2017-18 Vanguard

• G-001 THIS LINE'S NBA FLOOR IS NOT ESTABLISHED AND 2017-18 IS NOT CLAIMED AS IT. specs/2017-18_SCOPE_MEMO.md section 8a admits Vanguard scoped to 2017-18 only and claims no floor. Two sources describe a PRE-PANINI ANCESTOR IN A DIFFERENT SPORT, which is a naming hazard rather than a predecessor of this line: cardboardconnection writes "Formerly a Pacific brand across several sports from the early 2000s" (https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards) and Panini America's own blog writes "Hobby historians will instantly recognize that the base set for 2017-18 Vanguard Basketball borrows extensively from the 2000 Vanguard Football release." (https://blog.paniniamerica.net/the-panini-america-quality-control-gallery-2017-18-vanguard-basketball-50-images/). The Pacific-era Vanguard is a DIFFERENT line that this catalogue does not carry; reference/brands.csv gives manufacturer=Panini for this row and NOTHING from the Pacific era was merged into it. Every source page used in this build was verified to be 2017-18 Panini basketball by its own rendered heading, and no 2024-25 / 2013-14 / 2019-20 material was used. What remains UNRESOLVED: whether a Panini-era NBA Vanguard exists before or after 2017-18. tcdb's 2017 season index carries exactly one Vanguard entry (sid 177840) and no adjacent season was swept - deliberately, per the task scope. Not chased.
• G-002 THE PARENT CHECKLIST HEADER IS WRONG AND ITS OWN PARALLEL PROVES IT. tcdb parent sid 177840 states "Total Cards: 155" and renders 55 rows in the range 101-160, with 139, 142, 150, 151 and 153 absent (counted by me at https://www.tcdb.com/Checklist.cfm/sid/177840/2017-18-Panini-Vanguard?PageIndex=2 - the fetch's own self-reported tally of 60 disagreed with the 55 rows it printed - and confirmed tile for tile at https://www.tcdb.com/Gallery.cfm/sid/177840/2017-18-Panini-Vanguard?PageIndex=14 , ?PageIndex=15 and ?PageIndex=16 ). Its Gold parallel sid 184202 states "Total Cards: 160" and renders ALL SIXTY of 101-160 including those five (139 Jordan Bell, 142 Luke Kennard, 150 TJ Leaf, 151 Josh Hart, 153 Justin Patton) at https://www.tcdb.com/Checklist.cfm/sid/184202/2017-18-Panini-Vanguard-Gold?PageIndex=2 , and the Red sid 184204 does too. 100 + 30 + 30 = 160 also reconciles cardboardconnection's own set counts (https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards). SO THE FIVE NUMBERS ARE NOT UNISSUED - the parent page is five rows short - and 155 is a count of an incomplete page, not the count of a different unit. This is the opposite of the 2017-18 National Treasures finding recorded in specs/2017-18_SCOPE_MEMO.md section 7.2, where a header disagreeing with the row count turned out to be RIGHT because it counted distinct issued numbers. Both readings are available from the same evidence shape and only a parallel page distinguishes them. base_set_size is unaffected: it is 100 either way, and cards 1-100 are contiguous and complete.
• G-003 CARD COUNT CONTESTED, 25 versus 30, AND NOT SILENTLY RESOLVED. 25 is written. Evidence for 25: tcdb's parent renders 25 rows in the 131-160 range (counted by me) and sportscardchecklist lists "2017-18 Panini Vanguard Prime Prospects Signatures (25 Cards)" at https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-vanguard/trading-card-checklists-and-product-information. Evidence for 30: cardboardconnection states "30 cards." at https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards, and the number range 131-160 spans 30 numbers all of which EXIST - tcdb's own Gold tier renders all thirty. THE RECONCILIATION IS PARALLEL SCOPE WIDER THAN BASE, not a coverage hole: sportscardchecklist lists this set's Gold at 30 cards, its Red at 30 and its Purple at 29, i.e. it reproduces the asymmetry itself, and tcdb's three colour sids render exactly those counts. So five cards (139, 142, 150, 151, 153) appear to exist only at parallel tiers. That is a coherent physical claim and it is NOT verified - no marketplace listing was sought for any of the five, and no source states the asymmetry in words. ⚠ AND THE TIEBREAKER IS DISARMED BY G-009: sportscardchecklist matches tcdb's rendered counts on all 40 entries of this product, so it is not an independent second leg for 25. The honest position is one rendered count against one editorial count, with no third party. WHAT WOULD SETTLE IT: a seller's free-text title for any of those five card numbers showing a base-tier Prime Prospects card, or a fourth source stating the set size.
• G-004 CHANNEL IS UNKNOWN ON ALL 31 PARALLEL ROWS, DELIBERATELY, AND THE PHASE 0 'HOBBY-ONLY' CHARACTERISATION IS NOT SUPPORTED BY ANYTHING THIS BUILD COULD REACH. specs/2017-18_SCOPE_MEMO.md section 2c records Vanguard as 'hobby-only'. Four sources were read for an exclusivity statement and none makes one. The two strongest are PRESENCE statements per reference/caveats.csv CAV-010: Panini America's own blog, "On average, each hobby box of 2017-18 Vanguard Basketball (five cards per box, six boxes per case) will deliver four autographs and one base card or parallel." (https://blog.paniniamerica.net/the-panini-america-quality-control-gallery-2017-18-vanguard-basketball-50-images/), and cardboardconnection, "Formerly a Pacific brand across several sports from the early 2000s, each Hobby box of the basketball product averages four autographs." (https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards). cardboardconnection also renders "Product Configuration: 5 cards per pack, 1 pack per box, 6 boxes per case", "Each hobby box is slated to have: 4 Autographs" and "Boxes contain one base card, numbered to 49, or an even more limited parallel." - all of which describe what a hobby box CONTAINS. NONE says the product is available only through hobby, and the absence of a retail configuration is not a statement of exclusivity (EXTRACTION_SPEC 2.1). This is the totallycertified-bkb-2017-18 shape, not the prestige-bkb-2017-18 shape: prestige earned channel=hobby / sourced-exclusive on an explicit "is exclusive to Hobby boxes" sentence and no such sentence exists here. channel_basis and channel_confidence are therefore EMPTY throughout, which migrations/003_parallels.sql (basis_iff_channel_known) requires. One further fact, recorded because it cuts the other way and matches no channel enum value: the Panini blog says "The product will be available at the official online store of Panini America beginning Friday at 10 a.m." (https://blog.paniniamerica.net/the-panini-america-quality-control-gallery-2017-18-vanguard-basketball-50-images/) - a launch-venue statement, not a channel. WHAT WOULD RESOLVE THIS: a Panini sell sheet or press release stating hobby exclusivity, which would earn channel=hobby with sourced-exclusive on all 31 rows.
• G-005 NINE TIERS AND FIVE SETS CARRY A PER-CARD VARIABLE PRINT RUN, WHICH THE SCHEMA CANNOT EXPRESS. Each is PROVABLY serial numbered - every row carries an SN token - but the tokens within the tier disagree, and migrations/003_parallels.sql (run_and_serial_agree) forbids serial_numbered=yes with print_run=UNKNOWN, so both columns are UNKNOWN and the facts live here. Enumerated over every row of every tier, never sampled. PARALLEL TIERS (print_run=UNKNOWN in parallels.csv): Beyond the Arc Scripts Purple sid 184207 {SN25 x25, SN15 x1 = BA-KBR Kobe Bryant}; Cosmic Force Signatures Purple sid 184211 {SN25 x23, SN15 x4 = CF-CBK, CF-KAJ, CF-KML, CF-SON}; High Voltage Signatures Purple sid 184215 {SN25 x29, SN15 x5 = HV-AIV, HV-DLL, HV-DWD, HV-KBR, HV-KDR}; In Focus Autographs Purple sid 184223 {SN25 x30, SN15 x1 = IF-KDR}; Postseason Heroes Autographed Materials Gold sid 184226 {SN10 x17, SN9 x1 = PH-CMX, SN7 x2 = PH-BWL, PH-JKD}; Postseason Heroes Autographed Materials Purple sid 184227 {SN15 x10, SN10 x2 = PH-BWL, PH-JKD}; V-Team Signatures Swatches Gold sid 184230 {SN10 x46, SN7 x1 = VT-DRS, SN6 x1 = VT-RJK, SN3 x1 = VT-EKT}; V-Team Signatures Swatches Purple sid 184231 {SN25 x35, SN10 x1 = VT-RJK}. SETS whose own base tier is variable (recorded in sets.csv notes; sets.csv has no print_run column and EXTRACTION_SPEC 2.2 forbids parking a run in card_count): Beyond the Arc Scripts {SN99 x20, SN49 x5, SN25 x4}; Cosmic Force Signatures {SN99 x13, SN49 x14, SN25 x4}; High Voltage Signatures {SN99 x14, SN49 x16, SN25 x8}; In Focus Autographs {SN99 x20, SN49 x12, SN25 x7}; V-Team Signatures Swatches {SN99 x31, SN49 x9, SN25 x9}. ⚠ TWO EDITORIAL CEILINGS WERE TESTED AND BOTH ARE REAL CEILINGS, SO WRITING THE FIGURE WOULD HAVE BEEN FALSE RATHER THAN MERELY UNSUPPORTED: cardboardconnection's "Gold #/10 or less" (https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards) is wrong on 3 of 20 Postseason Heroes Gold rows and on 3 of 49 V-Team Gold rows. This is the EXTRACTION_SPEC 2.3 expressiveness gap, recurring on a tenth line.
• G-006 FOUR SETS HAVE A UNIFORM SET-LEVEL PRINT RUN THAT sets.csv CANNOT HOLD. EXTRACTION_SPEC 2.2 forbids putting a print run in card_count and there is no print_run column, so each figure is carried in that set's notes in words and here. Rookie Signatures: {SN99} over all 30 rows, and cardboardconnection states "30 cards. Serial numbered #/99." (https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards). Prime Prospects Signatures: {SN99} over all 25 rows tcdb renders. Hot off the Press Autographs: {SN99} over all 30 rows, cardboardconnection "30 cards. Serial numbered #/99." Postseason Heroes Autographed Materials: {SN25} over all 20 rows, cardboardconnection "20 cards. Serial numbered #/25." ⚠ Note the contrast with G-005: both of these flat editorial figures SURVIVED a full per-row census while the two 'or less' figures did not. The census is what distinguishes them and neither outcome was predictable from the phrasing alone.
• G-007 THE SAMPLES TIER'S QUANTITY IS EXPLICITLY UNKNOWN WHILE ITS SERIAL AXIS IS SETTLED. print_run=unnumbered / serial_numbered=no is written on the serial axis only, on a positive discriminative test: the flag column is demonstrably live on these rows - it renders "SAM" on all 100 - and carries no serial token on any, while the same renderer prints SN49 on the parent and SN10 / SN25 / SN1 on this product's three colour sids. tcdb's own set note says the quantity is not known: "These cards were issued early to dealers in a special pack, and do not have a known print run (though it is assumed to be relatively low)." So "unnumbered" must NOT be read as "we know how many there are". Also unresolved: the distribution fact "issued early to dealers in a special pack" matches no value in the channel enum, so per EXTRACTION_SPEC 2.3 it was described rather than forced into a near-miss value. And the tier's classification is 2-of-3: tcdb files it under a "Promo Sets" heading on the parent Inserts.cfm rather than "Parallel Sets", cardboardconnection lists it as a base-set parallel spelled "Sample*" (https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards), and sportscardchecklist does not list it at all (https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-vanguard/trading-card-checklists-and-product-information). Filed as a base-set parallel under tcdb's spelling.
• G-008 CARDBOARDCONNECTION'S PARALLEL LISTS OMIT THREE TIERS THAT tcdb CARRIES, AND THE OMISSION IS NOT A STATEMENT OF ABSENCE (EXTRACTION_SPEC 2.1). cardboardconnection's "PARALLEL CARDS" line for In Focus Autographs reads "Gold #/10, Red 1/1.", for Postseason Heroes Autographed Materials "Gold #/10 or less, Red 1/1." and for V-Team Signatures Swatches "Gold #/10 or less, Red 1/1." - no Purple in any of the three (https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards). tcdb carries In Focus Autographs Purple sid 184223 (31 rows), Postseason Heroes Purple sid 184227 (12 rows) and V-Team Purple sid 184231 (36 rows), each rendering SN tokens on every row. All three are filed on tcdb's evidence with source_confidence=single. sportscardchecklist also lists all three, but see G-009 before counting it as a second leg.
• G-009 MY 'TWO SOURCES' MAY BE ONE ON EVERY COUNT IN THIS RELEASE. sportscardchecklist matches the row counts I counted myself on tcdb for ALL 40 of its 2017-18 Vanguard entries, including every asymmetry - Rookie Signatures Red at 29 (tcdb omits card 104), Prime Prospects Purple at 29 (tcdb omits card 150), Postseason Heroes Purple at 12, V-Team Purple at 36, In Focus Purple at 31. CHASE_CHECKLIST_SPEC names exact agreement with tcdb's RENDERED counts as the shared-upstream signature rather than corroboration, so sportscardchecklist is NOT counted as an independent leg for any count, and source_confidence=multi in parallels.csv rests on cardboardconnection plus tcdb, never on sportscardchecklist plus tcdb. Where cardboardconnection is silent the row is single. https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-vanguard/trading-card-checklists-and-product-information
• G-010 AN EXCHANGE CARD IS PRESENT AND IS RECORDED RATHER THAN DROPPED. PH-KLV Kevin Love renders the flag token EXCH on the base tier ("AU, MEM, SN25, EXCH", sid 184225) and on all three colour tiers - Gold sid 184226 ("AU, MEM, SN10, EXCH"), Purple sid 184227 ("AU, MEM, SN15, EXCH") and Red sid 184228 ("AU, MEM, SN1, EXCH"). It is the only EXCH row anywhere in the 31 sids swept. The token is recorded as the literal string; no gloss of its meaning was accepted from a fetch (EXTRACTION_SPEC 3.2). No source read in this build states a redemption programme by name for this product, so the reason for the exchange is unresolved.
• G-011 rookie_flag IS UNKNOWN ON ALL 100 BASE CARDS, DELIBERATELY, AND THIS IS A JUDGEMENT WORTH REVIEWING. tcdb renders NO RC token on any of cards 1-100 while rendering RC on every row of 101-160 on the same sid - a within-sid positive discriminative test that the base rows carry no RC designation. EXTRACTION_SPEC 2.4 says to write yes/no only where the source explicitly shows an RC designation and UNKNOWN otherwise, so UNKNOWN is the literal reading. It is also the honest one here, because the base set MIXES eras: it contains 2017 draft-class rookies (Donovan Mitchell #5, Markelle Fultz #15, Jayson Tatum #25, Bam Adebayo #27, Lauri Markkanen #35, Lonzo Ball #72, De'Aaron Fox #85, Josh Jackson #95) alongside retired players (Kobe Bryant #9, Shaquille O'Neal #19, Larry Bird #29, Magic Johnson #39, Pete Maravich #49, Tim Duncan #59, Kevin Garnett #69, Allen Iverson #79, Drazen Petrovic #99), so writing no across the column would assert something false about roughly a dozen rows and writing it selectively would be derivation. No source read in this build states which base cards are rookie cards.
• G-012 beckett.com IS STILL UNREACHABLE FOR THIS SEASON, reproducing specs/2017-18_SCOPE_MEMO.md section 7.6 on a further product. A real Beckett page for this product exists in search results at https://www.beckett.com/news/2017-18-panini-vanguard-basketball/ and it 302s to https://maintenance.beckett.com/ . So the usual third structural source is absent here as elsewhere this season, and any count in this release resting on cardboardconnection plus tcdb has no Beckett check behind it. checklistcenter.com was searched for and returned NO 2017-18 Vanguard page in results; per reference/caveats.csv CAV-037 no constructed checklistcenter URL was tried and NO absence claim is made about that site - its slugs are irregular and a 404 on a built path would be evidence of nothing.
• G-013 pipeline/fix_schema.py's FAMILIES LIST HAS NO PATTERN FOR THIS LINE AND NONE WAS ADDED. parallel_family was derived by lifting FAMILIES and family_of() out of fix_schema.py by AST so the repo keeps exactly one definition, and all four parallel names in this release fall through to "other": Gold, Purple, Red and Samples. Per reference/caveats.csv CAV-039 no entry was added to the list - 268 existing rows carry hand-written families that family_of() does not produce, so the list must not be edited per release. Recorded so a taxonomy pass can decide whether a plain colour ladder deserves families of its own; note the catalogue already collapses colour ladders to "other" elsewhere, so this is consistent, not novel.
• G-014 THE Inserts.cfm NOTES COLUMN CARRIES ZERO PRINT-RUN INFORMATION ON THIS PARENT AND THE BOTH-DIRECTIONS SPLIT TEST CANNOT BE RUN AT ALL. Calibrated per reference/caveats.csv CAV-036 before use, by asking for each cell's text character for character at https://www.tcdb.com/Inserts.cfm/sid/177840/2017-18-Panini-Vanguard . The column EXISTS and every one of the 31 insert and parallel rows carries an image-completion percentage ("51.7% of images added", "6.9% of images added", "0.0% of images added"), with zero serial statements in prose form and zero in bare-token form, and only one non-percentage cell in the whole page - the Samples distribution note. This is the USELESS shape CAV-036's 2026-09-24 tally records as the most common state of the column across eleven 2017-18 parents, now measured on a twelfth. There are no blank cells to test either, so the split test is not merely failed but unrunnable. EVERY print run in this release therefore came from a per-sid row-level token sweep, 31 sids in all, and none was taken from this column.
• G-015 NO PRIOR-SEASON OR CROSS-PRODUCT SUBSET IS FILED UNDER THIS PARENT - stated positively because EXTRACTION_SPEC 5.5 makes the opposite finding easy to miss. Every one of the 32 entries on https://www.tcdb.com/Inserts.cfm/sid/177840/2017-18-Panini-Vanguard is a 2017-18 Panini Vanguard set (28 insert entries, 3 parallel entries, 1 promo entry), and all 40 entries sportscardchecklist indexes at https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-vanguard/trading-card-checklists-and-product-information likewise carry the 2017-18 Panini Vanguard prefix. Nothing was gapped instead of being filed, because there was nothing of that shape to file. The MIRROR case is NOT checked and stays open: no later-season Vanguard parent was examined for a 2017-18 Vanguard Update block, because tcdb's 2017 index carries only this one Vanguard entry and no adjacent season was swept.
• G-016 INTELLIGENCE FOR A LATER CHASE PASS, NOT A CLAIM ABOUT THIS BUILD'S FILES. Every one of the seven insert sets renders a MIXTURE of real letter codes and BARE ORDINALS in tcdb's card-number column: Beyond the Arc Scripts has a row numbered "28" among BA- codes; Cosmic Force Signatures has "19" and "30"; Hot off the Press Autographs has "11"; In Focus Autographs has "10", "14" and "24"; Postseason Heroes has "6"; V-Team Signatures Swatches has "10" and "15". Per CHASE_CHECKLIST_SPEC this is the dominant mixed code/ordinal pattern and a single real code settles the scheme for a set, so a chase pass should write the coded rows and withhold the ordinals. Separately, tcdb renders the code HV-JSK on TWO High Voltage Signatures rows - John Stockton (SN25) and Jerry Stackhouse (SN99) - which is a genuine two-player initials collision rather than a smear, and both are written verbatim if that set is ever filled. No insert_cards.csv was written by this build; that is a separate pass.
• G-017 Errors.cfm's SILENCE WAS NOT USED AS EVIDENCE OF ANYTHING. https://www.tcdb.com/Errors.cfm/sid/177840/2017-18-Panini-Vanguard returns "No errors or variations have been added for this set yet." Per reference/caveats.csv CAV-020 that sentence discriminates nothing - the endpoint returns it for sets that provably have UER rows - so NO conclusion is drawn about whether a variation set exists. What the clean base run rests on instead is an independent contiguous enumeration of 1-100 across two endpoints with an agreeing overlap. No b-suffixed or VAR-flagged row was seen anywhere in the 31 sids swept, and because Gallery.cfm strips the b suffix and Checklist.cfm truncates, that is unsupported rather than disproved.
• G-018 CONVENTION CHECK AGAINST THE EXISTING CATALOGUE - NO DIVERGENCE FOUND, recorded because EXTRACTION_SPEC section 6 requires the check and a clean result is worth stating. (1) EXCLUSION RULE: base_set_size=100 excludes the numbered AU block at 101-160. 20 of the 29 2017-18 releases already in data/ record an excluded 5.3 block in their release notes, so this call follows the majority convention of its own season and contradicts none of them. Note totallycertified-bkb-2017-18 reached the OPPOSITE call on a superficially similar shape - it INCLUDED cards 101-150 in base_set_size=150 - and was right to, because those rows carry no AU and no MEM and cardboardconnection counts them inside its base set; here the rows carry AU from 101 and MEM from 131, which is what makes them a block. (2) INTEGER base_print_run: 49 with basis=sourced is the normal premium shape in this catalogue, not a novelty - 67 other release rows carry an integer base_print_run, including dominion-bkb-2017-18 (75/100) and encased-bkb-2017-18 (99/100), which are the same size-and-run shape as this release. (3) set_type: all nine non-base sets are auto because every row of every one carries AU; the two memorabilia sets are signed memorabilia and there is no relic value in the enum. THIS RELEASE HAS NO PLAIN INSERT SET AT ALL, which is unusual and is consistent with cardboardconnection's framing of the product around "hard-signed autographs and hefty stock" (https://www.cardboardconnection.com/2017-18-panini-vanguard-basketball-nba-cards).

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