2018-19 National Treasures

• G-001 tcdb's parent header renders "Total Cards: 158" on https://www.tcdb.com/ViewSet.cfm/sid/198282/2018-19-Panini-National-Treasures while the parent renders 159 CHECKLIST ROWS. The two are reconcilable and the header is not stale: the checklist runs 1-160, tcdb's own header Notes line reads "#142 and #144 do not exist" (158 distinct numbers), and card 109 renders TWICE, as 109a and 109b. So on this parent "Total Cards" counts DISTINCT NUMBERS, not rendered rows -- another unit for the list in EXTRACTION_SPEC 5.3, which had recorded the header over-counting and under-counting but not this reconciliation. Neither figure was used to size anything: base_set_size was read off the last rendered page. Unresolved part: no source outside tcdb states a total for the parent at all, so the 158/159 reconciliation is single-sourced to tcdb. NOTE ALSO A MEMO DEFECT: the 2018-19 scope memo section 2 states "NT states 158, renders 160". The first half is right and the second is wrong in both readings -- 158 distinct numbers and 159 rendered rows, neither of them 160 -- and the memo cites it as evidence that this line's header totals are stale, which on this measurement they are not.
• G-002 Spec 5.3 convention divergence, recorded rather than smoothed over. This release applies the EXCLUSION rule: base_set_size=100, with the 101-150 and 151-160 autograph blocks as separate sets.csv rows. data/nattreasures-bkb-2019-20 makes the SAME call (base_set_size=100, Rookie Patch Auto and Rookie Auto as separate auto sets), so this line is internally consistent across the two seasons the catalogue now holds. The catalogue-level divergence named in the spec lies elsewhere and is NOT touched by this build: data/contenders-bkb-2019-20 and data/contenders-bkb-2021-22 both set base_set_size=165, including their Rookie Ticket autographs, against 100 in contenders 2020-21, 2022-23 and 2023-24 and in both contendersoptic seasons. Checked against the repo, not presumed.
• G-003 Uniform set-level print runs are recorded in sets.csv notes IN WORDS and deliberately NOT in card_count (spec 2.2), because sets.csv has no print_run column. The sets affected, with the single serial token tcdb renders on every one of their own rows: Base Set SN99; Rookie Patch Autographs SN99; Rookie Autographs SN99; All-Decade Materials SN99; All-NBA Materials SN99; Biography Materials SN99; Colossal Logoman SN1; Colossal Materials SN99; Colossal Rookie Materials SN99; Logoman Autographs SN1; Personalized SN1; Rookie Dual Materials SN99; Rookie Jumbo Materials SN99; Rookie Logoman SN5; Rookie Materials SN99; Rookie Triple Materials SN99; Treasured Tags SN1; Treasured Tags Autographs SN1; Treasured Threads SN99. Nineteen sets, and the figure is a property of each set with nowhere in the schema to hold it as a queryable value. This is a schema gap, not a research gap. On a case-hit product where essentially every card is serial numbered it is also the single largest category of fact this release could not store.
• G-004 PER-CARD VARIABLE PRINT RUNS -- the state EXTRACTION_SPEC 2.3 says the schema cannot express. On each tier listed below, tcdb renders MORE THAN ONE distinct SN token across that tier's own rows. Each is PROVABLY serial numbered with a run that varies card by card, and the forced print_run/serial_numbered pairing collapses that to UNKNOWN/UNKNOWN, which reads identically to "nothing was found". It is the opposite of that. 24 parallel rows are affected: All-Decade Signatures / Bronze (SN25, SN15); All-NBA Materials / Prime (SN25, SN10); Biography Materials / Prime (SN25, SN24, SN10); Century Materials / Prime (SN25, SN10, SN3); Clutch Factor Signatures / Prime (SN20, SN25, SN10, SN4, SN2); Colossal Material Autographs / Prime (SN25, SN9, SN16, SN10, SN13, SN2, SN20, SN22, SN11, SN24); Colossal Materials / Prime (SN6, SN10, SN25); Game Gear / Prime (SN25, SN10); Game Gear Autographs / Prime (SN25, SN10, SN9); Hometown Heroes / Bronze (SN25, SN15); International Treasures / Bronze (SN25, SN15); Lasting Legacies / Prime (SN25, SN22, SN10, SN24); Material Treasures / Prime (SN25, SN10); NBA Greats Signatures / Bronze (SN15, SN25); Peerless Signatures / Bronze (SN15, SN25); Penmanship / Bronze (SN25, SN15); Retro Materials / Prime (SN25, SN8, SN10, SN4, SN17, SN9); Rookie Triple Materials / Prime (SN25, SN9); Signatures / Bronze (SN15, SN25); Timeline Materials / Prime (SN25, SN10, SN5, SN9); Treasured Signatures / Bronze (SN25, SN15); Treasured Threads / Prime (SN25, SN10, SN19, SN9); Treasures of the Hall / Bronze (SN25, SN15); Tremendous Treasures / Prime (SN25, SN10, SN5). The same shape occurs on 20 PARENT sets, which have no print_run field at all to lose it in: All-Decade Signatures; Century Materials; Clutch Factor Signatures; Colossal Material Autographs; Game Gear; Game Gear Autographs; Hometown Heroes; International Treasures; Lasting Legacies; Material Treasures; NBA Finals Nameplate; NBA Greats Signatures; Peerless Signatures; Penmanship; Retro Materials; Signatures; Timeline Materials; Treasured Signatures; Treasures of the Hall; Tremendous Treasures. Where tcdb states the scheme in words it is quoted on the set row -- e.g. All-Decade Signatures "Autographed and serial numbered to 25, 49 or 99." and its Bronze tier "Autographed and serial numbered to 15 or 25." on https://www.tcdb.com/Inserts.cfm/sid/198282/2018-19-Panini-National-Treasures . A ceiling is not a print run and neither is a per-card variable run: nothing was rounded to the commonest token, and no maximum was written as if it were the run.
• G-005 Channel is UNKNOWN with an EMPTY channel_basis on all 91 parallel rows, and the reasoning sits here rather than being promoted into a value. The only channel evidence found anywhere is PRODUCT-level: steelcitycollectibles titles the box "2018-19 Panini National Treasures Basketball Hobby Box" https://www.steelcitycollectibles.com/i/2018-19-panini-national-treasures-basketball-hobby-box and cardboardconnection gives the configuration "10 cards per pack, 1 pack per box, 4 boxes per case" https://www.cardboardconnection.com/2018-19-panini-national-treasures-basketball-cards . Neither speaks to any individual tier, and channel_taxonomy.csv records that a single named channel value asserts EXCLUSIVITY to that channel -- which here would be an inference from absence, since no source states that no other configuration exists. Under CLAUDE.md prohibition 7 that makes the channel UNKNOWN and the basis empty. CONTRADICTION WITH EXISTING DATA: data/nattreasures-bkb-2019-20 writes channel=hobby on its four base-set parallels with channel_basis "unsupported" and "from-note" -- bases that by definition mean no real citation exists -- so the two seasons of this line now answer "what channel is Bronze?" differently on the same evidence shape. This build follows the spec; the divergence is recorded, not resolved unilaterally. data/prizm-bkb-2018-19 makes the same conservative call this build does on its Silver Prizm row.
• G-006 VARIATION ROW, correctly OUTSIDE the base set. https://www.tcdb.com/Checklist.cfm/sid/198282/2018-19-Panini-National-Treasures?PageIndex=2 renders card 109 twice: "109a Michael Porter Jr., Denver Nuggets, AU, MEM, RC, SN99" and "109b Michael Porter Jr., Denver Nuggets, VaR, AU, MEM, RC, SN49". https://www.tcdb.com/Errors.cfm/sid/198282/2018-19-Panini-National-Treasures independently returns exactly one row, "109b Michael Porter Jr. VAR: Horizontal". Both 109a and 109b sit in the Rookie Patch Autographs block (101-150), NOT in the base run, so base_cards.csv is unaffected and every base card_number parses as an integer. Recorded because the letter-suffixed number is the spec 2.4 trap and because it is why rendered rows exceed distinct numbers by one. The literal tokens are recorded as rendered -- "VaR" on the checklist row, "VAR: Horizontal" on the errors page -- and no gloss is written for either (spec 3.4). Note tcdb renders 109b at SN49 while 109a is SN99; the 48-card block is otherwise uniform at #/99 and the Rookie Patch Autographs Horizontal parallel sid 203221 is also #/49, so whether 109b is a member of that parallel or a distinct variation is NOT resolved here.
• G-007 CARDS 142 AND 144 ARE NOT ISSUED, and the absence is evidenced rather than assumed. tcdb states it in the parent header Notes on https://www.tcdb.com/ViewSet.cfm/sid/198282/2018-19-Panini-National-Treasures ("#142 and #144 do not exist") and repeats it in the per-set Notes of all four Rookie Patch Autographs parallels on https://www.tcdb.com/Inserts.cfm/sid/198282/2018-19-Panini-National-Treasures ; all four of those sids independently render "Total Cards: 48" over the range 101-150; and https://www.tcdb.com/Checklist.cfm/sid/198282/2018-19-Panini-National-Treasures?PageIndex=2 renders no row at either number. The spec 5.5 MIRROR CHECK was run before concluding this: https://www.tcdb.com/Inserts.cfm/sid/225917/2019-20-Panini-National-Treasures was searched for a 2018-19 Update block filed inside the NEXT season's parent and there is none -- the only 2018-19 rows there are printing plate blocks, and no row anywhere on that page contains the word "Update". Both numbers lie inside the Rookie Patch Autographs block, so base_set_size is unaffected either way.
• G-008 THE Inserts.cfm NOTES COLUMN ON THIS PARENT IS A MIXED SHAPE AND ITS BLANKS CARRY ZERO INFORMATION -- another measured shape for the list in EXTRACTION_SPEC 3.2, and one that differs BETWEEN SECTIONS OF THE SAME PAGE. On https://www.tcdb.com/Inserts.cfm/sid/198282/2018-19-Panini-National-Treasures the "Parallel Sets (4)" section carries a real serial statement on 4 of 4 rows, so there is no blank cell and no split to test. The "Insert Sets (133)" section on the SAME page carries a real serial statement on only 13 rows (the All-Decade Signatures family, the Treasured Signatures family, and the four Rookie Patch Autographs parallels), image-completion percentages on others -- a second fetch with the URL varied rendered "2017-18 Panini Flawless Distinguished Autos Printing Plates Cyan (2.5% of images added)" -- and nothing at all on the rest. The blank-annotation test was therefore NOT RUN, and must not be reported as run and passed. Every print run written in this release comes from a within-sid row-level probe of that tier's own Checklist.cfm page, never from the blankness of a cell. This matters because a blank cell here looks exactly like the Prizm shape in which blank does mean unnumbered, and on this parent that reading would have been FALSE on every tier: all 91 parallel rows sit on sets whose own pages render an SN token.
• G-009 SID LAYOUT: THIS PARENT IS CONTIGUOUS, AND IT IS THE EXCEPTION. The brief asked this build to test the 2018-19 scope memo's last surviving structural claim, already disproved on Prizm, Hoops and Certified. Measured here: all 137 rows of https://www.tcdb.com/Inserts.cfm/sid/198282/2018-19-Panini-National-Treasures resolve to sids 203134-203254 (120 sids: 4 parallel sets and 116 insert sets) plus 17 cross-season printing plate blocks at 269498-269501, 278803, 280528-280530, 344783-344787, 426097-426100 and 515273. Within 203134-203254 exactly ONE slot is unaccounted for, 203225, and it DOES NOT EXIST: https://www.tcdb.com/ViewSet.cfm/sid/203225/probe renders the tcdb HOME PAGE ("Welcome!", "Cards Listed: 35,360,341"), the documented non-existent-sid signature, sitting alphabetically between Rookie Patch Autographs Limited Edition Blue and Rookie Triple Materials. NO FOREIGN SET WAS FOUND ANYWHERE INSIDE THE BLOCK. Both boundaries were proved foreign rather than assumed: below, https://www.tcdb.com/ViewSet.cfm/sid/203132/probe renders "2019 Panini Contenders Draft Picks Collegiate - Season Ticket Signatures" (Total Cards 34) and https://www.tcdb.com/ViewSet.cfm/sid/203130/probe renders "2019 Panini Contenders Draft Picks Collegiate - Season Ticket Printing Plate Magenta Signatures" (35), both FOOTBALL, with https://www.tcdb.com/ViewSet.cfm/sid/203133/probe a second non-existent slot rendering the home page; above, https://www.tcdb.com/ViewSet.cfm/sid/203255/probe renders "2019 Leaf Metal Draft - Black" (87) and https://www.tcdb.com/ViewSet.cfm/sid/203256/probe renders "2019 Leaf Metal Draft - Black Wave" (87), both FOOTBALL. So the memo's easy-regime claim is TRUE on National Treasures and false on the other three parents tested this season -- it survives on one of four, which makes NT the exception rather than the rule. TWO CAVEATS against reading this as a rehabilitation of the memo: the PARENT sid 198282 sits nowhere near its own children, and the 17 cross-season blocks are scattered across five unrelated sid neighbourhoods, so even this parent is NOT enumerable as a single range. Interleaving remains the correct default and both boundaries should still be proved every time.
• G-010 SET NAME DRIFT ACROSS SOURCES for the two hidden blocks, and against this line's own sibling season. Written here: "Rookie Patch Autographs" and "Rookie Autographs". cardboardconnection uses exactly those two strings https://www.cardboardconnection.com/2018-19-panini-national-treasures-basketball-cards ; tcdb has no sid for either block but names its four parallels "Rookie Patch Autographs Horizontal / Horizontal Bronze / Limited Edition / Limited Edition Blue" https://www.tcdb.com/Inserts.cfm/sid/198282/2018-19-Panini-National-Treasures , which implies the parent form; checklistcenter uses "Rookie Patch Auto Set" and "Rookie Auto Set" https://www.checklistcenter.com/2018-19-panini-national-treasures-basketball-card-checklist/ ; breakninja uses "Rookie Patch Auto", "Rookie Patch Auto Horizontal" and "Rookie Auto" https://www.breakninja.com/basketball/18-19-Panini-National-Treasures-Basketball-Checklist.php ; and data/nattreasures-bkb-2019-20 files them as "Rookie Patch Auto", "Rookie Patch Auto Horizontal" and "Rookie Auto". A resolver matching across the two seasons of this line will NOT match these strings. Nothing was renamed to tidy this away. The same drift affects several insert sets, where tcdb gives the shorter form and cardboardconnection appends "Autographs" or substitutes "Relics": tcdb "Hometown Heroes" / cbc "Hometown Heroes Autographs"; tcdb "International Treasures" / cbc "International Treasures Autographs"; tcdb "Lasting Legacies" / cbc "Lasting Legacies Autographs"; tcdb "Penmanship" / cbc "Penmanship Autographs"; tcdb "Personalized" / cbc "Personalized Autographs"; tcdb "Treasures of the Hall" / cbc "Treasures of the Hall Autographs"; tcdb "Game Gear" / cbc "Game Gear Relics". tcdb's form is written throughout, because it is the form that carries a sid.
• G-011 CROSS-SEASON BLOCKS, BOTH DIRECTIONS, AND A COVERAGE HOLE IN EXISTING DATA. This parent carries 17 PRIOR-season printing plate blocks (2017-18 Flawless Distinguished Autos, 2017-18 National Treasures Colossal Logoman, 2017-18 National Treasures Material Treasures, 2017-18 National Treasures Personalized, 2017-2018 Flawless Patches) on https://www.tcdb.com/Inserts.cfm/sid/198282/2018-19-Panini-National-Treasures . They are listed in sets.csv and flagged per spec 5.5, not dropped. The MIRROR is also present and points at existing data: https://www.tcdb.com/Inserts.cfm/sid/225917/2019-20-Panini-National-Treasures carries SIXTEEN 2018-19 blocks filed inside the 2019-20 parent -- 2018-19 Flawless Printing Plates Black/Cyan/Magenta/Yellow (sids 274150-274153), 2018-19 National Treasure Lasting Legacies Printing Plate Black/Cyan/Magenta/Yellow (259185-259188), 2018-19 National Treasure Peerless Signatures Printing Plate Black/Cyan/Magenta/Yellow (296301-296304) and 2018-19 Panini National Treasures Colossal Logoman Printing Plates Black/Cyan/Magenta/Yellow (470702-470705). NONE of those 16 appears in data/nattreasures-bkb-2019-20/sets.csv, which carries 43 rows and no printing plate row at all. Four of them name 2018-19 National Treasures sets THIS release builds (Lasting Legacies, Peerless Signatures, Colossal Logoman). They are deliberately NOT written into this release: they are filed under a different parent sid and a different season, and writing them here would be this build guessing where another release's rows belong. Flagged for a repair pass on the 2019-20 release.
• G-012 tcdb RENDERS THE LITERAL STRING "TBD" AS A CARD NUMBER on some insert rows of this parent -- observed as the last rendered row on Colossal Material Autographs https://www.tcdb.com/Checklist.cfm/sid/203159/2018-19-Panini-National-Treasures-Colossal-Material-Autographs ("TBD, Nikola Mirotic, AU, MEM, SN99"), on its Prime and Super Prime tiers, and on Hometown Heroes https://www.tcdb.com/Checklist.cfm/sid/203174/2018-19-Panini-National-Treasures-Hometown-Heroes ("TBD, Terry Rozier") and all four of its colour tiers. Recorded here because the spec 6 placeholder check scans for that exact token: NO such value is written into any structured column of this release, and the token is quoted in this description as evidence rather than standing as a placeholder. It also means tcdb holds rows in those sets whose card number is not yet catalogued, so their "Total Cards" figures count a row whose identifier is unknown. A chase-checklist pass on those sets will hit this.
• G-013 BECKETT REMAINS UNREACHABLE, and this was tested rather than inherited from the brief. https://www.beckett.com/news/2018-19-panini-national-treasures-basketball-cards/ returns HTTP 302 to https://maintenance.beckett.com/ . Beckett is the usual third STRUCTURAL source, and its absence is why the two aggregator legs of this build are recorded as possibly one opinion (G-014). The release date itself does not depend on it: four independent sources state 2019-04-26 and none labels its date an estimate.
• G-014 POSSIBLE SINGLE OPINION COUNTED TWICE (spec 5.4). cardboardconnection https://www.cardboardconnection.com/2018-19-panini-national-treasures-basketball-cards and checklistcenter https://www.checklistcenter.com/2018-19-panini-national-treasures-basketball-card-checklist/ agree almost verbatim on the base set ("100 cards. Serial numbered #/99." against "100 Cards - Serial Numbered #/99"), on all four base parallels and their runs, and on both autograph block sizes (48 and 10). These sites are known to copy each other and may be one opinion. The independent leg for every one of those values is tcdb at ROW level inside each tier's own sid -- a different kind of observation, not a third restatement -- and for the Rookie Patch Autographs run also a marketplace lot title: lelands "2018-19 National Treasures #127 Luka Doncic RPA Rookie Patch Autograph 63/99 BGS MINT 9 - Auto 10" https://lelands.com/bids/2018-19-national-treasures-127-luka-doncic-rpa-rookie-patch-autograph-63-99-bgs-mint-9-auto-10 , which independently confirms that card 127 is Luka Doncic and that the block is numbered to 99. Source calibration (spec 4.6) was run before relying on either aggregator: cardboardconnection was tested on three questions tcdb answers independently -- Rookie Patch Autographs 48/#99, Rookie Patch Autographs Horizontal 48/#49, Rookie Autographs 10 -- and passed all three.
• G-015 Errors.cfm RETURNED A POSITIVE RESULT ON THIS SET, recorded because the spec was corrected today to say its silence proves nothing. https://www.tcdb.com/Errors.cfm/sid/198282/2018-19-Panini-National-Treasures does NOT return "No errors or variations have been added for this set yet." -- it returns one row, "109b Michael Porter Jr. VAR: Horizontal". So on this parent the endpoint fired, and its row independently corroborates the 109a/109b split found by enumerating the parent checklist. That is consistent with the corrected rule rather than against it: the endpoint's ROWS are evidence and only its silence is uninformative. It was NOT relied on as the detection path -- the clean base run rests on a contiguous enumeration of all 100 rows, and the variation was found on the checklist first. Note the endpoint does NOT report the 142/144 non-issues, so a build relying on it alone would still have missed those.
• G-016 SET_TYPE OF THE 17 CROSS-SEASON PRINTING PLATE BLOCKS is written as insert and is not fully settled. Four of the five families are plates of AUTOGRAPH sets by name ("Distinguished Autos", "Personalized"), which would make them auto if the plates themselves are signed. The evidence written from is a row-level probe: https://www.tcdb.com/Checklist.cfm/sid/278803/2018-19-Panini-National-Treasures-2017-18-Panini-Flawless-Distinguished-Autos-Printing-Plates-Black?sort=0 renders neither AU nor MEM on any of its first five rows, only "SN1". A printing plate is normally an unsigned 1/1 production plate, which is consistent. But tcdb's flag column has been measured dropping tokens, and no non-tcdb source names these blocks at all, so this is a single-source structural call on 17 rows. There is no relic or plate value in the set_type enum and there is not going to be one (spec 2.2); "Printing Plate" exists as a parallel_family in the catalogue, but these are filed as sets rather than parallels because their parent sets are not part of this release.
• G-017 TIER ROSTERS ARE NOT UNIFORM WITHIN A FAMILY, so no card_count in this release was ever computed as distinct x tiers (spec 5.4). Measured examples where a tier's own "Total Cards" differs from its parent set's: All-Decade Materials 30 with Prime 29 and Super Prime 29; All-NBA Materials 30 with Prime 29; Biography Materials 40 with Prime 39; Clutch Factor Signatures 50 with Prime 43 and Super Prime 46; Colossal Material Autographs 46 with Prime 41 and Super Prime 45; Colossal Materials 30 with Prime 29; Game Gear Autographs 45 with Prime 44 and Super Prime 46 -- where the Super Prime tier is LARGER than its parent set; Peerless Signatures 48 with Platinum 47; Timeline Materials 30 with Prime 29. Every card_count written in sets.csv is the figure rendered on THAT set's own page, never derived. The tier counts themselves are recorded in each parallel row's visual_notes, because parallels.csv has no card_count column to hold them.
• G-018 ONE ROW OF Treasured Signatures Platinum RENDERS AN EXTRA TOKEN. https://www.tcdb.com/Checklist.cfm/sid/203241/2018-19-Panini-National-Treasures-Treasured-Signatures-Platinum renders "SN1" on its rows and additionally "SN1, EXCH" on the Kelly Oubre Jr. row. The literal token "EXCH" is recorded as rendered and no gloss is written for it (spec 3.4 records that the fetch layer invents expansions for flag tokens). It does not change the tier's print run, which is uniform at 1/1. Whether an exchange card affects that tier's effective count of 23 is not resolved.
• G-019 FETCH-LAYER BEHAVIOUR MEASURED ON THIS PARENT, for the spec's running record. Checklist.cfm?PageIndex=1 truncated at row 73 of 100 (last row rendered: 73 Donovan Mitchell), inside the documented 70-86 band and near its floor; rows 74-100 were recovered from Gallery.cfm?PageIndex=8, 9 and 10, which returned 10 tiles each with no dropped tile. Checklist.cfm?PageIndex=2 delivered the whole 101-160 range without truncation, and ?PageIndex=3 rendered no rows. By contrast Inserts.cfm did NOT truncate at all on this parent: one fetch returned all 133 insert rows and all 4 parallel rows with their sids, slugs and Notes cells, which confirms the spec's note that truncation on that page is parent-dependent rather than size-dependent. The spec 4.1 RENUMBERING CONTROL was run on both transcribed pages by asking for the number beside a named player: Stephen Curry returned 57 and Kevin Durant 67 on page 1; Trae Young 103, Luka Doncic 127 and Shake Milton 157 on page 2. All five agree with the enumerated lists, so no constant shift was present on either page. Card 127 Luka Doncic is additionally confirmed outside tcdb by the lelands lot title quoted in G-014.
• G-020 SINGLE-SOURCE CARD COUNTS. Of the 56 sets.csv rows, 53 carry a card_count sourced ONLY from the "Total Cards:" line on that set's own tcdb page: the 36 native insert and autograph parent sets and the 17 cross-season printing plate blocks. cardboardconnection independently states a count for many of the native families and agrees with tcdb on every one it names (e.g. Clutch Factor Signatures 50, Colossal Logoman 98, Peerless Signatures 48, Penmanship 40, Personalized 90, Logoman Autographs 35, Treasured Tags Autographs 9, All-Decade Signatures 29), but it does not name NBA Finals Nameplate, Treasured Tags, Rookie Logoman or any cross-season block at all, and breakninja https://www.breakninja.com/basketball/18-19-Panini-National-Treasures-Basketball-Checklist.php lists set names with no counts. Only Base Set, Rookie Patch Autographs and Rookie Autographs have their counts stated by more than one site. No count anywhere in this release was derived by arithmetic from another figure.
• G-021 rookie_flag is UNKNOWN on all 100 base cards, not "no". tcdb renders NO RC token on any row 1-100 while rendering "RC" on every row from 101 up, on https://www.tcdb.com/Checklist.cfm/sid/198282/2018-19-Panini-National-Treasures?PageIndex=1 and ?PageIndex=2 . That is a real structural signal and it is what separates the base run from the autograph blocks, but it is NOT a statement that the players on rows 1-100 are not rookies -- and at least one of them plainly is: card 56 is Miles Bridges, a 2018-19 NBA rookie, rendered with no RC token. The honest reading is that this product reserves its RC designation for the autographed rookie blocks, so the base rows carry no rookie information at all. Per unknown_semantics.csv, "no" would express genuine absence and is not what was observed; UNKNOWN is written and this gap records why.
• G-022 NO NEW parallel_family WAS MINTED, and one judgement is recorded rather than acted on. The families written follow the catalogue's existing dominant conventions, checked against all parallels.csv files in data/: Bronze -> Bronze (93 existing rows), Emerald -> Emerald (205), Platinum -> Platinum (514), Gold -> other (1341 existing rows against 14 under "Metal", so "other" is the convention even though Gold is plainly a colour family). Prime -> other, matching all 217 existing "Prime" rows in the catalogue. Super Prime, Horizontal, Limited Edition and Limited Edition Blue have no precedent and are written "other"; Horizontal Bronze is written "Bronze" on the colour. OBSERVATION FOR A LATER PASS, not acted on here: Prime and Super Prime are a genuine patch-quality ladder of their own, distinct from any colour family, and the catalogue currently collapses 217+ of those rows into "other". Minting a "Prime" family in this one release would split them rather than fix them, so it was not done.

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