-out gains are +1.54, −0.57 and +3.03.
• On the exam at 0.5 it gains +0.7 [−0.4, +2.0], with 5 of 15 armies better and 7 worse. The placebo gains +0.9 [−0.1, +2.2] on seed 11 and +0.57 averaged over three seeds. Without Votann it gains +0.1 [−0.7, +0.9].
• It does what Jordan described for Votann. The Hekaton goes 10th → 5th of 22 and the Sagitaur 14th → 4th, against the reviewers' A for both. Votann rises +10.5 points at 0.5. But crediting random Votann vehicles gives +7.6 of that: the reviewers rate Votann's vehicles high whatever they carry.
• Three armies gain beyond their placebos. Drukhari +4.3 (placebo +1.4), Emperor's Children +2.6 (+0.4) and Votann +10.5 (+7.6).
• Five armies and the Tyranids pay. Agents −2.2 (the Imperial Rhino carries the Inquisitors, the only transport that can), Orks −1.5, Death Guard −1.1, Thousand Sons and Adeptus Mechanicus −0.6, Tyranids −0.5. The Tyranid cost comes from the Tyrannocyte, 48th → 23rd: cheap, a one-use delivery, its riders slow and shared with the Hierophant only.
• Space Marines barely move (−0.2). Their riders fit eight or nine transports each, so their exclusivity is about 0.11. The Land Raiders and Repulsors rise four to ten places, and the Rhino one.

How it works

For each transport in each army:

• Riders. Who may ride comes from tools/transports.js (the same reading and catalogue order as tools/carry.js). Each rider's space per model is the capacity ÷ the most of it that fit by keywords, so a Thunderkyn in the Hekaton takes two spaces.
• Rider's Value. compute()'s Value of the rider's own solo form at the run's sliders.
• Slowness: clamp(1 − Move ÷ 12, 0, 1). The choice: a game-wide 12", one turn of a typical transport or bike. M5 gives 0.58, M6 0.5, M8 0.33, and M12 or faster 0. The other option, 1 − Move ÷ the army's median Move, makes most of Votann's and Death Guard's infantry average for their army (both medians are M5), so its slowness would be about 0: the opposite of the idea.
• Exclusivity: 1 ÷ the number of the army's transports that may carry the rider. It is 0 when the rider has its own arrival on every one of its solo forms: Deep Strike, Scouts, Infiltrators and the like. Arrival that a detachment grants, such as the Orbital Assault Force's Deep Strike, doesn't count.
• The load: the best set of riders that fits the capacity, maximising Σ min(V × rider pts ÷ transport pts, V) × slowness × exclusivity. Each unit rides once, at one of its own forms; a load holds up to three squads and at most one CHARACTER a squad. The cap at V means a 70-point boat never gains more per point than its cargo is worth per point.
• The credit: w × that sum, added to the transport's ranking Value. It is already in Value per hundred of the transport's points, so it needs no reference P.
• Placebo: the same credits dealt to random VEHICLE or MONSTER units of the army (3 seeds).

A sensitivity, squads only with no characters in a load, is worse: held out +0.30 ± 0.99 against a placebo of +0.47. Lone characters' Values are high per point, which is why the Agents' Rhino and the Sagitaur (through Buri Aegnirssen) gain most from carrying them.

The table (Track A; pooled over the 15 armies on their 11th-edition lists)

Scored on consensus pairs where an army has two or more 11th-edition lists, else on its one list. The exam's per-list mean is beside.

Held out by army, consensus pairs: +1.33 ± 1.80 (w 1 in all three folds: +1.54, −0.57, +3.03); placebo +0.54.

The exam at 0.5 (tools/exam.js, 2,000 paired draws): +0.7 [−0.4, +2.0], armies resampled [−0.5, +2.6]. At 1 it is +1.3 [−0.4, +3.1], and the placebo (seed 11) at 0.5 +0.9 [−0.1, +2.2]. The Tyranids' three 11e lists lose 0.5 and their consensus pairs don't move (90.7).

The wall units at 0.5 (rank in the army, Value, the credit at 0.5, the load)

Prediction (written first): the Hekaton and the Land Raiders rise; cheap transports barely move. Half right.

• Right: the Hekaton rose (10 → 5), the Land Raiders and Repulsors rose (four to ten places), and the Space Marine Rhino barely moved (77 → 76).
• Wrong: cheap transports whose riders are exclusive and slow moved a lot. The Sagitaur went 14 → 4, the Tyrannocyte 48 → 23, the Agents' Imperial Rhino 27 → 10 and the Drukhari Raider 30 → 18. The Chaos Rhinos rose two to six places: their Values multiplied but stayed near the foot of their armies.

That is the shape Jordan asked for: exclusivity and slowness, not capacity per point. But a cheap transport still earns close to its cargo's own Value per point, because of the cap. The old line's fault, cheap boats rising, comes back wherever the cargo is exclusive.

For the cascade

Take it into the next cascade as an army-specific dial under Jordan's 5 Oct rule, not a game-wide one:

• Strengths: 0, 0.5 and 1.
• The game-wide value: 0. Held out it is no better than crediting random vehicles, and the per-army signs split 5 to 7.
• The armies that want it: Votann, Drukhari and Emperor's Children want 0.5 to 1. Orks, Agents and Death Guard want 0.
• The Tyranids at 0, or with the Tyrannocyte discounted as a one-use delivery (T-674's carryOnce, not built here). Otherwise it costs the headline 0.5.

Votann's gain should be read against its placebo. Any credit to Votann's vehicles wins about +7.6 there, the "big vehicles rated high" quirk seen at the wall. The exclusive-carry shape adds about 3 points beyond it.

Reproduce

node tools/exclusivecarry.js --measure --json reports/tiers/2026-10-05-exclusivecarry.json   # the tables above (about 5 s)
node tools/exclusivecarry.js --measure --chars 0                                              # squads only
node tools/exclusivecarry.js --army leagues-of-votann --w 1                                   # one army's loads
node tools/exam.js tools/ledger-benchmarks/2026-10-05-step34b-baseline.json \
  --post tools/exclusivecarry.js --post-args exclusiveCarry=0.5 \
  --post tools/exclusivecarry.js --post-args exclusiveCarry=1 \
  --post tools/exclusivecarry.js --post-args exclusiveCarry=0.5,placebo=11 --md                # the intervals

BSData is read from --bsdata or tools/carry.js's default locations. The ledger's frozen build/units.json is checked against its lock.