2017-18 Totally Certified

• G-001 CHANNEL IS UNKNOWN ON ALL 55 PARALLEL ROWS AND BOTH BASE TIERS, DELIBERATELY. The only channel evidence found anywhere is a PRODUCT-level sentence on cardboardconnection: "Each hobby box averages one autograph and one memorabilia card." (https://www.cardboardconnection.com/2017-18-panini-totally-certified-basketball-nba-cards), alongside a configuration line "5 cards per pack, 8 packs per box, 16 boxes per case (2 inners)". Neither speaks to any individual tier and neither asserts exclusivity, so no tier inherited a channel and channel_basis is empty throughout, per EXTRACTION_SPEC 2.3 (channel=UNKNOWN with channel_basis=unsupported is rejected by the database). No source read in this build says the product is hobby-only or names a retail configuration.
• G-002 PER-CARD VARIABLE PRINT RUN, INEXPRESSIBLE IN THE SCHEMA. The 45 rows tcdb renders at https://www.tcdb.com/Checklist.cfm/sid/164176/2017-18-Panini-Totally-Certified-Signed-Sealed-and-Delivered?sort=0 carry SIX distinct serial tokens: SN99 on 30 rows, SN75 on 10 (SSD-DA, SSD-FK, SSD-GM, SSD-GY, SSD-HG, SSD-KD, SSD-MG, SSD-RA, SSD-TL, SSD-ZI), SN56 on 1 (SSD-JJ J.J. Barea), SN50 on 1 (SSD-JK Jason Kidd), SN21 on 1 (SSD-GG Gail Goodrich) and SN15 on 2 (SSD-AM Alonzo Mourning, SSD-JS John Stockton). The set is PROVABLY serial numbered with a per-card variable run; sets.csv has no print_run column and parallels.csv cannot hold serial_numbered=yes with print_run=UNKNOWN, so the fact is recorded only here and in the sets.csv note. Row counts are this build's own count - the fetch's self-reported tally summed to 42 of its own 45 printed rows.
• G-003 CARD COUNT DISAGREEMENT, RESOLVED TO 47 AND RECORDED. tcdb's base page states "Total Cards: 45" and renders 45 rows; all three of its colour tiers state "Total Cards: 47" and render 47; cardboardconnection states "Signed, Sealed and Delivered Autographs: 47 cards" (https://www.cardboardconnection.com/2017-18-panini-totally-certified-basketball-nba-cards). Diffing the Gold roster at https://www.tcdb.com/Checklist.cfm/sid/164177/2017-18-Panini-Totally-Certified-Signed-Sealed-and-Delivered-Gold?sort=0 against the base roster identifies the two cards the base page lacks: SSD-BI (Brandon Ingram) and SSD-KB (Kent Bazemore). card_count=47 written. The base page being short means its variable-run census above covers 45 of 47 cards, not all 47.
• G-004 PER-CARD VARIABLE PRINT RUN. The 37 rows tcdb renders at https://www.tcdb.com/Checklist.cfm/sid/164203/2017-18-Panini-Totally-Certified-Fabric-of-the-Game-Rookies?sort=0 carry MEM on every row with SN249 on 36 and SN205 on exactly one - FR-KKZ, Kyle Kuzma. An unhedged "#/249" would be false on that card. No flat figure written; neither source states a print run for this set in words.
• G-005 CARD COUNT: tcdb's base page states "Total Cards: 37" and renders 37 rows, while all three colour tiers state "Total Cards: 38" and render 38, and cardboardconnection states "Fabric of the Game Rookies: 38 cards" (https://www.cardboardconnection.com/2017-18-panini-totally-certified-basketball-nba-cards). card_count=38 written. The one card the base page lacks was NOT identified - the tier rosters were probed for their token sets, not diffed name by name, so which card is missing is unresolved.
• G-006 PER-CARD VARIABLE PRINT RUN ACROSS FOUR TOKENS. The 50 rows at https://www.tcdb.com/Checklist.cfm/sid/164207/2017-18-Panini-Totally-Certified-Fabric-of-the-Game carry MEM on every row with SN199 as the majority token plus SN99 on eight named rows (Dragan Bender, Andre Drummond, Doug McDermott, Hassan Whiteside, Kemba Walker, Kawhi Leonard, Wade Baldwin IV, DeMar DeRozan), SN46 on Jae Crowder and SN25 on Giannis Antetokounmpo and Russell Westbrook. No flat figure written. The fetch's own per-token tally summed to 45 of the 50 rows it printed, so the exact per-token counts are unresolved even though the variability is established by tokens SEEN.
• G-007 THE COLOUR TIERS OF THIS SET DO NOT SHARE ITS ROSTER. Against a 50-card set, Gold (sid 164208) states and renders 45, Green (164209) 47 and Holo Blue (164210) 50 - three different roster sizes on one insert set, the same shape the 2017-18 memo section 7.7 records for Select's base parallels. parallels.csv has no card_count column so the sizes are carried in visual_notes. Which five cards Gold lacks, and which three Green lacks, is unresolved.
• G-008 Gold (sid 164239) and Green (sid 164240) each state "Total Cards: 8" and render 8 rows against a 10-card set, while Holo Blue (164241) renders 10. Whether the two 8-card tiers are genuinely short-printed rosters or an incomplete tcdb catalogue entry is unresolved; cardboardconnection states the parallels without card counts.
• G-009 AN UNHEDGED EDITORIAL PRINT RUN IS REFUTED. cardboardconnection states "Choice Signatures: 24 cards, serial numbered #/35" (https://www.cardboardconnection.com/2017-18-panini-totally-certified-basketball-nba-cards). tcdb renders SN35 on 23 of the 24 rows and SN99 on CS-KAT, Karl-Anthony Towns (https://www.tcdb.com/Checklist.cfm/sid/164216/2017-18-Panini-Totally-Certified-Choice-Signatures). The run is therefore per-card variable and no flat 35 was written. The two sources agree exactly on the count, 24.
• G-010 PER-CARD VARIABLE PRINT RUN. The 25 rows at https://www.tcdb.com/Checklist.cfm/sid/164224/2017-18-Panini-Totally-Certified-Autographs carry AU on every row with SN75 on 16, SN50 on 4 (George Gervin, Willis Reed, Julius Randle, LaMarcus Aldridge) and SN25 on 5 (Oscar Robertson, Dennis Rodman, Magic Johnson, Karl Malone, Ricky Rubio) - 16+4+5=25, the one variable-tier census in this release whose arithmetic closes exactly. No flat figure written. cardboardconnection states no base print run for this set.
• G-011 A UNIFORM SET-LEVEL PRINT RUN THAT THE SCHEMA CANNOT HOLD. Materials is serial numbered #/199 on all ten cards - tcdb renders MEM and SN199 on all 10 rows (https://www.tcdb.com/Checklist.cfm/sid/164238/2017-18-Panini-Totally-Certified-Materials) and cardboardconnection states "Materials: 10 cards, serial numbered #/199" (https://www.cardboardconnection.com/2017-18-panini-totally-certified-basketball-nba-cards). That is a property of the SET, not of a parallel tier, and sets.csv has no print_run column (EXTRACTION_SPEC 2.2), so the figure lives only in the sets.csv note and here. No parallels.csv row was invented to carry it.
• G-012 TWO SETS HAVE NO UN-COLOURED BASE PAGE ON tcdb, SO THEIR set_name IS THE COMMON PREFIX BOTH SOURCES PRINT. Rookie Duals exists on tcdb only as Camo/Gold/Green/Holo Blue (sids 164185-164188) and cardboardconnection's lowest entry is "Rookie Duals Autographs Camo: 10 cards, serial numbered #/25" (https://www.cardboardconnection.com/2017-18-panini-totally-certified-basketball-nba-cards). Rookie Roll Call exists on tcdb only as Red/Camo/Gold/Green/Holo Blue (164180-164184) while cardboardconnection names the set "Rookie Roll Call Autographs: 40 cards". sets.csv therefore carries "Rookie Duals" and "Rookie Roll Call"; neither string is a page heading on tcdb, and a resolver matching tcdb headings byte for byte will not hit them.
• G-013 A CHECKABLE DIVERGENCE BETWEEN THE TWO SOURCES, IN tcdb'S FAVOUR ON EXISTENCE. tcdb catalogues "2017-18 Panini Totally Certified - Rookie Roll Call Red" at sid 164180, "Total Cards: 40", 40 rows flagged AU with no serial token, reproduced on two endpoints. cardboardconnection lists this set's parallels as "Camo #/25, Gold #/10, Green #/5, Holo Blue 1/1" (https://www.cardboardconnection.com/2017-18-panini-totally-certified-basketball-nba-cards) and names no Red tier at all. Since its 40-card count equals cardboardconnection's unnumbered base, the two may be describing the same physical tier under different names rather than disagreeing about what exists - unresolved. Because checklistcenter has no locatable 2017-18 page for this product, this is the only checkable divergence found between the two sources, so their independence on this product is neither established nor refuted.
• G-014 tcdb's ViewSet.cfm AND Checklist.cfm DISAGREE ABOUT THE FLAG COLUMN ON THIS PRODUCT, IN THE FALSE-VALUE DIRECTION. https://www.tcdb.com/ViewSet.cfm/sid/164175/2017-18-Panini-Totally-Certified-The-Mighty-Holo-Blue renders an EMPTY flag column on all 20 rows, while https://www.tcdb.com/Checklist.cfm/sid/164175/2017-18-Panini-Totally-Certified-The-Mighty-Holo-Blue renders SN1 on all 20 - and cardboardconnection independently says Holo Blue is 1/1. Writing the ViewSet reading would have put unnumbered on a provably 1/1 tier. This is the OPPOSITE direction to the 2017-18 memo section 7.2 National Treasures finding (where ViewSet had the tokens and Checklist had none), which is exactly why the 2026-09-20 direction-neutral rule exists. Consequence for this build: every row-level serial probe was run on Checklist.cfm, and ViewSet.cfm was used only as the SECOND differently-varied fetch confirming an absence.
• G-015 Errors.cfm/sid/160148 returns "No errors or variations have been added for this set yet." (https://www.tcdb.com/Errors.cfm/sid/160148/2017-18-Panini-Totally-Certified). Per reference/caveats.csv CAV-020 that silence discriminates nothing, so NO conclusion about a variation set is drawn from it. What the base run does rest on is an independent contiguous enumeration of 1..150 with no hole and no duplicate. A b-suffixed variation layer would not be surfaced by that enumeration and is therefore unsupported rather than disproved. Gallery.cfm strips a b suffix, so the enumeration's Gallery legs could not have shown one either.
• G-016 THE SOURCE BASE FOR THIS RELEASE IS TWO SOURCES: tcdb and cardboardconnection. checklistcenter has NO locatable 2017-18 page for this product - a site search surfaced only its 2024-25 Totally Certified page, a DIFFERENT season, and per CAV-037 no constructed slug was tried as evidence of absence. beckett.com 302s to https://maintenance.beckett.com/ on its own 2017-18 Totally Certified article URL. gogts.net's checklist page (https://gogts.net/2017-18-panini-totally-certified-nba-basketball-cards-checklist/) carries no card counts, print runs or dates at all, only PDF/Excel downloads. sportscardradio.com is robots-blocked to this fetcher. No clean marketplace listing title for a base card of this product was located, so the /299 and /199 figures are not marketplace-corroborated; in the search attempt an eBay /p/ CATALOGUE page was seen carrying a tcdb flag string verbatim ("2017-18 Panini Contenders - Rookie Ticket Donovan Mitchell #113 /125 (AU, RC)"), the documented contamination fingerprint, on a different product. Every "multi" in parallels.csv therefore means cardboardconnection plus tcdb's own row-level token sweep, which is genuinely independent of it; it never means two aggregators of one lineage.
• G-017 NAMING DIVERGENCES BETWEEN THE TWO SOURCES, tcdb'S FORM WRITTEN IN EVERY CASE. (1) tcdb "2017" vs cardboardconnection "2017 (Graffiti)". (2) tcdb "Signed, Sealed, and Delivered" (Oxford comma, no "Autographs") vs cardboardconnection "Signed, Sealed and Delivered Autographs". (3) tcdb "Rookie Duals" / "Rookie Roll Call" tier pages vs cardboardconnection "Rookie Duals Autographs" / "Rookie Roll Call Autographs". A byte-for-byte matcher against cardboardconnection will miss all three.
• G-018 THIS IS THE FIRST RELEASE OF THIS LINE IN THE CATALOGUE AND NOTHING WAS INHERITED. brands.csv carries totallycertified / "Panini Totally Certified Basketball" (brd_01M3JAKF3G842VGQ3GV34FNB1B), admitted by owner decision on 2026-09-27 per specs/2017-18_SCOPE_MEMO.md section 8a, scoped to 2017-18 only with NO floor claimed. data/ holds no other season of this line, so there was no sibling structure to compare base_set_size, the tiering convention or the parallel ladder against - the EXTRACTION_SPEC section 6 adjacent-season check could not be run. IT MUST NEVER BE MERGED WITH THE certified LINE ("Panini Certified Basketball", brd_01M2B8M6BJKF4BVCZ1A77FRZ20): reference/data_line_coverage.csv states "'Panini Totally Certified' is a DIFFERENT brand ... do not merge the two", and certified is labelled verified-not-released for 2017-18. Consistent with that, tcdb's https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2017 index carries exactly one "Certified" entry - this product - and no plain Certified entry, so a loose search for 2017-18 Certified basketball returns THIS product. No value here was read from or written into any data/certified-bkb-* directory, and no source page used in this build conflated the two names. The depth of the Totally Certified line below 2017-18 was NOT investigated (the 2017-18 round brief records tcdb's year/2010 index carrying a Panini Totally Certified entry at sid 57627), so a backfill of unknown depth sits below this release.
• G-019 PRIOR-SEASON SUBSETS: NONE FOUND UNDER THIS PARENT, and this is a reported negative rather than an assumption. https://www.tcdb.com/Inserts.cfm/sid/160148/2017-18-Panini-Totally-Certified returns its whole listing in one fetch with no truncation - 61 rows under "Insert Sets (61)" and 6 under "Parallel Sets (6)" - and not one row carries a season prefix or names another product. The sid block 164172-164241 is fully bounded: 164171 is "2017 Bowman Prospects 5x7 - Gold 5x7" and 164242 is "2009 Magic the Gathering Zendikar - Rules Tips", both foreign. THREE FOREIGN SIDS ARE INTERLEAVED INSIDE THE BLOCK - 164189 "2008 Magic the Gathering Shards of Alara - Tips & tricks", 164190 "...Shards of Alara - Tokens" and 164211 "2008 Magic the Gathering Shadowmoor - Tokens" - which are exactly the three sids Inserts.cfm skipped, so its listing and the block reconcile with nothing hidden. The EXTRACTION_SPEC 5.5 mirror check (a number filed in the NEXT season's parent) was not run because it was not triggered: the base run has no hole to explain.
• G-020 DERIVED COLUMN parallel_family LANDS ON "other" FOR 5 OF THE 8 DISTINCT TIER NAMES, AND NO PATTERN WAS ADDED. Lifting FAMILIES and family_of out of pipeline/fix_schema.py by AST (so the repo keeps one definition) gives: "Holo Blue" -> Holo (16 rows), "Camo" -> Camo (3 rows), and "Gold" (16), "Green" (16), "Blue" (1), "Base Purple" (1), "Rookies Purple" (1) and "Red" (1) -> other. FAMILIES has no pattern for Gold, Green, Blue, Purple or Red, which are this line's entire visual system. Checked rather than assumed: FAMILIES also has NO "Mirror" pattern, so the Certified-family finish the pilot minted is not produced by family_of at all. Per CAV-039 no entry was added to fix_schema.py. 33 of the 55 rows would change value if patterns for Gold and Green were ever added.
• G-021 scarcity_tier IS EMPTY ON ALL 55 PARALLEL ROWS AND ON ALL 16 sets.csv ROWS: the scarcity pass is not part of this build. Per unknown_semantics.csv, empty means the pass never covered the row, which is the true state; UNKNOWN would wrongly assert that it looked and could not place them. No source read in this build states pack odds for any set, so stated_odds and its two evidence columns are empty throughout.
• G-022 EXCH / redemption rows: NONE seen on any of the 67 sids probed, and this is an observation rather than a finding - the probes asked for flag tokens by example (AU, MEM, SN99, EXCH, RC) and EXCH never appeared. Since the 2017-18 memo records EXCH rows on both Hoops and Select, its absence here is worth re-testing by someone filling insert_cards.csv, who will read every row rather than a token census.
• G-023 SUMMARISING-FETCH ARITHMETIC DISAGREED WITH ITS OWN PRINTED LIST ON THREE PAGES IN THIS BUILD, so every count in these files is this build's own count of enumerated items. (1) Checklist.cfm?PageIndex=2 on the Green base parallel (sid 164236) reported "Row Count: 40 cards displayed" above a list of 50 rows numbered 101-150. (2) The Signed, Sealed, and Delivered base census reported per-token counts summing to 42 against its own 45 printed rows. (3) The Fabric of the Game base census reported counts summing to 45 against its own 50 rows. Separately, the first fetch of the Signed, Sealed, and Delivered base page returned tcdb internal record ids (10507915-10507959) in place of the printed card-number column, the documented wrong-column failure; a differently-parameterised re-fetch asking for the letter code by example returned SSD-AM ... SSD-ZI correctly.

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