2017-18 Prestige

• G-001 RELEASE DATE rests on ONE source and three sources give two different days. Written: 2018-02-28, from tcdb's own parent set page, which renders the unqualified line "Release Date: February 28, 2018", reproduced on two differently-slugged fetches (https://www.tcdb.com/ViewSet.cfm/sid/164348/2017-18-Panini-Prestige and https://www.tcdb.com/ViewSet.cfm/sid/164348/Prestige-2017-18-Panini-Basketball-Cards). REFUSED: cardboardconnection's only date is "Estimated Release Date: 02/21/2018", which EXTRACTION_SPEC 3.2 says is not a release date, and which the 2017-18 scope memo lists among the season's three refused dates. ALSO NOT WRITTEN: blowoutbuzz's "Arrives: Feb. 21" - unqualified in form, but published 2017-11-18, three months before the fact, in an article headed First Buzz with early retail details, so it is the same forward-looking solicitation figure that cardboardconnection itself labels an estimate. The two 02/21 sources are therefore read as one forecast lineage and tcdb as the only retrospective record. A reader should know the 02/28 figure has no second source. https://www.cardboardconnection.com/2017-18-panini-prestige-basketball-nba-cards https://blowoutbuzz.wordpress.com/2017/11/18/first-buzz-2017-18-panini-prestige-basketball-cards-early-retail-details/
• G-002 base_print_run=unnumbered carries base_print_run_basis=assumed, not sourced, because NO source states in words that the base cards carry no serial numbering. What supports it is a within-product positive discriminative test: tcdb renders no serial token on any of the 200 base rows across Checklist.cfm pages 1-2 and Gallery.cfm pages 8-10 and 18-20, while rendering SN199, SN10 and SN1 on the three Crystal parallel sids of the same product on the same two endpoints. EXTRACTION_SPEC 2.1 is explicit that this shape is 'assumed' and that talking oneself into 'sourced' is the error. https://www.tcdb.com/Checklist.cfm/sid/164348/2017-18-Panini-Prestige?PageIndex=1
• G-003 THE PARENT RENDERS 201 ROWS OVER 200 ISSUED NUMBERS AND THE 201st IS NOT AN INTEGER. Gallery.cfm page 21 on the parent renders a single extra tile whose card number is the non-integer string NNO and whose title is "150 Points Reward" - a Panini Rewards points-redemption card - confirmed on two differently-parameterised fetches. The parent's "Total Cards: 200" therefore counts issued integer numbers and excludes it, which reconciles the header against the row count exactly rather than making either wrong. The row is EXCLUDED from base_cards.csv because migrations/002_core_entities.sql declares cards.card_number INTEGER NOT NULL and NNO cannot load; it is also NOT filed in sets.csv, because no source calls it a set. UNRESOLVED: whether the catalogue should carry redemption/points cards at all, and if so where. Nothing about it was silently dropped - this row is the record. https://www.tcdb.com/Gallery.cfm/sid/164348/2017-18-Panini-Prestige?PageIndex=21
• G-004 FOUR SET COUNTS DISAGREE ACROSS THREE LINEAGES; what was written and why, per set. (1) Old School Signatures - tcdb 19 (header 19, 19 rows rendered, all 19 codes read and counted), cardboardconnection 20, sportscardchecklist "(18 Cards)". No two of the three agree. WROTE 19. See G-008 for the reason: its own four parallel tiers each render 20 codes and the one code missing from the base tier is OS-THW, and the same base-runs-one-short-of-its-parallels shape is independently confirmed on Prestigious Signatures, so 19 is structural rather than a tcdb omission. sportscardchecklist's 18 is unexplained. (2) Prestigious Signatures - tcdb 48 (header 48, 48 rows rendered) and sportscardchecklist "(48 Cards)" agree; cardboardconnection says 50. WROTE 48, cardboardconnection refuted two-to-one. (3) Bonus Shots Signatures - tcdb 80 (header 80, and Gallery.cfm page 8 is the set's last page and renders 71-80) and cardboardconnection 80 agree; sportscardchecklist says "(70 Cards)" and is the outlier. WROTE 80. (4) Bonus Shots Signatures Crystal - tcdb 74, sportscardchecklist 73; tcdb reconciles internally (numbers 1-77 with 14, 48 and 59 unissued = 74) so 74 is recorded in that row's visual_notes. This is a parallel roster, not a sets.csv count. https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-prestige/trading-card-checklists-and-product-information https://www.cardboardconnection.com/2017-18-panini-prestige-basketball-nba-cards
• G-005 THE Inserts.cfm NOTES COLUMN ON THIS PARENT RENDERS TWO DIFFERENT FORMATS IN ITS TWO SECTIONS, which is CAV-036's shape on a new parent and is recorded so the next reader does not calibrate on the wrong one. Its Parallel Sets section renders BARE TOKENS - SN199, SN10, SN1 - and splits 3 annotated against 3 blank, and that split was confirmed in BOTH directions at row level (the annotated sids render their token on every row; the blank sids render no token on any row), so in THAT section the blank cell is discriminative. Its Insert Sets section on the SAME page is a different thing: 63 rows carrying mostly image-completion percentages ("62.5% of images added"), a handful of blanks, and exactly three cells carrying a real serial statement in PROSE ("Serial numbered to 1", "Serial numbered to 10") - so bare-token AND prose AND percentage formats all appear on one page, and in the Insert Sets section a blank cell carries no information and the both-directions test cannot be run. CONSEQUENCE: every one of the 51 insert-scope print runs in parallels.csv came from a per-sid row-level token sweep, not from this column. https://www.tcdb.com/Inserts.cfm/sid/164348/2017-18-Panini-Prestige
• G-006 EIGHT ROWS HAVE A SOURCED BOX SKU THAT THE channel ENUM CANNOT EXPRESS, so they carry channel=UNKNOWN with an empty channel_basis and the SKU in channel_sku. reference/channel_taxonomy.csv already records that jumbo, hanger, cello and fat-pack "HAS NO ENUM VALUE"; this release is mostly a retail product and three of its six base parallels plus two Micro Etch tiers are defined by exactly those SKUs. Affected: Mist (hanger, and blowoutbuzz states it EXCLUSIVELY - "Also only found here will be Mist parallels"), Rain (fat pack / jumbo - see G-007), Micro Etch Rookies Orange (hanger), Micro Etch Rookies Bright Green (jumbo), at base scope and set scope. Per EXTRACTION_SPEC 2.3 these were NOT forced to the near-miss 'retail'; note the corpus is split on exactly this question - 23 existing rows write channel_sku=hanger with channel=retail while 18 write channel_sku='Hanger/Fat Pack' with channel=UNKNOWN, and nothing on either row tells a reader which convention it is under. THE INFORMATION LOSS IS REAL: a strong sourced exclusivity claim is being recorded as an unknown channel. This is a schema expressiveness gap, not a research gap.
• G-007 RAIN'S SKU IS CONTESTED AND value_contested=yes IS SET ON THE BASE-SET ROW. cardboardconnection prints "Rain (Fat)" while blowoutbuzz puts it in jumbo packs: "Jumbo packs ($4.99) will be the home of bright-green parallels" followed by "These will also be the home of Rain parallels". reference/channel_taxonomy.csv treats fat-pack and jumbo as two separate box-sku values, so they are not synonyms in this repo's own vocabulary even though collectors use them interchangeably. Nothing reachable resolves which Panini actually shipped. channel_sku=fat-pack was written as the cardboardconnection wording, with blowoutbuzz's in visual_notes. https://blowoutbuzz.wordpress.com/2017/11/18/first-buzz-2017-18-panini-prestige-basketball-cards-early-retail-details/
• G-008 A PARALLEL'S ROSTER IS NOT ITS BASE INSERT SET'S ROSTER IN THIS RELEASE, measured on three sets, so no tier's count may be assumed from another. (1) Old School Signatures renders 19 codes while its Crystal, Crystal Gold, Crystal Blue and Gold tiers each render 20; the single code present in all four parallels and absent from the base tier is OS-THW, found by diffing the two rendered code lists rather than inferred. (2) Prestigious Signatures renders 48 while its Blue and Gold tiers each render 49 - the same +1 shape - and sportscardchecklist independently reproduces 48/49/49, which is what makes the shape look structural rather than like a tcdb omission. (3) Bonus Shots Signatures rosters differ four ways: 80 base, 74 Crystal, 78 Crystal Gold, 80 Crystal Blue, 78 Gold. UNRESOLVED: whether a parallel can legitimately carry a card its own base tier does not, or whether tcdb's base-tier pages are one row short on both letter-coded autograph sets. This is the 2017-18 scope memo 7.7 Select finding recurring on a second line: every parallel needs its own count fetch.
• G-009 NO CLAIM IS MADE THAT THIS RELEASE HAS NO VARIATION SET. Errors.cfm on the parent returns the page headed "Errors / Variations" with the sentence "No errors or variations have been added for this set yet." Per CAV-020 and EXTRACTION_SPEC that silence discriminates nothing - the identical sentence is returned for sets that provably carry UER rows - so it is cited here as having produced no rows and nothing more. What the clean base run actually rests on is the independent contiguous enumeration of 1-200 with two cross-endpoint overlaps agreeing row for row. A b-suffixed variation layer would not surface in a base-run enumeration and Gallery.cfm strips the b suffix, so absence of one is unsupported rather than disproved. https://www.tcdb.com/Errors.cfm/sid/164348/2017-18-Panini-Prestige
• G-010 TWO OF THE USUAL SOURCES WERE NOT AVAILABLE AND NEITHER ABSENCE IS TREATED AS EVIDENCE. (1) checklistcenter.com: no 2017-18 Prestige page surfaced through any search result. Per CAV-037 no URL was constructed and no 404 was interpreted, so this is recorded as not reached, NOT as the page not existing. Nothing in this release rests on checklistcenter, and the count corroboration that would normally come from it came from sportscardchecklist instead. (2) beckett.com: its own 2017-18 Prestige news URL returns 302 to https://maintenance.beckett.com/, reproducing the 2017-18 scope memo's finding on this product. CONSEQUENCE for reading source_confidence in this release: the independent legs available were tcdb, cardboardconnection, sportscardchecklist and blowoutbuzz; cardboardconnection and checklistcenter could not be compared for a per-product divergence the way CAV-037 asks, because only one of the two was reachable.
• G-011 NO PACK ODDS ARE RECORDED FOR ANY OF THE 13 SETS because no reachable source states any. cardboardconnection carries box configuration only - "6 cards per pack, 24 packs per box, 16 boxes per case" and "Hobby boxes also average a pair of autographs" - and no odds line; blowoutbuzz gives per-SKU contents ("One autograph per 12-pack jumbo box (12 boxes per case)", "Eight-pack blaster boxes ($19.99 suggested retail price) will have two Micro-Etch RCs and eight inserts or parallels") but no insertion odds; beckett is unreachable (G-010). The one scarcity statement found was used: Playmakers carries scarcity_tier=case-hit off "the Playmakers insert is a case hit". Every other set's scarcity_tier is left EMPTY rather than UNKNOWN, matching the corpus's dominant convention for sets.
• G-012 THIS LINE IS OLDER THAN THIS SEASON AND NO FLOOR IS CLAIMED. The owner decision recorded in 2017-18_SCOPE_MEMO.md 8a admitted Prestige scoped to 2017-18 ONLY, and the predecessor was deliberately not chased by this build. Recorded incidentally and without being sought: a search result surfaced cardboardconnection's own "2016-17 Panini Prestige Basketball Checklist, Boxes, Set Info, Reviews" page at https://www.cardboardconnection.com/2016-17-panini-prestige-basketball - existence only, the page was NOT opened - which corroborates the memo's 2c note that Prestige is in tcdb's 2016-17 index. So a backfill of unknown depth sits below this release and 2017-18 is NOT this line's floor. Nothing was measured about how much further back it runs.
• G-013 PRESTIGE'S WEATHER LADDER HAS NO FAMILY AND NONE WAS MINTED. parallel_family was derived by AST-lifting family_of() out of pipeline/fix_schema.py, per CAV-034, and nothing was hand-written: 27 rows land in the existing family Crystal and 30 land in 'other'. The 30 are Horizon, Mist and Rain - a coherent three-step visual system with a sourced finish description (blowoutbuzz: "Horizon parallels with holographic stock", and tcdb's own set note "Cards say 'HORIZON' on the backs, to the right of the card number.") - plus the plain colour names Gold, Blue, Bright Green, Orange, Pink and Red. EXTRACTION_SPEC 2.3 permits minting a family and CAV-039 warns that 268 hand-written values already exist that family_of() does not reproduce; minting a Horizon/Mist/Rain family here would add to that pile in one release, so it was NOT done and is flagged for a decision instead. SEPARATE TAXONOMY NOTE: family_of() puts all 27 Crystal rows in family 'Crystal', while cardboardconnection says the finish IS Cracked Ice - "Cracked Ice technology, apparently referred to as Crystal in this release" - and 'Cracked Ice' is itself a family in the list. The derivation was left to stand; whether Crystal should collapse into Cracked Ice is a corpus-wide question, not a one-release one.
• G-014 TWO BASE ROWS CARRY A TEAM THAT DOES NOT MATCH THEIR SURROUNDING TEAM BLOCK, written as tcdb renders them rather than corrected. The base run is grouped five cards per team; within the Brooklyn Nets block #83 Sean Kilpatrick renders "Milwaukee Bucks", and within the Phoenix Suns block #122 Eric Bledsoe also renders "Milwaukee Bucks". Bledsoe was traded from Phoenix to Milwaukee in November 2017, so tcdb is plausibly rendering a post-trade team rather than the team printed on the card; the Kilpatrick row has no such explanation. #83's identity is not a fetch artifact - the same player at the same number was reproduced on four separate sids (the parent and the Crystal, Crystal Gold and Crystal Blue parallels) - but its TEAM was read from only the parent's Gallery tile. UNRESOLVED: whether team_or_school here means the team on the card or tcdb's current-team field.
• G-015 ROOKIE FLAGS ARE SOURCED HERE, WHICH IS UNUSUAL FOR A PANINI RELEASE, and the no side rests on a discriminative test rather than on a statement. tcdb renders the RC token on exactly cards 151-200 and on none of cards 1-150, measured across Checklist.cfm page 2 and Gallery.cfm pages 18-20 for the rookies and Checklist.cfm page 1 and Gallery.cfm pages 8-10 for the veterans; so rookie_flag=yes on 50 rows and no on 150 rows, with none UNKNOWN. CAV-003 records that most Panini checklist sources do not print RC designations, so this release is an exception worth noting. The split corroborates cardboardconnection structurally: "200-card base set includes 150 veterans grouped by team ... the final 50 cards are all for rookie players." One consequence a reader should see: Ben Simmons is card #1 inside the veteran block and carries no RC token, which is correct for tcdb's scheme (drafted 2016) but will surprise anyone who remembers he was the 2017-18 Rookie of the Year.
• G-016 "PRESTIGE" NAMES TWO DIFFERENT OBJECTS IN THIS CATALOGUE AND THIS IS THE STANDALONE ONE. CAV-021 records that Panini Chronicles carves its base run into sub-brand blocks one of which is called Prestige, and data/chronicles-bkb-2019-20/release.csv states it in words: its base is "carved into 26 sub-brand blocks, each a real standalone Panini product line (Chronicles, Prestige, Threads, ...)". That Chronicles block is NOT this release. Checked on our own data: data/chronicles-bkb-2017-18/sets.csv carries only 4 rows (Base Set, Chronicles Autographs, Chronicles Signature Swatches, Chronicles Swatches) and files no Prestige-named row, so there is no name collision inside the 2017-18 season as currently built - but a resolver matching the string Prestige across seasons will hit both. No row was renamed to avoid this; recording it is the fix.

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