2016-17 Preferred

• G-001 NO PACK-COMPLETABLE BASE SET, AND EXTRACTION_SPEC 5.3 CANNOT BE APPLIED TO THIS RELEASE. All 298 issued card numbers render the flag token "AU" -- enumerated row by row on Checklist.cfm pages 1-3 and on all thirty Gallery.cfm pages of https://www.tcdb.com/ViewSet.cfm/sid/154373/2016-17-Panini-Preferred -- and cardboardconnection says so in words on https://www.cardboardconnection.com/2016-17-panini-preferred-basketball : "Technically the base set, the 2016-17 Panini Preferred Basketball checklist consists of multiple signed inserts, each with a distinct look and on-card signatures." 5.3 says a block filed past base_set_size is EXCLUDED from the base set; here there is no base run for the blocks to sit past, and excluding all of them yields base_set_size=0, which migrations/002_core_entities.sql rejects (CHECK (base_set_size > 0)). base_set_size=298 is therefore written as the whole enumerated numbered run, which is what both sources treat as the technical base set, and the six constituent blocks are additionally filed as auto rows in sets.csv carrying a dated RECLASSIFIED marker so a consumer can see them without double-counting (the CAV-032 treatment). THE DECISION NEEDED FROM THE OWNER: whether an all-hits release should carry base_set_size = the whole numbered run, or whether the schema should gain a way to say "this release has no base set". Nothing in the catalogue currently answers it: the smallest base_set_size across all 240 release rows is 90 and no existing release has no base portion at all.
• G-002 FOUR OF THE SIX BASE-NUMBERED BLOCKS HAVE A PER-CARD VARIABLE SERIAL RUN AND THE SCHEMA CANNOT HOLD IT. Enumerated over EVERY row, never sampled. Rookie Silhouettes Autograph Relics (cards 1-40): {"SN99"} uniform. Silhouettes Autograph Relics (41-100 less 50, 59 cards): eight distinct tokens {"SN75","SN60","SN50","SN49","SN47","SN45","SN38","SN35"}. Autographs (101-140): four distinct {"SN99","SN75","SN49","SN35"}. Crown Royale Autographs (141-180): five distinct {"SN99","SN75","SN60","SN49","SN35"}. Panini's Choice Autographs (181-200): three distinct {"SN99","SN75","SN35"}. Unparalleled Autographs (201-300 less 203, 99 cards): {"SN50"} uniform. sets.csv has no print_run column (EXTRACTION_SPEC 2.2) so the figures live in that file's notes in words and here. The two uniform blocks are corroborated by cardboardconnection on https://www.cardboardconnection.com/2016-17-panini-preferred-basketball , which states "Rookie Silhouettes Autograph Relics: 40 cards, #/99" and "Unparalleled Autographs: 99 cards, #/50"; the four variable blocks have no single figure to state and none was invented.
• G-003 base_print_run=UNKNOWN AND base_print_run_basis=UNKNOWN ARE NOT MISSING RESEARCH -- THEY ARE THE EXTRACTION_SPEC 2.3 EXPRESSIVENESS GAP AT RELEASE LEVEL, AND base_card_tiers.csv IS ILLEGAL HERE. The 298 cards are provably serial numbered (an SNxx token on every row, two endpoints), so "unnumbered" is refuted; the runs differ by block and by card, so no integer is writeable; and "varies-by-tier" requires base_card_tiers.csv rows, which CANNOT be written because migrations/009_triggers_and_views.sql's deferred check_tier_partition requires card_count = range_end - range_start + 1 for every tier and contiguous tiling. The natural tiers are 1-40 (40 cards), 41-100 (59 cards over a 60-number span -- card 50 unissued), 101-140 (40), 141-180 (40), 181-200 (20) and 201-300 (99 over a 100-number span -- card 203 unissued); two of the six therefore fail the card_count=span test, and splitting them at the hole produces a range gap the same trigger rejects. This is the flawless-bkb-2018-19 shape named in 2016-17_SCOPE_MEMO 5d, and it is now measured on a second release. The block names are carried on every base card as card_tier FREE LABELS instead, which CAV-043 licenses for releases with no tier file; cards.tier_id will load NULL for all 298 rows BY DESIGN and that is not a defect to repair.
• G-004 TWO COLUMN-VOCABULARY GAPS ON base_cards.csv, BOTH RECORDED RATHER THAN FORCED. (1) card_type: migrations/002 allows only ("base","rookie-auto") or NULL. 76 rows carry the tcdb flag RC and are written "rookie-auto". The other 222 are autographed memorabilia or autograph cards of VETERANS and LEGENDS sharing the base numbering -- Kobe Bryant at 200 and 257, Kareem Abdul-Jabbar at 133 and 243, Larry Bird at 41, 170 and 189 -- for which no enum value exists: "base" would be false because CAV-017 defines these as hits seeded at autograph odds rather than pack-completable base cards. card_type is EMPTY on those 222 rows. (2) rookie_flag is "UNKNOWN" on those same 222 rows rather than "no", because THIS RELEASE REFUTES ITS OWN DISCRIMINATIVE TEST FROM THE INSIDE: card 190 is Wade Baldwin IV rendering "AU, SN99" with NO RC on both Checklist.cfm?PageIndex=2 and Gallery.cfm?PageIndex=19, while the same player in the same release renders RC at card 5 and card 285; and card 215 is Rodney McGruder, a 2016-17 rookie, rendering "AU, SN50" with no RC. So an absent RC token here carries no information about rookie status (CAV-042, CAV-003) and writing "no" would have put a confident sourced-looking wrong value on at least two provable rookies.
• G-005 ONE PARALLEL SID SPANS TWO OF THIS RELEASE'S SETS AND THE SCHEMA HAS NO SCOPE FOR IT. tcdb sid 157787 "Silhouettes Prime" states "Total Cards: 98" and renders cards 1-100 less 49 and 50 -- that is the Prime parallel of BOTH "Rookie Silhouettes Autograph Relics" (cards 1-40) AND "Silhouettes Autograph Relics" (cards 41-100). parallels.scope_set_id holds one id, so no scope_type=set value is correct; the row is written scope_type=UNKNOWN with scope empty, which per CAV-027 means the scope was NOT RECORDED and specifically does NOT mean it spans the whole base set. Evidence: https://www.tcdb.com/Checklist.cfm/sid/157787/2016-17-Panini-Preferred-Silhouettes-Prime (80 of 98 rows) plus https://www.tcdb.com/Gallery.cfm/sid/157787/2016-17-Panini-Preferred-Silhouettes-Prime?PageIndex=9 and ?PageIndex=10 (the remaining 18).
• G-006 FOURTEEN OF THE THIRTY-SIX PARALLEL TIERS CARRY A PER-CARD VARIABLE RUN, SO print_run AND serial_numbered ARE BOTH UNKNOWN ON THEM AND THE MODAL TOKEN WAS NOT WRITTEN. The distinct token sets, each enumerated over EVERY row of the tier and never sampled, are recorded here because the CSV cannot carry them: Silhouettes Prime {SN25,SN35,SN20,SN15,SN10,SN8,SN6,SN5,SN3} over 98 rows; Autographs Blue {SN15,SN25} over 40; Autographs Purple {SN25,SN35,SN49} over 40; Crown Royale Blue {SN15,SN25} over 40; Crown Royale Purple {SN25,SN35,SN49} over 40; Panini's Choice Blue {SN15,SN25} over 20; Panini's Choice Purple {SN25,SN49} over 20; 2016 NBA Finals Prime {SN10,SN9,SN8,SN4,SN3,SN2} over 14; Board Members Prime {SN10,SN1} over 16; Rookie Playbooks Prime {SN10,SN8} over 36; Stat Line Memorabilia Prime {SN25,SN23,SN20,SN16,SN15} over 26; Trending Upward Prime {SN25,SN2} over 10; Triple Memorabilia Prime {SN10,SN5} over 6; Vs: One on One Prime {SN10,SN9,SN7,SN3} over 17. FOUR of these turn on a SINGLE ROW and each of those four was deliberately confirmed on a second differently-varied fetch, because a three-row sample would have written a clean sourced-looking flat integer: Rookie Playbooks Prime card 9 Henry Ellenson is SN8 against SN10 on the other 35; Board Members Prime card 6 Andre Drummond is SN1 against SN10 on the other 15; Trending Upward Prime card 8 is SN2 against SN25 on the other 9; and in the insert sets, Stat Line Memorabilia card 10 Marc Gasol is SN125 against SN149 on the other 25. cardboardconnection independently flags the non-uniformity with its own "or less" suffix on https://www.cardboardconnection.com/2016-17-panini-preferred-basketball : "PARALLEL CARDS: Purple #/49 or less, Blue #/25 or #/15, Gold #/10, Platinum 1/1" and "PARALLEL CARDS: Prime #/25 or less, Super Prime 1/1." -- the CAV-036 Select calibration, reproduced on a different line and a different season.
• G-007 ONE CONTESTED CARD COUNT, RESOLVED AGAINST tcdb BY tcdb'S OWN PARALLEL. "Playbook Jumbos": tcdb sid 157794 states "Total Cards: 35" and renders 35 rows spanning 1-36 with card 30 absent -- self-consistent, and still wrong. cardboardconnection states "Playbook Jumbo Relics: 36 cards, #/75" on https://www.cardboardconnection.com/2016-17-panini-preferred-basketball. 36 is written, because tcdb's OWN parallels of the set render card 30: https://www.tcdb.com/Checklist.cfm/sid/157795/2016-17-Panini-Preferred-Playbook-Jumbos-Prime renders 29 rows including 30, and https://www.tcdb.com/Checklist.cfm/sid/157796/2016-17-Panini-Preferred-Playbook-Jumbos-Super-Prime renders 35 rows including 30 (and lacking 29). That is CAV-044 exactly: a page rows short of its own parallel. The two Playbook Jumbos parallel rows therefore each render a number its own base page does not, which is also why their card ranges are recorded in visual_notes rather than inferred.
• G-008 tcdb FILES THE 1-40 PRIME TIER TWICE AND THE TWO RENDERINGS ARE IDENTICAL. sid 157970 "Silhouettes Rookies Prime" states "Total Cards: 40" and renders cards 1-40 with "AU, MEM, SN25" on all forty; sid 157787 "Silhouettes Prime" states 98 and renders those same forty cards with the same players and the same "SN25" as the first 40 of its own range. Both are filed as parallels.csv rows because both are real tcdb sids with distinct names and removing either would be a judgement about which tcdb row is wrong. UNRESOLVED: whether the product has one Prime parallel of the whole Silhouettes family or two Prime parallels one of which is a rookie-only subset. SEPARATELY: sid 157787 also omits card 49 (Tim Hardaway, which the parent renders as "AU, MEM, SN47"), and 100 - 49 - 50 = 98 reconciles its stated total exactly, so the omission is tcdb's own record of the set rather than a dropped row -- but no second source states that card 49 has no Prime version.
• G-009 channel IS UNKNOWN ON ALL 36 PARALLEL ROWS AND channel_basis IS EMPTY ON ALL 36, BY DELIBERATE REFUSAL. No source consulted states a trade channel, a SKU exclusivity or a region for this product. cardboardconnection gives only "Product Configuration: 2 cards per pack, 2 packs per box, 8 boxes per case" and a section heading containing the words "Hobby Box Break" on https://www.cardboardconnection.com/2016-17-panini-preferred-basketball -- a heading about a box-break video is not a channel statement, and per CLAUDE.md promoting it would be from-note at best. EXTRACTION_SPEC 2.3 requires channel_basis to be EMPTY when channel is UNKNOWN (migrations/003_parallels.sql basis_iff_channel_known), so the empty basis is the schema saying there is no channel for a basis to describe, not a loss. channel_confidence is likewise empty rather than carrying an out-of-vocabulary value (CAV-038).
• G-010 THE Inserts.cfm NOTES COLUMN CARRIES ZERO INFORMATION ON THIS PARENT AND THE BOTH-DIRECTIONS SPLIT TEST IS UNRUNNABLE. https://www.tcdb.com/Inserts.cfm/sid/154373/2016-17-Panini-Preferred renders the heading "Insert Sets (47)" and NO "Parallel Sets" section at all -- asked directly, the page has none. All 47 Notes cells are image-completion percentages of the form "5.0% of images added": zero serial statements in any format, zero blanks, so there is nothing for the blank-annotation test to discriminate. This is CAV-036's modal worthless state, now measured on a thirteenth parent, and it means EVERY print run in this release came from a per-sid row-level token sweep -- 47 sid fetches plus second fetches, not one saved. The page did NOT truncate: its 47 rows match its own heading and match the 47 entries of the "Included Sets:" list reached at an out-of-range Checklist.cfm?PageIndex, counted by me both times. Note one summarising fetch of that overview page self-reported "46 included subset collections" over a list of 47; I took the list.
• G-013 NO PRIOR-SEASON OR CROSS-PRODUCT SUBSET IS FILED UNDER THIS PARENT, AND THE EXTRACTION_SPEC 5.5 MIRROR COULD NOT BE RUN. All 47 entries of the parent's "Included Sets:" list and all 47 rows of its "Insert Sets (47)" section name 2016-17 Panini Preferred's own sets; not one carries another season's or another product's prefix, counted by me on both renderings. So nothing had to be filed under EXTRACTION_SPEC 5.5 and nothing was gapped away. THE MIRROR DIRECTION IS UNTESTED FOR A REASON THAT IS ITSELF A FINDING: https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2017 was asked directly whether any entry anywhere on the page contains the word "Preferred" and answered NO, listing 27 Panini "Major Releases" entries none of which is Preferred -- so there is no 2017-18 Preferred parent under which a "2016-17 Panini Preferred Update" block could sit. That does NOT exclude such a block being filed under a DIFFERENT 2017-18 parent, and ViewAll.cfm is known to under-count subset-filed products (2017-18_SCOPE_MEMO 5a), so this is a bounded negative and not a clean one.
• G-014 NO MARKETPLACE CORROBORATION WAS ATTEMPTED, SO THE 22 INTEGER PRINT RUNS REST ON TWO CATALOGUE LINEAGES AND NOT ON A CARD IN HAND. The integers written are 1 on fourteen tiers, 10 on seven and 25 on one. Each comes from the tcdb flag token on EVERY row of the tier, with the rendered row count equal to the page's own stated total so that no row can have been silently dropped; nineteen of the twenty-two are additionally matched by cardboardconnection's ladder line ("Gold #/10", "Platinum 1/1", "Super Prime 1/1") and are written source_confidence=multi, while the four Prime tiers written as a flat 10 are source_confidence=single because cardboardconnection states "Prime #/10 or less" in at least one place and a CEILING is not a print run (EXTRACTION_SPEC 2.3). 2016-17 is the Simmons/Ingram/Hield/Brown rookie year so the raw material for an eBay or goldin check exists; 2016-17_SCOPE_MEMO 9 records that no 2016-17 marketplace corroboration has been attempted by anyone yet, and this build did not change that.
• G-015 FOUR INSERT SETS HAVE INTERIOR NUMBERS THEIR OWN PAGES DO NOT RENDER, AND ONLY TWO OF THE FOUR ARE SETTLED. "Quads" states 11 and renders cards 1-10 and 12 with no card 11; cardboardconnection independently states "Quads Memorabilia: 11 cards", and tcdb's own Quads Super Prime (sid 157793) likewise renders 2-10 and 12 with no 11, so 11 is written and card 11 appears to be unissued -- but no source says so in words. "Playbook Jumbos" is the opposite case and is resolved in G-007. "Vs: One on One" renders all 18 of its letter-coded VS- cards while its Prime renders 17 and its Super Prime 16, and "Rookie Playbooks" renders all 36 while its Super Prime renders 34 -- those shortfalls are PARALLEL-SCOPE, i.e. not every card got every tier, which is ordinary and is recorded in each parallel row's visual_notes as a card range rather than treated as a coverage hole. Card-level checklists for these sets are a separate pass (CHASE_CHECKLIST_SPEC) and insert_cards.csv was NOT written by this build.

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