2017-18 National Treasures

• G-001 tcdb's parent header renders "Total Cards: 155" on https://www.tcdb.com/ViewSet.cfm/sid/170088/2017-18-Panini-National-Treasures while the parent renders 201 CHECKLIST ROWS and a highest card number of 159. The header is RIGHT and the disagreement is a question about the UNIT, not staleness: it counts distinct ISSUED card numbers. Enumerated exactly -- 100 base (1-100), 46 Rookie Patch Autographs (101-149 less the unissued 115, 145 and 147), and 9 Rookie Autographs (150-159 less the unissued 152) = 155; rendered rows are 100 + 46x2 + 9 = 201 because each Rookie Patch Autographs number also renders a "b"-suffixed variation row. Neither figure was used to size anything: the block end was read off the last rendered page (https://www.tcdb.com/Gallery.cfm/sid/170088/2017-18-Panini-National-Treasures?PageIndex=21 renders the single final row, card 159). RESOLVED as a reconciliation; left open only because no source outside tcdb states a parent total, so the 155/201 reconciliation is single-sourced to tcdb.
• G-002 INSTRUCTION CONFLICT, RAISED RATHER THAN FOLLOWED QUIETLY. This build's task prompt states that the 20 prior-season cross-season rows carried on the parent's Inserts.cfm listing "get gaps, not set rows". That contradicts EXTRACTION_SPEC 5.5, whose own appended paragraph reads "Prior-season Update blocks belong IN sets.csv, listed and flagged" and records that a 2023-24 briefing which told an agent to exclude them was REVERSED after the agent flagged it. It also contradicts this line's own adjacent season: data/nattreasures-bkb-2018-19/sets.csv carries 16 such cross-season rows (2017-18 Flawless and 2017-18 National Treasures printing plates) as ordinary set rows. Following the prompt would have made 2017-18 the only season of this line to hide them, and would have been a silent, unfilterable omission. The 20 rows are therefore WRITTEN into sets.csv, each reproducing tcdb's own name verbatim INCLUDING its season prefix so a consumer can filter them mechanically, sourced to https://www.tcdb.com/Inserts.cfm/sid/170088/2017-18-Panini-National-Treasures . The orchestrator should confirm which convention it wants; whichever it is, it should be the same one 2018-19 uses.
• G-003 The 20 prior-season cross-season rows carry an EMPTY card_count. Their existence and their names are sourced from the parent listing https://www.tcdb.com/Inserts.cfm/sid/170088/2017-18-Panini-National-Treasures , which has no card-count column, and their own pages were not opened -- 20 fetches that would buy counts for a different season's product. Empty is "unresolved", not "zero". The sids, for a later pass, are: 425862 2014-15 Eminence Optimum Autographed Patches Printing Plates Black; 425863 2014-15 Eminence Optimum Autographed Patches Printing Plates Cyan; 425864 2014-15 Eminence Optimum Autographed Patches Printing Plates Magenta; 425865 2014-15 Eminence Optimum Autographed Patches Printing Plates Yellow; 425800 2015-16 National Treasures Hometown Heroes Printing Plates Black; 425801 2015-16 National Treasures Hometown Heroes Printing Plates Cyan; 425802 2015-16 National Treasures Hometown Heroes Printing Plates Magenta; 425803 2015-16 National Treasures Hometown Heroes Printing Plates Yellow; 301452 2016-17 Flawless Autographs Printing Plates Black; 301453 2016-17 Flawless Autographs Printing Plates Cyan; 301454 2016-17 Flawless Autographs Printing Plates Magenta; 301455 2016-17 Flawless Autographs Printing Plates Yellow; 222441 2016-17 Flawless Printing Plates Black; 222442 2016-17 Flawless Printing Plates Cyan; 222443 2016-17 Flawless Printing Plates Magenta; 222444 2016-17 Flawless Printing Plates Yellow; 468394 2016-17 Penmanship Printing Plates Black; 468393 2016-17 Penmanship Printing Plates Cyan; 468395 2016-17 Penmanship Printing Plates Magenta; 411956 2016-17 Penmanship Printing Plates Yellow . Note these sit in FIVE separate neighbourhoods far outside the main 174313-174430 block, and that the 2016-17 Penmanship Printing Plates family is SPLIT across two of them -- Black 468394, Cyan 468393, Magenta 468395 but Yellow at 411956. A sibling sid here can never be computed by offset.
• G-004 PHASE 0 FINDING REFUTED, AND IT CHANGES source_confidence ON THIS PRODUCT. The 2017-18 scope memo section 7.6 and reference/caveats.csv CAV-037 both assert that cardboardconnection and checklistcenter are NOT one opinion on 2017-18, and the ONLY evidence given is that "cardboardconnection carries a 46-card variation-layer fact for National Treasures that checklistcenter does not have", since "two sites copying each other cannot do that". checklistcenter DOES carry it. https://www.checklistcenter.com/2017-18-national-treasures/ renders the heading "2017-18 National Treasures - Rookie Patch Auto Horizontal Set" above the line "46 Cards - Serial Numbered #/49", confirmed on a second differently-parameterised fetch asking specifically whether any such line exists. The two sites also phrase the shared figures near-identically ("100 cards. Serial numbered #/99." against "100 Cards - Serial Numbered #/99"). The independence argument therefore fails on its own only piece of evidence, and the two are treated in this build as plausibly ONE lineage. Consequence, applied throughout: a figure supported only by those two sites is source_confidence=single, and "multi" is written only where tcdb's own row-level token sweep is the second, genuinely independent leg. This does not show the two sites are identical -- only that the one fact said to prove otherwise is not a difference between them.
• G-005 channel is UNKNOWN with an EMPTY channel_basis on 91 of the 92 parallel rows. The single exception is Limited Edition, which is FOTL on two independent statements: the tcdb Inserts.cfm Notes cell "1st Off the Line box exclusive" (https://www.tcdb.com/Inserts.cfm/sid/170088/2017-18-Panini-National-Treasures) and cardboardconnection's "Limited Edition #/15 (FOTL)" (https://www.cardboardconnection.com/2017-18-panini-national-treasures-basketball-cards). That Notes cell is the ONLY non-blank cell in the entire 142-row Inserts.cfm listing, and no source consulted states a trade channel, box type or region for any other tier of this product. Per EXTRACTION_SPEC 2.3 a declined channel is UNKNOWN with an EMPTY basis, never channel_basis=unsupported, which the database rejects. Unresolved: whether this is a hobby-only product at all. No box-configuration or hobby statement was found for 2017-18 National Treasures in this build.
• G-006 SOURCES DISAGREE ON THE EXISTENCE OF A TIER, AND THE ROW WAS NOT WRITTEN. cardboardconnection lists the Rookie Patch Autographs parallel ladder as "Bronze #/25, Limited Edition #/15 (FOTL), Gold #/10, Emerald #/5, Platinum 1/1, Logoman 1/1." (https://www.cardboardconnection.com/2017-18-panini-national-treasures-basketball-cards). tcdb positively enumerates the absence of the Platinum member: https://www.tcdb.com/Checklist.cfm/sid/174318/2017-18-Panini-National-Treasures-Platinum renders "Total Cards: 109" -- which is exactly 100 base cards plus the 9 Rookie Autographs -- and its ?PageIndex=2 renders cards 150-159 and NOTHING between 101 and 149, while the Bronze, Gold and Emerald sids each render "Total Cards: 155" and do carry the 101-149 rows. No "Platinum" row scoped to Rookie Patch Autographs is written. It is possible cardboardconnection is describing the Logoman 1/1 twice, or that tcdb lacks a real tier; nothing consulted settles it.
• G-007 cardboardconnection states that the Rookie Patch Autographs HORIZONTAL variation layer has its own parallel: "46 cards. Serial numbered #/49" with "Bronze #/25" (https://www.cardboardconnection.com/2017-18-panini-national-treasures-basketball-cards). No such tier is written. tcdb carries no sid for a Horizontal Bronze anywhere in the 142-row Inserts.cfm listing, and the Bronze sid 174313 renders 155 distinct numbers with NO "b"-suffixed rows at all, so tcdb positively renders its absence from that sid. Note the adjacent season DOES carry such a tier -- data/nattreasures-bkb-2018-19/parallels.csv names "Horizontal Bronze" in an ambiguous_with field -- so this is a plausible tier that this season's tcdb does not catalogue, not an implausible claim. Unresolved.
• G-008 The Limited Edition tier carries 45 of the 46 issued Rookie Patch Autographs cards, not 46. https://www.tcdb.com/Checklist.cfm/sid/174316/2017-18-Panini-National-Treasures-Limited-Edition renders "Total Cards: 45" and its rendered numbers omit card 104 (Josh Jackson) in addition to the 115, 145 and 147 that are unissued product-wide. Single-sourced to tcdb; no editorial source states a roster size for this tier. The row is written at print_run 15 regardless, since the run is what the tier's own 45 rows uniformly carry ("AU, SN15") and cardboardconnection states the same figure.
• G-009 THE SCHEMA CANNOT EXPRESS WHAT WAS MEASURED, AND THIS IS THE LARGEST SINGLE LOSS IN THIS RELEASE. 19 of the 78 set-scope tiers probed carry a PER-CARD VARIABLE print run: they are PROVABLY serial numbered, and the run differs card by card. EXTRACTION_SPEC 2.3 records that the schema cannot hold "serial_numbered=yes with print_run=UNKNOWN", so all of them collapse to UNKNOWN/UNKNOWN, which is indistinguishable from "we found nothing". They are not nothing; every one was ENUMERATED rather than sampled. The distinct token sets, which the CSV cannot carry, are: All-Decade Materials Prime ("SN10" x11, "SN5" x4, "SN4" x1) | All-Decade Memorabilia Signatures Gold ("SN10" x18, "SN9" x1, "SN5" x1) | Century Materials Bronze ("SN25" x50, "SN24" x1, "SN15" x1, "SN10" x8, "SN21" x1, "SN20" x1) | Century Materials Gold ("SN10" x72, "SN5" x10, "SN4" x1, "SN3" x1, "SN2" x2, "SN9" x1) | Colossal Jersey Autographs Bronze ("SN25" x41, "SN10" x5, "SN16" x1, "SN9" x1) | Colossal Materials Prime ("SN4" x1, "SN25" x1, "SN2" x1, "SN10" x6) | Game Gear Prime ("SN25" x31, "SN10" x9, "SN14" x1, "SN22" x1, "SN23" x1) | Game Gear Autographs Prime ("SN10" x9, "SN25" x18) | Game Gear Dual Autographs Prime ("SN10" x8, "SN25" x14) | Game Gear Duals Prime ("SN25" x19, "SN10" x5, "SN6" x1) | Game Gear Triples Prime ("SN25" x12, "SN10" x2, "SN5" x1) | Lasting Legacies Prime ("SN10" x12, "SN25" x13) | Material Treasures Prime ("SN25" x45, "SN10" x4, "SN19" x1, "SN4" x1, "SN5" x1, "SN21" x1) | Retro Materials Bronze ("SN25" x44, "SN8" x1, "SN4" x1, "SN10" x3, "SN5" x1, "SN20" x1, "SN15" x2, "SN21" x1, "SN13" x1, "SN17" x1) | Retro Materials Gold ("SN10" x40, "SN5" x5, "SN3" x1) | Rookie Triple Materials Gold ("SN10" x22, "SN8" x1) | Treasured Threads Prime ("SN25" x39, "SN10" x5, "SN24" x2, "SN23" x1, "SN21" x1, "SN16" x1) | Tremendous Treasures Bronze ("SN25" x23, "SN10" x12, "SN20" x1, "SN19" x1, "SN16" x1, "SN13" x1) | Tremendous Treasures Gold ("SN10" x16, "SN5" x11, "SN7" x2, "SN8" x1, "SN9" x1) . Writing the modal token on any of these would have produced a confident false sourced integer. This is a schema problem, not a research problem.
• G-010 0 set-scope tiers rendered UNIFORM serial tokens over a PARTIAL enumeration and were therefore written as UNKNOWN rather than as the token seen, because a sample of rows is not a tier-level print run (EXTRACTION_SPEC 3.2). Each would very likely resolve with two or three Gallery.cfm top-ups: . Note the contrasting cases that WERE resolved this way and are written as integers: All-Decade Signatures Gold (16 of 20 on Checklist, completed by Gallery ?PageIndex=2), Rookie Jumbo Materials Gold (20 of 23, completed by ?PageIndex=3), Colossal Materials Super Prime (30 of 37, completed by ?PageIndex=4) and Century Materials Platinum (52 of 97, completed by ?PageIndex=6 through 10 -- five top-ups to turn one UNKNOWN into a sourced 1/1).
• G-011 FOUR INTERIOR CARD NUMBERS ARE NOT ISSUED: 115, 145 and 147 inside the Rookie Patch Autographs block 101-149, and 152 inside the Rookie Autographs block 150-159. Evidenced by contiguous enumeration on THREE renderings that agree: https://www.tcdb.com/Checklist.cfm/sid/170088/2017-18-Panini-National-Treasures?PageIndex=2, https://www.tcdb.com/Gallery.cfm/sid/170088/2017-18-Panini-National-Treasures?PageIndex=18 through 21, and https://www.tcdb.com/Gallery.cfm/sid/170088/2017-18-Panini-National-Treasures?PageIndex=13&sort=0 (which renders 114 then 116 with no 115). It also reconciles arithmetically with the header, 155 = 100 + 46 + 9. None of these numbers falls inside base_set_size, so base_cards.csv is a clean contiguous 1-100 and is unaffected. NOT DONE, and named as the one check this build owes: EXTRACTION_SPEC 5.5's MIRROR test -- the 2018-19 National Treasures parent was not checked for a "2017-18 Update" block that might contain these four numbers. On 2022-23 Contenders Optic five independent renderings agreed a card was unissued and it was sitting in the next season's Update block. Errors.cfm was also not exercised on this parent; per CAV-020 its silence would have been worthless, but its ROWS would have been evidence.
• G-012 COUNT DISAGREEMENTS BETWEEN tcdb AND checklistcenter, tcdb written in every case because its counts were confirmed against rendered rows. checklistcenter says Retro Materials 50, tcdb header 43 (40 rows rendered); checklistcenter says Rookie Materials 23, Rookie Jumbo Materials 23 and Rookie Dual Materials 23 where the tcdb BASE sids render headers of 21, 21 and 21 -- note 23 is exactly the roster size of each of those sets' Bronze, Gold and Platinum parallels, so checklistcenter may be quoting a PARALLEL roster rather than the base set (EXTRACTION_SPEC 5.4 unit 4); checklistcenter says Colossal Rookie Materials 38 where tcdb renders 35 for the base and 38 for the Super Prime, the same shape again. checklistcenter also lists "NBA Greats Signatures Bronze Set - 11 Cards" where tcdb renders 20. Unresolved: no third independent catalogue was reachable, Beckett being confirmed gone for this season. Sources: https://www.checklistcenter.com/2017-18-national-treasures/ and the per-sid tcdb pages cited in each sets.csv row.
• G-013 THE DOCUMENTED Gallery.cfm DROPPED-TILE DEFECT REPRODUCED, AND THE DOCUMENTED FIX WORKED. https://www.tcdb.com/Gallery.cfm/sid/174343/2017-18-Panini-National-Treasures-Colossal-Materials-Super-Prime?PageIndex=4 returned 6 tiles; the identical page with "&sort=0" appended returned 7, recovering CM-39 Michael Kidd-Gilchrist. Without the cache-bust this tier would have been recorded as 36 of 37 rows enumerated and written UNKNOWN instead of 1. Recorded because EXTRACTION_SPEC 3.2 describes this defect and it is worth a fourth confirmation.
• G-014 DELIBERATE DIVERGENCE FROM THE ADJACENT SEASON, DECLARED. This build writes rookie_flag=no on all 100 base cards. data/nattreasures-bkb-2018-19/base_cards.csv writes UNKNOWN on all 100 of its base cards. The basis here is a WITHIN-SID POSITIVE DISCRIMINATIVE TEST confirmed on two different endpoints: the renderer prints the "RC" token on every row from 101 to 159 and prints it on none of rows 1-100, on https://www.tcdb.com/Checklist.cfm/sid/170088/2017-18-Panini-National-Treasures?PageIndex=1 and ?PageIndex=2 and again on https://www.tcdb.com/Gallery.cfm/sid/170088/2017-18-Panini-National-Treasures?PageIndex=8, 9, 10, 13, 18, 19, 20 and 21. EXTRACTION_SPEC 2.4 says yes/no only "where the source explicitly shows an RC designation", and this source does show one, as a real per-row column whose absence is a rendered value rather than an unexamined field. Recorded rather than smoothed over, per EXTRACTION_SPEC 6: if the catalogue wants one convention across the line, this row and the 2018-19 rows should move together.
• G-015 THE Inserts.cfm NOTES COLUMN IS UNUSABLE ON THIS PARENT, extending the CAV-036 tally. The column EXISTS and is blank on 141 of 142 rows across both sections (https://www.tcdb.com/Inserts.cfm/sid/170088/2017-18-Panini-National-Treasures). The single non-blank cell is prose -- "1st Off the Line box exclusive" -- and carries no serial content of any kind. So the column is neither the Prizm shape nor the bare-token shape nor the image-percentage shape: it is a column with exactly one distribution note and 141 blanks, and the both-directions split test CANNOT BE RUN on it, because there is no annotated-versus-blank split to test. Treating any blank here as "unnumbered" would have been FALSE on essentially every tier in the release. Every print run in this build therefore came from a per-sid row-level token sweep; the saving this column advertises was unavailable, exactly as CAV-036 now advises budgeting.
• G-016 FIVE SETS ASSERT A UNIFORM PRINT RUN IN sets.csv NOTES OVER A PARTIAL ENUMERATION and should be re-probed before that wording is relied on: Personalized (47 of 79 rows) | Rookie Dual Materials (16 of 21 rows) | Rookie Jumbo Materials (17 of 21 rows) | Rookie Materials (20 of 21 rows) | Rookie Triple Materials (20 of 21 rows) . For Personalized the partial tcdb sweep (47 of 79 rows, all "SN1") is corroborated in words by cardboardconnection "79 cards. Serial numbered 1/1" (https://www.cardboardconnection.com/2017-18-panini-national-treasures-basketball-cards) and checklistcenter "Personalized Autos Set - 79 Cards" (https://www.checklistcenter.com/2017-18-national-treasures/), so that one is the least exposed. The others rest on tcdb alone. Nothing in a structured column depends on any of this -- sets.csv has no print_run column (EXTRACTION_SPEC 2.2) and these figures are in prose only.
• G-017 A PHASE 0 MEASUREMENT DID NOT REPRODUCE, IN THE HARMLESS DIRECTION. The 2017-18 scope memo section 7.2 reports that National Treasures Treasured Tags "renders no serial token on Checklist.cfm and SN1-SN5 on ViewSet.cfm", and uses it as the season's exemplar of a flag column lost in the direction that writes a false value. On this build https://www.tcdb.com/Checklist.cfm/sid/174418/2017-18-Panini-National-Treasures-Treasured-Tags rendered the tokens perfectly well -- "SN1" x6, "SN2" x3, "SN3" x2, "SN4" x1, "SN5" x18 over 31 rows. The memo's CONCLUSION stands and is if anything strengthened: the flag column is lost STOCHASTICALLY per fetch, not deterministically per endpoint, so a negative flag result from one fetch of one endpoint is evidence of nothing and the direction-neutral rule is the right one. But the specific sid-and-endpoint pairing the memo names as reproducible is not reproducible.
• G-018 FETCH-LAYER DEFECTS OBSERVED IN THIS BUILD, recorded because the specs ask for them. (1) SELF-COUNTS WRONG, four times: the https://www.tcdb.com/Checklist.cfm/sid/170088/2017-18-Panini-National-Treasures?PageIndex=2 fetch asserted "Total rows listed: 78" over a list of 72; the Bronze ?PageIndex=2 fetch asserted 59 over a list of 55; two others miscounted by small amounts. Every row count in this build was obtained by counting the enumerated items, never by taking the fetch's total. (2) UNREQUESTED GLOSSES ADDED, twice: asked only for flag tokens, fetches of Gallery ?PageIndex=19 and ?PageIndex=13 both appended the word "horizontal" to the VAR rows, which appears nowhere in the token column. The gloss happens to be TRUE -- cardboardconnection independently calls the layer a "horizontal version" (https://www.cardboardconnection.com/2017-18-panini-national-treasures-basketball-cards) -- which is precisely the dangerous case EXTRACTION_SPEC 3.4 names: a quote being true does not establish that it was quoted. The word "Horizontal" is used as the parallel_name on the authority of the two editorial sources, NOT on the authority of those two fetches. (3) Token censuses that do not sum to their own row counts appeared on several tiers; where the tier was already established as variable this changes no written value, and it is noted in G-009.
• G-019 TWO EDITORIAL CLAIMS TESTED AGAINST tcdb, ONE CONFIRMED AND ONE REFUTED, both as the task brief predicted. CONFIRMED: cardboardconnection describes Hometown Heroes with a per-player variable run, and https://www.tcdb.com/Checklist.cfm/sid/174362/2017-18-Panini-National-Treasures-Hometown-Heroes renders exactly that over all 50 rows -- "SN99" x24, "SN49" x14, "SN35" x12. An editorial source correctly declining to give a flat figure is worth recording. REFUTED IN THE TAIL: the same lineage misses the single outlier row in Century Materials -- https://www.tcdb.com/Checklist.cfm/sid/174330/2017-18-Panini-National-Treasures-Century-Materials renders "SN25" on exactly one row beside "SN49" x39 and "SN99" x34. Correct in shape, incomplete in tail, which is the standing verdict on this lineage for this product.
• G-020 MARKETPLACE CORROBORATION NOT INDEPENDENTLY ESTABLISHED BY THIS BUILD, stated plainly rather than borrowed. The task brief reports a Phase 0 eBay title matching tcdb on card number, print run and set. This build ran its own check and reached only https://cardladder.com/ladder/card/rDnHFCUi2xtTIXWtOW9g , titled "2017 National Treasures Donovan Mitchell Rookie Patch Autograph (RPA) /99 #113 BGS 9.5/Auto 10" -- which agrees with tcdb's #113 Donovan Mitchell "AU, MEM, RC, SN99" on all three axes and carries no tcdb flag string, but is a price-index CATALOGUE page, and CHASE_CHECKLIST_SPEC is explicit that a catalogue page is not a seller listing. No free-text seller title carrying a serial off a card in hand was reached. The #/99 figure therefore rests on tcdb's row-level tokens plus the cardboardconnection/checklistcenter lineage, not on a marketplace leg. Recorded also as a contamination example: the same search returned an eBay listing titled "2017-18 Panini Contenders - Rookie Donovan Mitchell #113 Playoff Ticket /65 (AU, RC)" -- a DIFFERENT product carrying the tcdb flag string verbatim in its title, rejected on both counts, and a reminder that card number 113 collides across 2017-18 products.
• G-021 PARALLEL ROSTER SIZES DIFFER FROM THEIR BASE SETS ACROSS THIS RELEASE, so no parallel's card count may be inherited from its parent set. Measured: Century Materials base 98 against Bronze 91, Gold 96, Platinum 97; Retro Materials base 43 against Bronze 49, Gold 46, Platinum 49; Tremendous Treasures base 50 against Bronze 41, Gold 41, Platinum 50; Colossal Rookie Materials base 35 against Prime 36 and Super Prime 38; Rookie Materials, Rookie Jumbo Materials, Rookie Dual Materials and Rookie Triple Materials each base 21 against colour tiers of 23; Signatures base 70 against Platinum 69; Treasured Threads base 50 against Platinum 49 and Prime 46; All-Decade Signatures base 20 against Bronze 7; All-Decade Memorabilia Signatures base 20 against Bronze 10; Penmanship base 40 against Bronze 35. parallels.csv has no card-count column so none of this is written as structured data; it is recorded because it bears directly on G-012, where checklistcenter's counts for four sets match a PARALLEL roster rather than the base set.
• G-022 THE TASK BRIEF'S VARIABILITY DECOMPOSITION IS HALF CONFIRMED AND HALF REFUTED, MEASURED AT n=116 INSTEAD OF n=16, AND THE REFUTED HALF IS THE PLANNING TOOL. The brief, from the 2017-18 scope memo section 7.5, predicted that this parent's 137 insert rows decompose as about 24 printing-plate + 56 colour-metal + 57 parent/Prime/Tag, and that "parent / Prime / Tag tiers are 11 of 12 variable (92%); colour-metal parallels (Bronze / Gold / Platinum / Emerald) are 0 of 4". THE DECOMPOSITION IS EXACT: 24 printing-plate rows (20 prior-season plus 4 this-season Penmanship), 56 colour-metal parallel rows, and 57 remaining (38 parent sets, 18 Prime / Super Prime / Prime Tag parallels, and the Rookie Patch Autographs Logoman) = 137. THE VARIABILITY RATES DO NOT SURVIVE THE LARGER SAMPLE. Every one of the 116 enumerable tiers was swept for its DISTINCT token set. Colour-metal parallels are 9 of 56 variable, 16%, NOT 0 of 4 -- the variable ones are Century Materials Bronze and Gold, Retro Materials Bronze and Gold, Tremendous Treasures Bronze and Gold, Colossal Jersey Autographs Bronze, All-Decade Memorabilia Signatures Gold and Rookie Triple Materials Gold. Note the shape inside that: all 9 are Bronze or Gold and NONE is Platinum or Emerald; all 30 Platinum tiers and all 3 Emerald tiers are uniform. Prime / Super Prime / Prime Tag parallels are 10 of 18 variable, 56%, not 92% -- and the same split appears there, since every one of the 8 Prime Tag and Super Prime tiers is uniform at #/1 while 10 of 10 plain "Prime" tiers are variable. Printing plates are 0 of 4, uniform at #/1, as predicted. Parent sets are 27 of 38 variable, 71%. OVERALL 46 of 116 tiers are variable, 40%, against the memo's 75% on 16 tiers. THE USABLE RULE, which is sharper than the one the memo offers and was not visible at n=16: variability tracks the SCARCEST END OF A LADDER BEING UNIFORM. A tier whose run is 1 is essentially never variable, because 1 is the floor; the variability lives in the middle of each ladder, at Bronze, Gold and plain Prime. Budget full enumerations for Bronze, Gold and Prime tiers and for parent sets; expect Platinum, Emerald, Super Prime, Prime Tag and printing plates to resolve cheaply and uniformly. The memo's own warning that "75% is NT's number, not the premium tier's and certainly not the season's" should be extended: it was not even NT's number.

If you know a source that answers one of these, tell us — name the question's label and where to look.