2018-19 Obsidian

• G-001 DEBUT CLAIM TESTED AND CONFIRMED for basketball, with two qualifications the scope memo does not carry. The memo asserts 2018-19 is Obsidian's debut season and that there is nothing before it. Nothing before it in NBA is supported: tcdb's full 2017-18 basketball product index https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2017 contains no Obsidian entry and does render National Treasures, Noir, Opulence, Prestige and Prizm, so the alphabetical neighbourhood is populated. Qualification one, the line is not new: https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards states "The trading card line was initially issued as a 2018 NFL set and also appeared in the multi-brand 2018-19 Chronicles NBA." and tcdb confirms "2018 Panini Obsidian" in football at sid 188545 via https://www.tcdb.com/ViewAll.cfm/sp/Football/year/2018. Because brands.csv is unique on brand plus sport, a football Obsidian does not collide with the basketball brand row, but "debut season" is true of the sport and not of the line. Qualification two, the section 5.1 brand-identity trap fires exactly as anticipated: the Obsidian name ships inside another NBA product in the SAME season as a subset. https://www.sportscardchecklist.com/set-173319/2018-19-panini-chronicles-base-obsidian-preview-basketball-trading-card-checklist renders "2018-19 Panini Chronicles Base Obsidian Preview Basketball Checklist" with 15 cards numbered 571 to 585 inside the Chronicles base run, and web search surfaces Beckett pages for its own colour tiers. Because that subset is same-season rather than earlier, and because a name inside Chronicles is not that line shipping, it does not predate the standalone debut and must not be counted as an Obsidian release. Note tcdb does NOT list it: https://www.tcdb.com/Inserts.cfm/sid/205013/2018-19-Panini-Chronicles renders no set whose name contains "Obsidian", consistent with tcdb filing Chronicles subsets as ranges of the parent base run rather than as named sets, so an agent checking only tcdb would have concluded the Chronicles appearance did not exist. brands.csv was NOT touched; the catalogue-level correction implied is that the Obsidian basketball line begins at 2018-19 rather than 2019-20.
• G-002 RELEASE DATE CONTESTED by one day and the value written picks a side. tcdb renders "Release Date: August 28, 2019" at https://www.tcdb.com/ViewSet.cfm/sid/213957/2018-19-Panini-Obsidian. https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards renders "Release Date: August 29, 2019" and https://www.checklistcenter.com/2018-19-panini-obsidian-basketball/ renders "August 29, 2019". Neither source labels its date an estimate, so this is a genuine disagreement and not an estimate-versus-fact question. 2019-08-29 is written because two sources state it against tcdb's one, and release.csv has no column in which to flag a contested value. The two agreeing sources are both aggregators of the kind the extraction spec warns may be one opinion counted twice, so the majority is weaker than two-to-one suggests. Beckett was not consulted; the brief records it as unreachable for six consecutive passes. A single fetch of a fourth independent source would settle this.
• G-003 base_print_run=unnumbered carries base_print_run_basis=assumed, not sourced. No source says in words that the base cards carry no serial number. The evidence is a strong WITHIN-PRODUCT positive discriminative test, the strongest form the extraction spec describes: the same tcdb renderer prints no flag token whatever on base rows 1-70 and only RC on rows 71-100, across https://www.tcdb.com/Checklist.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=1 and gallery pages 8, 9 and 10, while printing SN49, SN25, SN15, SN10, SN5 and SN1 on the rows of this product's six base parallel tiers. Strong indirect evidence is still not a statement, so the basis is assumed.
• G-004 THE BLANK-NOTES DISCRIMINATIVE TEST IS REFUTED ON THIS PARENT, and this is a calibration result the extraction spec asks for by name. The spec records that a blank Notes annotation on tcdb Inserts.cfm is a positive discriminative signal for unnumbered, measured on the 2018-19 Prizm parent where 25 of 37 base parallel rows carried "Serial numbered to N" and 11 carried nothing. On the Obsidian parent https://www.tcdb.com/Inserts.cfm/sid/213957/2018-19-Panini-Obsidian the Notes column is BLANK on ALL 25 rendered rows - 19 under "Insert Sets (19)" and 6 under "Parallel Sets (6)". There is no split, and the blank cells sit beside tiers that provably ARE numbered: Electric Etch Purple has a blank Notes cell, and its own page https://www.tcdb.com/ViewSet.cfm/sid/217627/2018-19-Panini-Obsidian-Electric-Etch-Purple renders SN49 on its rows while https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards states "Purple #/49". Writing unnumbered off the blank annotation would have been FALSE on all six base tiers, not merely unsupported. The operative rule is that the signal requires a parent that renders a SPLIT; an all-blank Notes column means the parent carries no annotation data, which is not evidence about any tier. The spec's warning that truncation on this page is parent-dependent should be widened: the ANNOTATION is parent-dependent too.
• G-005 print_run and serial_numbered are UNKNOWN on all 15 set-scope parallel rows. No source consulted states a print run for any colour tier of Galaxy Autos, Matrix Autographs or Rookie Jersey Autographs: the tcdb Inserts.cfm Notes column is blank on every row of this parent, and https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards and https://www.checklistcenter.com/2018-19-panini-obsidian-basketball/ give card counts for three of the Galaxy tiers and no print run for any tier. Two tiers were probed directly and DO carry serial tokens, so the tiers are provably serial numbered with runs that a later pass can recover one sid at a time: https://www.tcdb.com/ViewSet.cfm/sid/276708/2018-19-Panini-Obsidian-Galaxy-Autos-Electric-Etch-Yellow renders "Total Cards: 30" with the tokens "AU, SN10" on the five rows quoted back, and https://www.tcdb.com/ViewSet.cfm/sid/221546/2018-19-Panini-Obsidian-Matrix-Autographs-Electric-Etch-Orange renders "Total Cards: 30" with "AU, SN50" on the five rows quoted back. Five rows of thirty is not the tier, so no integer was written. This is the schema expressiveness gap the extraction spec records as recurring: serial_numbered=yes with print_run=UNKNOWN cannot be expressed, so the held fact - these tiers ARE numbered, to 10 and 50 respectively on the rows seen - lives only here. Recovering all 15 costs 15 fetches.
• G-006 FETCH-LAYER ASSERTION REJECTED. The fetch of https://www.tcdb.com/ViewSet.cfm/sid/221546/2018-19-Panini-Obsidian-Matrix-Autographs-Electric-Etch-Orange quoted five rows all carrying "AU, SN50" and then asserted in its own words that the serial token varies between SN50 and SN25 across the tier depending on player rarity. No row carrying SN25 was produced. The extraction spec's rule that a summarising fetch's own arithmetic and glosses are not evidence was applied and the SN25 claim was not written anywhere. Recorded because the assertion may well be TRUE - the Rookie Jersey Autographs parent does render a mixed SN75 and SN50 - and a later pass should test it on the rows rather than inherit it from this note.
• G-007 CARD COUNT ANOMALY on Galaxy Autos, unresolved. The set renders "Total Cards: 14" at https://www.tcdb.com/ViewSet.cfm/sid/276703/2018-19-Panini-Obsidian-Galaxy-Autos, and https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards and https://www.checklistcenter.com/2018-19-panini-obsidian-basketball/ independently state 14. Its own colour tiers render LARGER rosters: https://www.tcdb.com/ViewSet.cfm/sid/276708/2018-19-Panini-Obsidian-Galaxy-Autos-Electric-Etch-Yellow renders "Total Cards: 30" for the Yellow tier and https://www.checklistcenter.com/2018-19-panini-obsidian-basketball/ states "Galaxy Auto Electric Etch Orange Set: 20 Cards", "Galaxy Auto Electric Etch Green Set: 25 Cards" and "Galaxy Auto Electric Etch Yellow Set: 30 Cards". A colour parallel cannot contain more cards than the set it parallels, so one of two things is true and neither was established: the 14 is a partial roster that tcdb and both aggregators have copied, or the colour tiers are not parallels of this set but separate sets sharing its prefix. This is the section 5.4 problem - the unit inside card_count is not settled. 14 is written for the set because three sources state it, and the tiers stay in parallels.csv because tcdb files them under the parent's Insert Sets heading. Note the same tier carries a plain integer card number "1" for Kobe Bryant alongside GA- prefixed codes, in both the 14-card parent and the 30-card Yellow tier, which is another sign the parent roster is not simply a subset.
• G-008 SET NAME DIVERGENCE on Galaxy Autos, recorded because a resolver has to match what sources print. tcdb renders "Galaxy Autos" at https://www.tcdb.com/Inserts.cfm/sid/213957/2018-19-Panini-Obsidian; https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards renders "Galaxy Autographs"; https://www.checklistcenter.com/2018-19-panini-obsidian-basketball/ renders "Galaxy Auto Set". The sibling release obsidian-bkb-2019-20 carries "Galaxy Autographs" in its sets.csv. tcdb's spelling is written here because tcdb is the enumerating source for this release's set inventory, which is the convention the prizm-bkb-2018-19 pilot used for the same situation, but the consequence is that this release and its own sibling season are not joinable on set_name without an alias.
• G-009 PARALLEL FAMILY MINTED: "Electric Etch" is written on all 21 parallel rows, and it does not exist anywhere else in the catalogue. Reported as the extraction spec requires. Electric Etch is Panini's own name for this line's etched-foil treatment and it is part of every tier name the sources print, so it is a genuine visual system rather than a colour label. DIVERGENCE: the sibling release obsidian-bkb-2019-20 files the identical tiers with parallel_family "other" for the five colours and "Mojo" for Electric Etch White Mojo. Two seasons of one line now answer the family question differently and nothing on either row says which convention it is under. "other" is the information-free value and "Mojo" splits a tier whose printed name begins "Electric Etch", which is why this build diverged rather than copied; a catalogue-level decision should settle it across the whole line and the other Obsidian seasons.
• G-010 CHANNEL IS INHERITED FROM A PRODUCT-LEVEL STATEMENT, not sourced per tier. All 21 rows carry channel=panini-direct with channel_basis=sourced-exclusive on the strength of two sources describing the whole product as a Panini-website exclusive: https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards renders "the set is exclusive to Panini's online store" and https://www.checklistcenter.com/2018-19-panini-obsidian-basketball/ renders "2018-19 Panini Obsidian is a release that is exclusive to the Panini website". No source names a channel for any individual tier. The inheritance is judged legitimate here because the claim is exhaustive about where any card in the product could be bought, which is a different shape from the product-level hobby note the extraction spec records an agent correctly declining to push onto tiers it did not cover. Recorded so a reviewer can disagree: if the inheritance is rejected, every one of these rows becomes channel=UNKNOWN with an empty channel_basis. Note also that the product being website-exclusive is why no FOTL, blaster, mega or retail tier exists in this release, unlike the sibling season, which carries Electric Etch Blue FOTL and Electric Etch Contra FOTL.
• G-011 No source found states a licensing grantor for this product. license_status=licensed (NBA), licensed=yes and league=NBA are written to match the catalogue-wide value and are supported only by the product being an NBA-team, NBA-player checklist of 100 current and retired NBA players on NBA franchises. This is the same absence recorded catalogue-wide rather than a gap specific to this release. Sources: https://www.tcdb.com/Checklist.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=1 ; https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards
• G-012 rookie_flag rests on a single structural source. Thirty of the 100 base rows, numbers 71-100, carry a tcdb RC designation and the other 70 carry none; no second source was consulted for rookie status on any card. The no values are written rather than UNKNOWN because the same renderer marks RC on sibling rows of the same page, which is a positive discriminative test, but the whole column is one source deep. Sources: https://www.tcdb.com/Checklist.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=1 ; https://www.tcdb.com/Gallery.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=8 ; https://www.tcdb.com/Gallery.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=9 ; https://www.tcdb.com/Gallery.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=10
• G-013 The base checklist is assembled from two endpoints with a measured delivery truncation, so the joins matter. https://www.tcdb.com/Checklist.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=1 delivered rows 1-77 of the server's 100-row page one and stopped; rows 78-100 came from https://www.tcdb.com/Gallery.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=8, https://www.tcdb.com/Gallery.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=9 and https://www.tcdb.com/Gallery.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=10, each of which rendered exactly 10 tiles. Rows 71-77 were rendered by BOTH endpoints and agree on card number, player name, team and the RC designation, which is the only cross-endpoint check available. Rows outside that overlap - 1-70 from Checklist only and 78-100 from Gallery only - rest on one rendering each. Card 15 was additionally corroborated by a third rendering, the Electric Etch Red tier page.
• G-014 TEMPORAL ODDITY recorded, not repaired. Card 15 is rendered "Anthony Davis, Los Angeles Lakers" on a product whose season is 2018-19, a season Davis played for the New Orleans Pelicans; he was traded to the Lakers in July 2019. The temporal-impossibility check was run and does NOT fire: this product released at the end of August 2019, after the trade, and cards 25 and 35 in the same run are rendered Julius Randle and Jrue Holiday on the New Orleans Pelicans, so the checklist is not uniformly post-trade. The team value is confirmed by two independent tcdb renderings - https://www.tcdb.com/Checklist.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=1 and https://www.tcdb.com/ViewSet.cfm/sid/217628/2018-19-Panini-Obsidian-Electric-Etch-Red - so it is what the source prints rather than a fetch artefact. team_or_school is written as rendered. What is NOT established is whether the card itself pictures Davis in a Lakers uniform or whether tcdb has applied a later team assignment.
• G-015 No pack odds were found for any set or parallel tier in this release. stated_odds, odds_evidence_url and odds_evidence_quote are empty on every row of sets.csv and parallels.csv, and scarcity_tier is empty rather than guessed. Neither https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards nor https://www.checklistcenter.com/2018-19-panini-obsidian-basketball/ prints an odds figure, and tcdb prints none anywhere. Recorded so the emptiness reads as an evidenced absence of stated odds rather than as unfinished work.
• G-016 Card counts for the 15 set-scope parallel tiers were not sourced except where noted. tcdb's Inserts.cfm carries no card-count column, so each tier's size costs its own fetch; two were spent, on the Galaxy Autos Yellow and Matrix Autographs Orange tiers, both rendering "Total Cards: 30". https://www.checklistcenter.com/2018-19-panini-obsidian-basketball/ states counts for three Galaxy tiers. sets.csv carries no rows for these tiers because they are parallels rather than sets, and parallels.csv has no card_count column, so the counts that WERE observed are recorded only in prose here and in visual_notes.
• G-017 The section 5.3 no-sid block pattern is ABSENT in this season, and that is a divergence from this line's own next season worth stating. Checked both boundaries: https://www.tcdb.com/Checklist.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=2 renders no checklist rows at all, and https://www.tcdb.com/Gallery.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=11&sort=0 renders no tiles with the gallery pagination offering pages 1 through 10 only, so the base run genuinely ends at 100 and the last rendered page rather than the header was used to establish it. The parent's own header agrees at "Total Cards: 100", which is the case in which the header happens to be right. By contrast obsidian-bkb-2019-20 files Rookie Jersey Autographs at 201-242 inside its parent's numbering past a 200-card base, and its release notes record tcdb stating "Total Cards: 241". Same line, adjacent seasons, opposite structure. Here that is a fact about the products rather than a convention divergence: there was no block to include or exclude, so base_set_size=100 is not a choice between the two conventions the extraction spec records in conflict.
• G-018 No cross-filed or hidden Obsidian set was found and the search was NOT exhaustive. Checks run: https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2018 lists exactly one Obsidian product for the season, the parent at sid 213957; https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2019 lists only "2019-20 Panini Obsidian" at sid 227999 and no entry titled 2018-19, so there is no next-season Update block holding 2018-19 Obsidian cards - the mirror check the extraction spec requires before concluding a number is unissued; https://www.tcdb.com/Checklist.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=2 renders no rows and https://www.tcdb.com/Gallery.cfm/sid/213957/2018-19-Panini-Obsidian?PageIndex=11&sort=0 renders no tiles; and the parent's error and variation page https://www.tcdb.com/Errors.cfm/sid/213957/2018-19-Panini-Obsidian renders "No errors or variations have been added for this set yet.", so no variation row exists to widen the base run or to force a non-integer base card number. What was NOT done is a probe of every sid gap inside the tier block 217625-217630 or the insert blocks 221544-221550 and 276697-276708, so a set with a sid in those ranges that Inserts.cfm omits would not have been seen. The extraction spec records that Inserts.cfm can omit real sets entirely. [CORRECTED 2026-09-20: THE Errors.cfm SENTENCE QUOTED ABOVE IS NOT EVIDENCE AND THE CONCLUSION DRAWN FROM IT DOES NOT HOLD. Measured 2026-09-20 on tcdb sid 182809: Errors.cfm returns that identical sentence for a set whose own renderings carry UER on at least two cards (card 260 Mo Bamba, "UER: Statline says Undrafted, was actually drafted 6th", confirmed under &sort=0; card 67 Justin Jackson, "Career stat line is that of Kosta Koufos"). The endpoint returns the same string on a set that provably HAS errata and on one that appears not to, so it discriminates nothing and its silence is uninformative. WHAT SURVIVES: any claim here about base_set_size, about the base run being contiguous, or about no numbered block sitting past the base run, because those rest on the independent contiguous enumeration also cited in this note - the last rendered Gallery page, the empty page after it, and the per-row flag column. WHAT DOES NOT SURVIVE: the narrower claim that NO VARIATION SET EXISTS. A variation set renders as b-suffixed rows (101a/101b) that a base-run enumeration is not designed to surface, and Errors.cfm was the documented route to it, so that claim is now unsupported rather than disproved - it may still be true. Not re-fetched; recorded so it is not read as settled. See reference/caveats.csv CAV-020.]
• G-019 POSSIBLE SINGLE-OPINION CORROBORATION. cardboardconnection and checklistcenter agree with each other on the release date, the base set size, all six base parallel names and all six print runs, and on four of the five set names and counts. They are the two aggregators the extraction spec names as copying each other, and no third editorial source was reached: Beckett is recorded unreachable and was not spent on. What makes this build stronger than that caveat is that the print runs do NOT rest on those two alone - each of the six was independently confirmed by a tcdb serial token rendered on that tier's own set page, and tcdb is a structurally different kind of source. The release DATE has no such independent confirmation and is where the single-opinion risk actually bites.
• G-020 BRAND IDENTITY CHECKED AND CLEAN, recorded because the check is required. The product's name is rendered "2018-19 Panini Obsidian Basketball Cards" by https://www.cardboardconnection.com/2018-19-panini-obsidian-basketball-nba-cards and the set title "2018-19 Panini Obsidian" by tcdb at https://www.tcdb.com/ViewSet.cfm/sid/213957/2018-19-Panini-Obsidian, which drops the sport word as tcdb does for every basketball product. reference/brands.csv carries the Obsidian brand with product_name "Panini Obsidian Basketball", the season-stripped form of the cardboardconnection name, so product_name was written byte for byte from brands.csv and asserted against it in the build script. Nothing was minted and brands.csv was not touched. The one identity question this release does raise is the first-season question recorded in G-001, which is a catalogue decision and not a name conflict.
• G-021 STRUCTURAL NOTE on scope_ambiguous, and on one tier that is genuinely base-only. Five of the six Electric Etch colour names - Green, Orange, Red, White Mojo and Yellow - occur in FOUR scopes each in this release: base-set plus Galaxy Autos, Matrix Autographs and Rookie Jersey Autographs. Those 20 rows carry scope_ambiguous=yes with the scope list. Electric Etch Purple occurs in the base scope ONLY: no Purple tier is catalogued for any of the three autograph sets at https://www.tcdb.com/Inserts.cfm/sid/213957/2018-19-Panini-Obsidian, which renders five colour tiers per set and six under Parallel Sets. Its row carries scope_ambiguous=no. Separately, the literal strings the SOURCES print for the set-scope tiers carry the set name as a prefix - tcdb renders "Matrix Autographs Electric Etch Orange", not "Electric Etch Orange" - so a consumer matching on printed strings will not see the collision that scope_ambiguous is flagging, and a consumer matching on the names written here will. The sibling release obsidian-bkb-2019-20 writes the bare tier name in the same way.
• G-022 FABRICATION CHECKS RUN AND THEIR RESULTS. Copy check: Vitreous 1-10 is not base 1-10 renamed - the two rosters share no player in any position - and no other set's roster matches the base run. Set-name cross-check: all five set names were verified on at least two independent sources and the one divergence is gapped. Temporal check: the only anomaly is card 15 and it is gapped; Kobe Bryant and Charles Barkley appear only in Vitreous and Galaxy Autos, which are legends content and not season-impossible. Suspicious uniformity: the six base tiers ARE a clean 49/25/15/10/5/1 ladder, which is a template smell, but it is stated verbatim by two editorial sources and independently confirmed by six distinct tcdb serial tokens on six distinct sids, so it is a real ladder and not a template fill. Source calibration: cardboardconnection was tested on six questions this build could check independently - the six print runs - and passed six of six against tcdb tokens, before being relied on for the release date and the debut wording. Narrow verbatim re-fetch: used on the three cardboardconnection sentences underlying the debut finding, which the first fetch had truncated mid-sentence; the re-ask was shaped as three enumerated sentence requests and returned all three in full.

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