2020-21 Absolute Memorabilia

• G-001 base_print_run recorded as unnumbered/assumed. NO SOURCE STATES that the 2020-21 Absolute Memorabilia base cards are unserialised. The positive evidence is indirect and of exactly the shape EXTRACTION_SPEC 2.1 calls `assumed`, not `sourced`: tcdb annotates eight of the nine base parallel sets "Serial numbered to 199." / "to 99." / "to 75." / "to 49." / "to 25." / "to 10." / "to 5." / "One of One." and annotates nothing of the kind on the base set (https://www.tcdb.com/Inserts.cfm/sid/257946/2020-21-Panini-Absolute-Memorabilia); the parent Checklist.cfm flag column carries no SN token on any base row (https://www.tcdb.com/Checklist.cfm/sid/257946/2020-21-Panini-Absolute-Memorabilia); and cardboardconnection gives a "#/" figure to every base parallel except the Retail/Dollar Tree one and none to the base set (https://www.cardboardconnection.com/2020-21-panini-absolute-memorabilia-basketball-cards). The SAME reasoning is applied to the base "Retail" parallel row in parallels.csv, where the file has no basis column at all to record it in.
• G-002 rookie_flag is UNKNOWN on 98 of the 100 base cards because the ONLY source found is demonstrably incomplete, and an empty flag cell is therefore not evidence of genuine absence. tcdb renders an "RC" token on exactly two rows - card 18 Isaac Okoro and card 60 Jordan Nwora (https://www.tcdb.com/Checklist.cfm/sid/257946/2020-21-Panini-Absolute-Memorabilia) - and those two are written yes. A NARROW VERBATIM RE-FETCH (EXTRACTION_SPEC 4.1) of the same page was run precisely because two looked too few for a 2020-21 rookie class, and it confirmed the page: card 12 LaMelo Ball, card 15 Patrick Williams, card 59 Anthony Edwards and card 67 Obi Toppin each render with NO "RC" in the row. Those are indisputable 2020-21 rookies, so tcdb has simply not populated the RC column for this set, and writing rookie_flag=no would assert a genuine absence the evidence contradicts (CLAUDE.md: UNKNOWN never means "there is none"). SECOND HOLE: the flag column could not be read AT ALL for cards 79-100, because Checklist.cfm truncated at row 78 through the fetch layer and Gallery.cfm renders no flag column. CONTRAST WITH THE SIBLING: absolute-bkb-2019-20 got RC on 44 of its 100 base cards from the same endpoint, so this is a per-set data hole at tcdb, not an endpoint limitation. A second RC source (beckett, unreachable - see G-016) would resolve this.
• G-004 CEILINGS AND ALTERNATIVES NOT RESOLVED - 12 parallel rows carry UNKNOWN/UNKNOWN rather than a number, because the only figure available is a ceiling or an either/or and EXTRACTION_SPEC 2.3 forbids writing either as a print run. cardboardconnection (https://www.cardboardconnection.com/2020-21-panini-absolute-memorabilia-basketball-cards) publishes: Retired Autographs Level 1 - "Level 2 - #/25 or less, Level 3 - #/10 or #/5, Level 4 - #/5 or #/3"; Veteran Autographs Level 1 - the identical line; Tools of the Trade Three / Four / Six Swatch Signatures Level 1 - "Level 2 - #/49 or #/25"; Jumbo Hat Team Logo - "Brand Logo #/10 or less, Button #/10 or #/5, Laundry Tag #/10 or #/5". That is 3 rows (Retired Autographs Levels 2-4) + 3 rows (Veteran Autographs Levels 2-4) + 3 rows (one Level 2 each for the three Tools of the Trade sets) = 9 Level rows, plus the 3 Jumbo Hat rows = 12 rows in total, and the Jumbo ceilings are independently PROVEN real by tcdb's per-row SN flags rather than being loose wording: Jumbo Hat Brand Logo renders SN10, SN9 and SN5 on different rows, Button renders SN10 and SN5, Laundry Tag renders SN10 on 36 cards and SN5 on 2. Resolving these needs a per-card source (tcdb Level-tier subsets carry per-row SN flags and each has its own sid, listed at https://www.tcdb.com/Inserts.cfm/sid/257946/2020-21-Panini-Absolute-Memorabilia); that is 44 further fetches and was out of budget for this pass.
• G-005 SERIAL NUMBERED, RUN UNKNOWN - the state EXTRACTION_SPEC 2.3 says the schema cannot express, on 12 of this release's 58 parallel rows. Every row listed in G-004 is PROVABLY serial numbered: cardboardconnection writes each as "#/N or ..." rather than as unnumbered, and for the three Jumbo Hat rows tcdb prints an explicit SN token on every single card (https://www.tcdb.com/Inserts.cfm/sid/257946/2020-21-Panini-Absolute-Memorabilia and the subset pages cited in parallels.csv). The forced pairing of print_run and serial_numbered collapses all fifteen to UNKNOWN/UNKNOWN, which a consumer cannot distinguish from "we found nothing at all". Held here because the CSV cannot carry it. This is the same loss the 2019-20 sibling recorded in its G-004 on 6 of 61 rows.
• G-007 STRUCTURAL CALL: the Level 2-5 tiers are filed as 44 insert-scope parallel rows, NOT as 44 sets.csv rows, even though tcdb catalogues each of them as a separate subset with its own sid and its own Total Cards (https://www.tcdb.com/Inserts.cfm/sid/257946/2020-21-Panini-Absolute-Memorabilia lists all 65 insert sids). Two reasons: cardboardconnection frames them as parallels in so many words ("PARALLEL CARDS: Level 2 - #/25, Level 3 - #/10, Level 4 - #/5, Level 5 - 1/1.", https://www.cardboardconnection.com/2020-21-panini-absolute-memorabilia-basketball-cards), and the 2019-20 sibling absolute-bkb-2019-20 made the same call, so a consumer comparing the two seasons of this line sees one convention. THE COST IS REAL AND IS RECORDED HERE: tcdb holds a card_count for each of those 44 tiers and this release does not carry it, because parallels.csv has no card_count column and sets.csv would be the wrong home for a tier. If the catalogue later decides Level tiers are sets, this release and absolute-bkb-2019-20 must both be reworked, not just one.
• G-008 STRUCTURAL CALL: five tcdb subsets are filed as insert-scope parallels rather than as sets - Jumbo Basketball NBA Logo (264173) and Jumbo Basketball Spalding Logo (264174) scoped to "Jumbo Basketball Spalding Name", and Jumbo Hat Brand Logo (264176), Jumbo Hat Button (264177) and Jumbo Hat Laundry Tag (264178) scoped to "Jumbo Hat Team Logo". EVIDENCE FOR THE CALL, which is stronger here than the aggregator framing alone: all three Jumbo Basketball subsets render the IDENTICAL 38 card codes JB-AED..JB-ZNN in the same order and differ only in the memorabilia piece and the serial flag (SN20 / SN15 / SN5), and all four Jumbo Hat subsets render the identical 38 codes JH-AED..JH-ZNN. cardboardconnection agrees, listing them under the parent set headings as "PARALLEL CARDS: Spalding Logo #/15, NBA Logo #/5." and "PARALLEL CARDS: Brand Logo #/10 or less, Button #/10 or #/5, Laundry Tag #/10 or #/5." (https://www.cardboardconnection.com/2020-21-panini-absolute-memorabilia-basketball-cards). The 2019-20 sibling made the same split. COUNTER-ARGUMENT RECORDED: a hat button and a hat brand logo are different physical objects, not a colour variation, so a reader who thinks of a parallel as a finish will find these filed where they do not expect them. WHICH SIBLING IS THE "SET" IS ARBITRARY: Spalding Name and Team Logo were chosen because they are the highest-print tier and because the 2019-20 sibling chose them.
• G-009 sets.csv CANNOT HOLD A SET'S OWN PRINT RUN (EXTRACTION_SPEC 2.2) and eight sets in this release have one, so the figures live only in notes prose and here. UNIFORM, therefore a real set-level property: Future Signatures Level 1 - tcdb "AU, SN49" on every row and cardboardconnection "39 cards. Serial numbered #/49."; Jumbo Basketball Spalding Name - tcdb "MEM, SN20" on every row and cardboardconnection "38 cards. Serial numbered #/20."; Rookie Laundry Tag - tcdb "MEM, SN5" and cardboardconnection "20 cards. Serial numbered #/5."; Tools of the Trade Six Swatch Signatures Level 1 - tcdb "AU, MEM, SN99" on every row. NOT UNIFORM, therefore no figure exists to record even in prose: Jumbo Hat Team Logo (SN20, but SN10 on LaMelo Ball, RJ Hampton, Vernon Carey Jr. and Zeke Nnaji); Tools of the Trade Three and Four Swatch Signatures Level 1 (SN199, but SN99 on Anthony Edwards, James Wiseman and LaMelo Ball); Veteran Autographs Level 1 (SN49, but SN25 on Donovan Mitchell, Ja Morant, Kristaps Porzingis, Lonzo Ball, RJ Barrett and Trae Young, and SN10 on Stephen Curry). All read off the subset pages cited in sets.csv notes. Same defect as the sibling's G-006.
• G-010 TWO NAME/CODE COLLISIONS INSIDE THIS RELEASE, recorded rather than renamed. (1) CARD-CODE PREFIX COLLISION: "Retired Autographs Level 1" uses RA- (RA-BOB Bob ... RA-WFR, 30 codes, sid 264181) and "Rookie Autographs Level 1" ALSO uses RA- (RA-AED Anthony Edwards ... RA-ZNN, 48 codes, sid 264185). The rosters are disjoint - retired players against 2020-21 rookies - so no single code is claimed by both, but a resolver keyed on code alone cannot tell the sets apart. The 2019-20 sibling recorded the identical collision in its G-009, so this is a property of the line, not a one-season slip. (2) SET-AGAINST-PARALLEL NAME COLLISION: the insert set "Rookie Laundry Tag" (sid 264195, RL- codes) and the parallel "Laundry Tag" scoped to "Jumbo Hat Team Logo" (sid 264178, JH- codes) are different objects with near-identical names; name_ambiguous=yes and ambiguous_with="Rookie Laundry Tag" are set on the parallel row. (3) WITHIN "Rookie Autographs Level 1" the rendered code list contains "RA-TMA" TWICE - two different players sharing initials, most plausibly Theo Maledon and Tyrese Maxey, both of whom are in the base set. The same doubled "JH-TMA" appears in the Jumbo Hat Team Logo code list. Recorded verbatim, not corrected; the card-level pass (CHASE_CHECKLIST_SPEC) must resolve it rather than deduplicating.
• G-011 CONVENTION DIVERGENCE AGAINST THE 2019-20 SIBLING, DELIBERATE, ON 3 ROWS. absolute-bkb-2019-20 writes print_run=10 / serial_numbered=yes on its "Brand Logo", "Button" and "Laundry Tag" rows scoped to Jumbo Hat Team Logo. This release writes UNKNOWN/UNKNOWN on the same three names in the same scope, because cardboardconnection states them as ceilings ("Brand Logo #/10 or less, Button #/10 or #/5, Laundry Tag #/10 or #/5", https://www.cardboardconnection.com/2020-21-panini-absolute-memorabilia-basketball-cards) and because EXTRACTION_SPEC 2.3, as revised 2026-09-12, says a ceiling is not a print run and explicitly REVERSES the earlier practice. SAME LINE, ADJACENT SEASONS, SAME EVIDENCE SHAPE, OPPOSITE VALUE, and nothing in either row tells a reader which convention it is under. A repair pass over absolute-bkb-2019-20 would make the line internally consistent; until then a consumer comparing the two seasons will read the 2019-20 rows as more precise when they are not.
• G-012 card_count UNIT CHECK (EXTRACTION_SPEC 5.4): every count in sets.csv is DISTINCT CARDS in that named set. Each was read two ways - the tcdb Total Cards header on the set's own page AND the rendered rows or enumerated card codes on that same page - and then compared against cardboardconnection's section heading count (https://www.cardboardconnection.com/2020-21-panini-absolute-memorabilia-basketball-cards). All 17 rows agree across both sources: 100, 30, 39, 25, 38, 38, 30, 48, 48, 20, 40, 25, 32, 32, 32, 29, 30. None of the four wrong units is present: no count here is a sum of parallel-tier lines (the Level tiers are in parallels.csv), none is a signer count (the code lists enumerate one code per card), and none is a parallel roster. TWO EXCEPTIONS TO "READ TWO WAYS": Rookie Laundry Tag rendered only 16 of its stated 20 rows (see G-019), and Jumbo Hat Team Logo needed a second narrow fetch before its 38 codes could be enumerated (see G-017). ⚠ THE TWO SOURCES MAY BE ONE OPINION: cardboardconnection is one of the aggregators EXTRACTION_SPEC 3.3 and 5.4 warn copy each other, and beckett - the usual third structural source - is unreachable (G-016). The agreement is still meaningful as a CALIBRATION: cardboardconnection matched tcdb on 17 of 17 counts including the base set, which is why its single-source figures elsewhere in this release (the Level ladder print runs) were trusted enough to write.
• G-014 BRAND-NAME TITLE ANOMALY, CHECKED AND DISMISSED, RECORDED SO THE NEXT AGENT DOES NOT RE-OPEN IT. cardboardconnection's page TITLE for this product reads "2020-21 Panini Absolute Basketball Checklist, Boxes, Set Details, Reviews" - dropping "Memorabilia" - while the SAME page's H1 reads "2020-21 Panini Absolute Memorabilia Basketball Cards" and its URL keeps the word (https://www.cardboardconnection.com/2020-21-panini-absolute-memorabilia-basketball-cards). This is the editorial house-style shortening EXTRACTION_SPEC 2.1 warns about, not a name change: tcdb (https://www.tcdb.com/ViewSet.cfm/sid/257946/2020-21-Panini-Absolute-Memorabilia), beckett, gogts.net and checklistcenter all render "Absolute Memorabilia" for this season. product_name is therefore "Panini Absolute Memorabilia Basketball", matching reference/brands.csv byte for byte, and NO brand-identity escalation is raised. The task briefing asked this to be verified rather than assumed; it was, and the scope memo is correct.
• G-015 LAST-SEASON CLAIM RESTS ON ONE SOURCE, POSITIVELY BUT NARROWLY. tcdb's season index for 2021 basketball (https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2021) returns NO product whose name contains "Absolute", while the 2020 index (https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2020) returns "2020-21 Panini Absolute Memorabilia" (sid 257946) - so on tcdb, 2020-21 is the last Absolute Memorabilia basketball release. The same 2021 index carries "2021-22 Panini Chronicles" (sid 320462), which is consistent with the briefing that Absolute becomes a Chronicles SUBSET from 2021-22; that subset was NOT opened or verified in this pass and no object in this release is a Chronicles object. An absence on one catalogue is weaker evidence than a presence: a Panini announcement or a second catalogue would be needed to state the discontinuation as fact rather than as "not indexed".
• G-016 beckett.com could not be consulted. A search result exposes the article title "2020-21 Panini Absolute Memorabilia Basketball Checklist, Team Set Lists and Details" at https://www.beckett.com/news/2020-21-panini-absolute-memorabilia-basketball-cards/ , and that title is used in release.csv notes ONLY as brand-name corroboration, which is all a search result can support. The page itself was not fetched - EXTRACTION_SPEC 3.3 records that beckett 302-redirects to a maintenance page, and the 2019-20 sibling recorded the same in its G-012. The practical cost is that the usual THIRD independent structural source is missing, so every card count in this release rests on tcdb plus one aggregator that may be echoing it (see G-012), and the rookie_flag hole in G-002 has no second opinion available.
• G-017 TWO FETCH-LAYER ANOMALIES ON tcdb SUBSET PAGES, both recorded rather than silently smoothed. (1) Jumbo Hat Team Logo (sid 264179): the first summarizing fetch reported "31 rows rendered" against a header of Total Cards "38" - which is exactly the shape EXTRACTION_SPEC 3.2 says to treat as a question. A narrow re-fetch enumerating every code returned 38 codes ending JH-ZNN and re-quoted Total Cards "38", so 38 is written and the 31 was a fetch artefact. THE CHECK FIRED CORRECTLY AND CHANGED NOTHING, which is the point of running it. (2) Rookie Autographs Variation Level 1 (sid 264190): the fetch reported the FIRST rendered card code as "JUM-THA" while the last is "RAV-ZNN" and every other row it described carries the RAV- prefix. "JUM-" belongs to no set in this release. Either tcdb really carries a stray code on that row or the fetch layer corrupted it; it was not resolved, and the card-level pass must read that row directly rather than assuming RAV-.
• G-018 team_or_school RECORDS tcdb'S CURRENT PLAYER-TEAM ATTRIBUTION, WHICH IS NOT NECESSARILY THE TEAM PRINTED ON THE CARD. The base run is ordered by team, and four rows break that ordering in a way that shows the attribution was updated after a trade: card 10 Jarrett Allen is "Cleveland Cavaliers" but sits inside the Brooklyn block (8-9) ahead of Charlotte (11-13), with Cleveland proper at 17-19; card 36 James Harden is "Brooklyn Nets" but sits in the Houston block beside card 38 Eric Gordon "Houston Rockets"; card 37 Russell Westbrook is "Washington Wizards" in that same Houston block, with Washington proper at 98-100; card 60 Jordan Nwora is "Milwaukee Bucks" inside the Minnesota block (58-59, 61). All four players moved during 2020-21, and this product released 2021-03-24, after those moves. The values are written as the source renders them (https://www.tcdb.com/Checklist.cfm/sid/257946/2020-21-Panini-Absolute-Memorabilia) and are NOT corrected. A consumer matching cards by team will mismatch these four. INCIDENTALLY THIS IS THE TEMPORAL CHECK PASSING (EXTRACTION_SPEC 4.3): every roster placement is consistent with a March 2021 release and none is impossible for the season.
• G-019 SINGLE-SOURCE RENDERED-ROW COUNT: Rookie Laundry Tag (sid 264195) states Total Cards "20" but only 16 rows survived the fetch, RL-AED through RL-THA, so the count was NOT confirmed by counting rendered rows the way the other sixteen sets were. It is written as 20 because the tcdb header and cardboardconnection's "20 cards. Serial numbered #/5." (https://www.cardboardconnection.com/2020-21-panini-absolute-memorabilia-basketball-cards) agree. The four unseen rows are presumably players after "T" alphabetically by the set's own ordering. EXTRACTION_SPEC 3.2 warns that Total Cards counts different things on different lines; here it could not be checked against the rendered rows, so it is a single unconfirmed structural figure in an otherwise doubly-read file.
• G-020 EXTRACTION_SPEC 4.5 SUSPICIOUS-UNIFORMITY CHECK RUN AGAINST THE 2019-20 SIBLING, AND PASSED - recorded because a template-fill would look exactly like this pass if it were not checked. The two seasons of this line genuinely differ: base parallels went 8 -> 9 (Teal #/49 is new in 2020-21); the set inventory went 18 -> 17 rows with "Limitless Signatures Level 1" present in 2019-20 and ABSENT in 2020-21; and counts moved independently rather than together - Future Signatures 38 -> 39, Rookie Autographs 40 -> 48, Rookie Autographs Variation 40 -> 48, Tools of the Trade (all three) 33 -> 32, Jumbo families 40 -> 38, Veteran Autographs 29 -> 29, Rookie Threads 40 -> 40, Glass 25 -> 25, base 100 -> 100. The Level ladders also differ: 2019-20 Established Threads reads "Level 2 - #/25 or less" where 2020-21 reads "Level 2 - #/25" flat. The structure is inherited because it is the same product line, not because a template was filled.

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