Yardstick step 11's lists, before any of them joins the yardstick. Numbers from node tools/scratch/sm-reviewers.js; the ledger's from node tools/ledgerstep.js --base 2026-10-03-step6-baseline --lists sm.auspex,sm.auspexFull,sm.tobias,sm.tistaminis.
The lists
| Key | List | Panel file | Graded | Matched to the ledger | Use |
|---|---|---|---|---|---|
| AF | Auspex Tactics, full transcript (4 Aug, before the codex) | 2026-08-04-space-marines-auspex-full.json | 84 | 84 | fitting; replaces the summary |
| BT | Brother Tobias (8 Jul, before the codex; an Ultramarines list) | 2026-07-08-space-marines-tobias.json | 90 | 82 (8 Ultramarines characters and units out) | fitting |
| GB | Grimdark Breakdown, five videos (17 to 19 Sep, leaked codex, before points) | 2026-09-17-space-marines-grimdark.json | 79 | 61 (17 named or chapter characters and units, and the Captain on Bike, out) | reference, partial |
| TR | TacticalRock40K, ranged infantry (26 Sep, codex) | 2026-09-26-space-marines-tacticalrock.json | 12 | 12 | reference, partial |
| TM | Tista Minis (22 Jul), for context | 2026-10-02-space-marines-tistaminis.json | 30 | 30 | fitting, out of the yardstick since 3 Oct |
| AS | Auspex Tactics, the summary (superseded) | 2026-10-03-space-marines-auspex.json | 46 | 46 | superseded by AF |
Letters on one ruler, S A B C D F (F below D). Auspex's seven grades fold to five letters as before (Tier 2 upper and lower both A, and so on). Grimdark's plus, minus and split grades fold to the nearest letter, an exact tie to the lower one (the panel's scale says how; its heard keeps the grade as given).
Every pair
ρ is Spearman on the shared units' letters; pairwise is the share of pairs both lists separate that they order the same way.
| AF | BT | GB | TR | |
|---|---|---|---|---|
| AF | – | 0.73 · 92% · n 82 | 0.25 · 67% · n 61 | 0.19 · 62% · n 12 |
| BT | – | 0.24 · 64% · n 61 | 0.22 · 64% · n 12 | |
| GB | – | 0.69 · 89% · n 12 |
For context: Tista Minis against AF 0.63 · 89% (n 30), BT 0.57 · 82% (30), GB 0.33 · 71% (22), TR −0.22 · 45% (6). The superseded summary against AF, the same video: 0.94 · 99% (n 46).
Each list's noise ceiling (leave one out)
Each list against the mean letter of the other four independent lists (AF, BT, GB, TR, TM; the summary left out, being AF's video), on the units it shares with at least one of them.
| List | ρ | Pairwise | n | On units two or more others grade |
|---|---|---|---|---|
| AF | 0.73 | 90% | 82 | 0.61 · 85% · n 69 |
| BT | 0.63 | 80% | 82 | 0.51 · 74% · n 69 |
| TM | 0.61 | 79% | 30 | the same 30 |
| GB | 0.32 | 67% | 61 | the same 61 |
| TR | 0.28 | 61% | 12 | the same 12 |
The two full pre-codex lists agree well (0.73), and Tista Minis with both. The two post-codex lists agree with each other (0.69, but on 12 units) and poorly with the pre-codex three: the codex rewrote datasheets between them (Grimdark grades whether a datasheet got better or worse), so they are a different question, not just noisier answers.
Where they disagree most
Units two or more letters apart between the independent lists, with the spread of each list's own percentile for the unit (0 to 100: how far apart the lists put it once each list's own harshness is taken out). Tobias grades a letter harder than Auspex across the board (the same letter on only 9 of 82 units, yet ρ 0.73), so a gap of two with a small spread is mostly that offset. The 21 with a spread of 50 or more:
| Unit | Letters apart | Spread | AF | BT | GB | TR | TM |
|---|---|---|---|---|---|---|---|
| Aggressor Squad | 4 | 92 | S | C | D | – | B |
| Terminator Squad | 4 | 87 | A | D | A | S | C |
| Rhino | 4 | 84 | B | D | S | – | – |
| Librarian in Terminator Armour | 3 | 82 | S | B | C | – | – |
| Lieutenant | 3 | 82 | S | C | C | – | – |
| Intercessor Squad | 3 | 73 | S | A | A | C | A |
| Land Raider Crusader | 3 | 71 | B | D | A | – | C |
| Chaplain with Jump Pack | 3 | 65 | A | D | A | – | – |
| Hellblaster Squad | 3 | 60 | A | C | B | D | B |
| Vanguard Veteran Squad with Jump Packs | 3 | 54 | A | D | B | – | A |
| Heavy Intercessor Squad | 2 | 59 | A | B | C | C | – |
| Techmarine | 2 | 59 | A | B | C | – | – |
| Ballistus Dreadnought | 2 | 57 | S | A | B | – | B |
| Captain in Gravis Armour | 2 | 56 | A | S | B | – | – |
| Lieutenant with Combi-weapon | 2 | 56 | A | S | B | – | A |
| Storm Speeder Hammerstrike | 2 | 56 | A | S | B | – | – |
| Storm Speeder Hailstrike | 2 | 56 | B | C | A | – | – |
| Eradicator Squad | 2 | 54 | A | B | B | B | S |
| Outrider Squad | 2 | 52 | A | C | C | – | C |
| Terminator Assault Squad | 2 | 51 | S | B | B | – | – |
| Drop Pod | 2 | 50 | A | C | C | – | – |
Two of Tobias's D grades, the Chaplain with Jump Pack and the Vanguard Veterans, are only because their new models weren't out yet. A further 24 units are two letters apart with a spread under 50, most of them Tobias's harder scale: just under, Infernus Squad (49), Redemptor Dreadnought (46) and Ancient in Terminator Armour (39); then the Apothecary and Biologis, the Chaplains, the Captains in Phobos and Terminator Armour, Invictor, Judiciar, Devastators, Razorback, Tactical Squad, the two aircraft and others.
The ledger against each list (step-6 baseline)
| List | n | Pairwise (pairs) | 2+ tiers apart (pairs) | ρ |
|---|---|---|---|---|
| Auspex summary (the yardstick's list to step 10) | 46 | 68% (778) | 80% (339) | 0.41 |
| Auspex full transcript | 84 | 63% (2,440) | 74% (859) | 0.28 |
| Auspex full, on the summary's 46 units only | 46 | 68% (751) | 81% (321) | 0.40 |
| Brother Tobias | 82 (79 in pairwise: his three F units sit out, as Maelstrom's do) | 55% (2,375) | 57% (1,180) | 0.20 |
| Tista Minis | 30 | 54% (323) | 58% (126) | 0.10 |
| Grimdark Breakdown (reference) | 61 | 52% (1,239) | 48% (354) | 0.05 |
| TacticalRock40K (reference) | 12 | 43% (56) | 34% (29) | −0.20 |
Three readings. The transcript's grades barely differ from the summary's (9 of 46 units a grade apart, 5 a letter): on the same 46 units the ledger scores the same (0.40 against 0.41). The fall to 0.28 is the 38 units the summary never named, so the summary flattered the ledger by grading the units it already places well. Against Tobias the ledger is near chance in pairwise (55%) while the two reviewers agree at 92%. And the ceilings put the ledger well short: 0.28 and 0.20 against leave-one-out ceilings of 0.73 and 0.63.