2017-18 Revolution

• G-001 CHANNEL DECLINED on 28 of this release's 29 parallel rows: channel=UNKNOWN with channel_basis EMPTY on the nine non-Chinese-New-Year base-set tiers (Astro, Fractal, Galactic, Groove, Impact, Cosmic, Sunburst, Cubic, Lava) and on all nineteen insert-scope tiers. The only distribution evidence found anywhere is product-level box configuration: cardboardconnection renders "*Hobby*: 5 cards per pack, 8 packs per box, 16 boxes per case (2 inners)" and "*Int Retail*: 5 cards per pack, 12 packs per box, 8 boxes per case." That is a pack/box configuration for the PRODUCT, and under CAV-010 a named channel value asserts EXCLUSIVITY to that channel; a configuration line attributes no tier to any channel and would be sourced-presence at best even if it did. Note the product ships in BOTH a hobby and an international retail configuration, so "hobby" would be affirmatively wrong as a product-level default here, not merely unsupported. Under EXTRACTION_SPEC 2.3 and CLAUDE.md prohibition 7 that is channel=UNKNOWN with an EMPTY basis, never channel=hobby with basis=unsupported (migrations/003_parallels.sql enforces (channel = 'UNKNOWN') = (channel_basis IS NULL), so the latter is rejected outright). RESOLVING THIS needs a per-tier distribution statement: a Panini sell sheet, the blog.paniniamerica.net first-look for this product, or retailer copy naming a tier and a channel together. https://www.cardboardconnection.com/2017-18-panini-revolution-basketball-nba-cards
• G-002 SET SIZE CONTESTED ON THE TWO AUTOGRAPH SETS, AND THE CONTEST IS NOT RESOLVED. sets.csv records Autographs=28 and Rookie Autographs=20. Evidence for Autographs=28: https://www.tcdb.com/Checklist.cfm/sid/165697/2017-18-Panini-Revolution-Autographs states "Total Cards: 28" and renders 28 rows, and https://www.tcdb.com/Checklist.cfm/sid/165698/2017-18-Panini-Revolution-Autographs-Cubic states 28 and renders a byte-identical 28-code list (A-AD ... A-ZL). Evidence for Autographs=29: https://www.tcdb.com/Checklist.cfm/sid/165699/2017-18-Panini-Revolution-Autographs-Kaleido states "Total Cards: 29" and renders 29 codes - the same 28 plus "A-AM" - and cardboardconnection states 29 (https://www.cardboardconnection.com/2017-18-panini-revolution-basketball-nba-cards). Evidence for Rookie Autographs=20: tcdb renders 20 on all three of its pages (sids 165694, 165695, 165696). cardboardconnection states 19. WHY 28/20 WAS WRITTEN RATHER THAN 29/19: cardboardconnection's two figures sum to 48 and tcdb's also sum to 48, so cardboardconnection's 29 is equally consistent with it having filed one rookie signer under Autographs as with the Autographs base page being one card short - and its 19 for Rookie Autographs is contradicted by three independent tcdb pages. The CHASE_CHECKLIST_SPEC 1c-ter tiebreaker (an independent third party siding with the parallel page proves the base page short) therefore does NOT fire cleanly here, because the third party is wrong on the neighbouring set. RESOLVING THIS needs the identity of "A-AM": a marketplace seller title carrying that code, or a second independent roster of the Autographs set. Whoever resolves it should also check whether A-AM is a Kaleido-only card.
• G-003 base_print_run=unnumbered is recorded with basis="assumed", not "sourced", and this is a known expressiveness loss rather than missing research. The evidence is strong and indirect: no row of the 150-card parent renders a serial token on any of three differently-varied fetches, while four sibling sids in the same block render SN100/SN75/SN50/SN10 on all 150 of their own rows through the identical request shape - a within-block positive discriminative test. EXTRACTION_SPEC 2.1 reserves "sourced" for a source that says so in words you can quote, and no source found states in words that the 2017-18 Revolution base cards carry no serial numbering. RESOLVING THIS needs a sentence, not another checklist: a Panini sell sheet or first-look article stating the base cards are unnumbered.
• G-004 NAME AND TIER-COUNT QUESTION ON THE CHINESE NEW YEAR BASE PARALLEL, NOT RESOLVED. parallels.csv records one base-set tier named "Chinese New Year", which is the name tcdb prints for sid 165665 and the name cardboardconnection uses in its own PARALLEL CARDS line. But cardboardconnection's PROSE calls the tier "Chinese New Year Red" - quoted verbatim: "Issued in a special format, the Chinese New Year Red base parallels are exclusive to international retail boxes." - and the adjacent 2018-19 release carries THREE Chinese New Year tiers (Red, Emerald #/88, Holo Gold #/8) where this season's tcdb listing carries one sid and cardboardconnection's list carries one entry. Two readings are open and nothing found settles them: (a) 2017-18 genuinely shipped a single Chinese New Year tier which cardboardconnection's prose calls Red, or (b) further tiers exist and neither source catalogues them. The channel value on this row rests on the quote above, so if reading (b) is right the channel is attributed to the correct tier but the release is under-covered by however many tiers are missing. RESOLVING THIS needs an international-retail box listing or Panini copy enumerating the Chinese New Year tiers for 2017-18. https://www.cardboardconnection.com/2017-18-panini-revolution-basketball-nba-cards
• G-005 MISSING RELEASE BELOW THIS LINE'S FLOOR IN data/: this build found that 2017-18 is NOT the first Revolution basketball season. tcdb's 2016-17 season index renders "2016-17 Panini Revolution" at sid 140756 (https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2016). The repository currently holds revolution-bkb-2018-19 through revolution-bkb-2024-25 and, with this build, 2017-18; there is no revolution-bkb-2016-17. This is the same class of finding the 2017-18 scope memo records for chronicles-bkb-2017-18, one season further back, and the memo does not mention it: its section 4 predecessor-chain table does not cover Revolution at all, and its section 2a files revolution under "maps cleanly" without a floor claim. This gap asserts only that a 2016-17 Revolution product exists on tcdb's own season index; it does not assert that the line has no seasons below 2016-17 either. RESOLVING THIS is a scoping decision for a human, not a research task.
• G-006 SOURCE-RELIABILITY FINDING ON tcdb ViewAll.cfm, recorded because EXTRACTION_SPEC 3.1 recommends that endpoint as the first fetch when opening a new season and this build was nearly misled by it. A first fetch of https://www.tcdb.com/ViewAll.cfm/sp/Basketball/year/2017 asked for the sid of the Revolution entry and returned 165179. That sid is NOT Revolution: https://www.tcdb.com/ViewSet.cfm/sid/165179/zzz-probe-slug renders the heading "2017-18 Donruss Optic" with Total Cards 200, and the 2017-18 scope memo section 5a independently uses 165179 as the Donruss Optic parent. A second fetch of the same index, varied with ?sort=0 and asked to quote the href STRING rather than name the sid, returned /ViewSet.cfm/sid/162384/2017-18-Panini-Revolution together with /ViewSet.cfm/sid/165179/2017-18-Donruss-Optic - i.e. it reproduced the correct mapping for both, and the first response had attached a neighbouring product's sid to the right product name. 162384 was then confirmed against its own page heading ("2017-18 Panini Revolution", Total Cards 150). CONSEQUENCE FOR THE SPEC: a sid taken from ViewAll.cfm must be confirmed against its own ViewSet.cfm heading before any enumeration is built on it, and asking for the href string is measurably safer than asking for the sid. RESOLVING THIS is a spec amendment, not a fetch.
• G-007 METHOD DIVERGENCE AGAINST AN EXISTING CATALOGUE RELEASE, RECORDED SO IT CAN BE RE-RUN RATHER THAN ARGUED. Every one of this release's 29 parallel rows was written from a DISTINCT-TOKEN ENUMERATION over every row of its tier (150 rows each for the ten base-set tiers, via Checklist.cfm pages 1 and 2 plus Gallery.cfm top-ups, and the full roster for each insert-scope tier). The adjacent revolution-bkb-2018-19 release wrote its tier print runs from a SAMPLE: its Cosmic row's own visual_notes read "every sampled row of this sid (101, 110, 120, 128, 135, 140, 145, 150) renders the literal token SN100". This build makes no claim that any 2018-19 value is wrong - it did not fetch that season - but EXTRACTION_SPEC 3.2 records that a three-row sample was wrong on 22 percent of one release's tiers, and the 2017-18 scope memo section 7.5 measures variable-run rates of 17 to 75 percent on this season. A sampled tier is an unverified tier. RESOLVING THIS means re-enumerating the distinct token set of each revolution-bkb-2018-19 parallel tier. It is NOT a gap in this release's data.
• G-008 NO PACK ODDS SOURCED FOR ANY SET OR TIER IN THIS RELEASE. Every sets.csv row has stated_odds, odds_evidence_url and odds_evidence_quote empty, and so does every parallels.csv row. cardboardconnection publishes box configuration for this product but no per-set or per-tier insertion odds, and tcdb carries none. No odds statement was found on any source fetched. This is an evidenced absence of odds DATA, not an evidenced absence of odds: Panini printed odds on the pack wrappers and sell sheets, and nothing found reproduces them. RESOLVING THIS needs a pack wrapper image, a sell sheet, or a first-look article that reproduces the odds table.
• G-009 INSERTS.CFM NOTES COLUMN WAS UNUSABLE ON THIS PARENT AND WAS NOT USED - recorded because CAV-036 makes this the season's highest-cost silent failure. The Notes column on https://www.tcdb.com/Inserts.cfm/sid/162384/2017-18-Panini-Revolution is populated on exactly TWO of 36 rows: "Rookie Autographs Cubic" reads "Serial numbered to 50" and "Rookie Autographs Kaleido" reads "Serial numbered 1/1". The other 34 cells are empty, including the cells of Cosmic, Sunburst, Cubic and Lava, which this build then proved carry SN100, SN75, SN50 and SN10 on all 150 of their rows. So the blank-annotation test CANNOT be run on this parent and was not run: a blank cell here is the Obsidian shape and carries ZERO information. Writing unnumbered off those blanks would have been FALSE on four base-set tiers and on seven insert-scope tiers, not merely unsupported. FORMAT, for the record CAV-036 asks for: this parent renders PROSE ("Serial numbered to 50"), like the 2017-18 Hoops parent, not Select's bare-token shape. The format trap did not fire here; the mostly-blank trap would have. Every one of the 29 tiers was instead probed at row level on at least one endpoint, and every tier written as unnumbered was probed on two.

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