2017-18 Threads

• G-001 RELEASE DATE UNRESOLVED TO A DAY. Eight sources tried, none states one. cardboardconnection renders only "**Release Date**: May/June 2018" (https://www.cardboardconnection.com/2017-18-panini-threads-basketball-nba-cards) - a month range, not a date, and not an "Estimated Release Date" label either, so it is imprecise rather than self-declared as a forecast. tcdb renders no Release Date line for parent sid 176608, confirmed on two differently-slugged fetches (https://www.tcdb.com/ViewSet.cfm/sid/176608/2017-18-Panini-Threads and https://www.tcdb.com/ViewSet.cfm/sid/176608/zzz-threads-probe). Panini America's own blog post, first-party and off the cardboardconnection lineage, is dated "May 25, 2018" and says "2017-18 Threads Basketball is starting to arrive on store shelves right about now (primarily Target locations nationwide)" (https://blog.paniniamerica.net/panini-americas-2017-18-threads-basketball-delivers-games-biggest-memorabilia-cards/) - it corroborates the May/June window independently and still names no day. beckett.com 302s to https://maintenance.beckett.com/ . gogts.net, which supplied an off-lineage date for 2017-18 Encased, carries no 2017-18 Threads page. No checklistcenter page surfaced through search; per CAV-037 that is NOT an absence claim, only a failure to find. steelcitycollectibles.com shows no date. release_date is UNKNOWN. This is a REGRESSION against data/threads-bkb-2018-19, which carries 2019-04-24 from two sources, and a reader comparing the line will see a hole where there was data.
• G-002 A CANDIDATE RELEASE DATE WAS FOUND AND REFUTED, and it is recorded so a later pass does not re-find it and write it. dacardworld.com prints "Release Date: January 1st, 2017" on its product page for this box (https://www.dacardworld.com/sports-cards/2017-18-panini-threads-basketball-5x7-jumbo-jersey-box). That is temporally impossible: the checklist's 40 rookies (cards 61-100, including Lonzo Ball, Jayson Tatum, Donovan Mitchell and De'Aaron Fox) were drafted in June 2017, and Panini's own post has the product arriving in stores in late May 2018 (https://blog.paniniamerica.net/panini-americas-2017-18-threads-basketball-delivers-games-biggest-memorabilia-cards/). Treated as a retailer placeholder, not a date.
• G-003 Errors.cfm on the parent returns "No errors or variations have been added for this set yet." (https://www.tcdb.com/Errors.cfm/sid/176608/2017-18-Panini-Threads). Recorded as a reproduction of the CAV-020 false negative and used to conclude NOTHING. What the base run rests on instead is an independent contiguous enumeration of all 100 numbers across one Checklist.cfm and four Gallery.cfm pages, with the boundary proved by Gallery ?PageIndex=11&sort=0 rendering no tiles. No claim is made here that this release has no variation set; in fact it demonstrably has one, on the Box Topper Memorabilia Rookies sid, which Errors.cfm on the parent does not mention.
• G-004 IMAGE VARIATION on Box Topper Memorabilia (the veteran set) IS SINGLE-SOURCED AND CONTRADICTED BY tcdb. sportscardchecklist lists "2017-18 Panini Threads Box Topper Memorabilia Image Variations (25 Cards)" (https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-threads/trading-card-checklists-and-product-information) and its set page renders exactly the 25-name Box Topper Memorabilia roster (https://www.sportscardchecklist.com/set-146202/2017-18-panini-threads-box-topper-memorabilia-image-variations-basketball-trading-card-checklist), which is a roster match rather than a name match. cardboardconnection carries an unscoped "Image Variation" entry with no print run (https://www.cardboardconnection.com/2017-18-panini-threads-basketball-nba-cards). But tcdb sid 176610 renders 25 rows with NO "VAR" token on any of them, on two differently-parameterised fetches. The row is written with source_confidence=single. Needs a third lineage or a marketplace title showing a veteran BTM variation card in hand. Note that sportscardchecklist is the source CAV-040 has just proved capable of mixing two products' inventories in one index, so its uncorroborated entries on this line deserve exactly this much weight and no more.
• G-005 PRINT RUN REFUSED ON BOTH IMAGE VARIATION ROWS - the EXTRACTION_SPEC 2.3 expressiveness gap, twice. For Box Topper Memorabilia Rookies the tier IS rendered (50 rows over 25 numbers on https://www.tcdb.com/Checklist.cfm/sid/176612/2017-18-Panini-Threads-Box-Topper-Memorabilia-Rookies?sort=0, the extra 25 carrying the literal token "VAR") and no row carries any token beginning with "SN" on either of two fetches - but the non-variation rows of the SAME sid are equally tokenless, so there is no within-sid discriminative test and unnumbered would be unsupported rather than merely cautious. For the veteran Box Topper Memorabilia tcdb does not carry the tier at all. No source states a run for either. Both rows are UNKNOWN/UNKNOWN, which is legal here only because scope_type=set; a base-set-scope row in this state would have had to be removed.
• G-006 A CLASSIFICATION DIVERGENCE FROM tcdb, RECORDED RATHER THAN SMOOTHED. tcdb files "Box Topper Memorabilia Prime" (sid 176611) and "Box Topper Memorabilia Rookies Prime" (sid 176613) under the "Insert Sets" heading of https://www.tcdb.com/Inserts.cfm/sid/176608/2017-18-Panini-Threads , as siblings of their base sets rather than as parallels. They are recorded here as parallels.csv rows scoped to their base sets, because their card codes are the same 25 codes as those sets (BTM-AWG ... BTM-ZLV byte-identical; BTR-CSW ... BTR-ZCL the same 25 players without the a/b suffixes) and a parallel's checklist belongs in parallels.csv. cardboardconnection agrees with this reading, listing them under parallels as "Prime 1/1" and "Prime #/5" (https://www.cardboardconnection.com/2017-18-panini-threads-basketball-nba-cards); sportscardchecklist agrees with tcdb, listing them as sets (https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-threads/trading-card-checklists-and-product-information). The consequence is that a consumer counting this release's sets gets 3 here and 5 from tcdb, and the difference is a modelling choice, not missing data. The adjacent season data/threads-bkb-2018-19 made the same choice, filing Prime as parallel rows.
• G-007 TOTAL CARDS AND ROW COUNT DISAGREE ON Box Topper Memorabilia Rookies AND THE HEADER IS RIGHT - a unit question, resolved, recorded because the next pass will meet it again. tcdb states "Total Cards: 25" while https://www.tcdb.com/Checklist.cfm/sid/176612/2017-18-Panini-Threads-Box-Topper-Memorabilia-Rookies?sort=0 renders 50 rows. The header counts distinct ISSUED numbers and the extra 25 rows are an image-variation layer duplicating them. Reconciled three ways: the a/b suffix pairing visible in the codes themselves, the "VAR" token on exactly the duplicate rows, and the Prime tier (sid 176613) rendering exactly 25 unsuffixed codes for the same 25 players. card_count=25 is written.
• G-008 THE CAV-040 MIRROR CHECK WAS RUN FROM THIS SIDE AND CAME BACK CLEAN, WITH ONE REFINEMENT TO THE CAVEAT'S OWN DESCRIPTION. CAV-040 predicts that a name-keyed search on this line would pull 2017-18 Encased pages, because sportscardchecklist's Encased index is contaminated with eighteen Threads sets. No Encased page was returned by any search run for this build, and sportscardchecklist's own Threads index (https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-threads/trading-card-checklists-and-product-information) lists 12 entries, all 12 of them genuinely Threads. Reconciled against the rendered tcdb enumeration: tcdb carries 10 sids for this product (parent 176608, inserts 176610-176613, parallels 177656-177660); sportscardchecklist carries those same 10 plus two Image Variation entries tcdb does not file separately, giving 12. The refinement: CAV-040 describes the contamination as "six Box Topper Memorabilia (25 cards each), six Threads (60) and six Rookies Threads (40), with Dazzle/Black/Blue/Gold/Red tiers", which reads as though the Box Toppers carry five Dazzle colour tiers. They do not. On the Threads side the six Box Topper entries are Memorabilia, Memorabilia Image Variations, Memorabilia Prime, Memorabilia Rookies, Memorabilia Rookies Image Variation and Memorabilia Rookies Prime, and it is the 100-card BASE set that carries the five Dazzle colour tiers - which is also why the Encased index could split it into a 60-card "Threads" and a 40-card "Rookies Threads": this base run really is 60 veterans then 40 rookies, confirmed by the "RC" token rendering on cards 61-100 and on no card 1-60.
• G-009 parallel_family ON ALL 9 ROWS IS "other", DERIVED BY AST-LIFTING family_of() FROM pipeline/fix_schema.py, AND IT DISAGREES WITH THE ADJACENT SEASON BY DESIGN. data/threads-bkb-2018-19 carries 93 hand-written values of "Dazzle" and "Premium" that family_of() does not produce - the largest single block of the 268 rows CAV-039 describes. family_of("Dazzle"), family_of("Dazzle Red"), family_of("Dazzle Blue"), family_of("Dazzle Gold"), family_of("Dazzle Black"), family_of("Prime") and family_of("Image Variation") all return "other", because FAMILIES contains no Dazzle, Prime or Image Variation pattern. Nothing was added to fix_schema.py and the sibling was not touched. The consequence is that a consumer filtering parallel_family="Dazzle" gets the 2018-19 rows and not the 2017-18 ones, for the same visual treatment under the same product name. This is CAV-039's decision, not this build's, and it should be settled corpus-wide before either side is changed. One negative result worth recording: no short FAMILIES pattern swallowed any of these names, so the "Ink" inside "Pink" failure mode did not fire here.
• G-010 card_tier IS EMPTY ON ALL 100 BASE ROWS, and a sourced sub-division of the run exists that was deliberately not written into it. The base run is 60 veterans (1-60) then 40 rookies (61-100): tcdb renders the literal token "RC" on 61-100 and on no row 1-60; dacardworld states "The first 60 cards feature NBA veterans and the last 40 cards are for the rookie players." (https://www.dacardworld.com/sports-cards/2017-18-panini-threads-basketball-5x7-jumbo-jersey-box); and per CAV-040 sportscardchecklist's 2017-18 Encased index splits this very set into a 60-card "Threads" and a 40-card "Rookies Threads". It is not written as card_tier because tcdb renders no sub-section heading and the split carries no print-run or scarcity difference - all 100 cards are unnumbered and all five Dazzle tiers run the full 100 - so no base_card_tiers.csv is written either and base_print_run is unnumbered rather than varies-by-tier. rookie_flag already carries the distinction losslessly. Recorded in case a later pass wants the block names.
• G-011 NO channel_sku IS WRITTEN ANYWHERE IN THIS RELEASE, although the product has exactly one box configuration. The two retailers reached title it differently: dacardworld "2017/18 Panini Threads Basketball 5x7 Jumbo Jersey Blaster Box" (https://www.dacardworld.com/sports-cards/2017-18-panini-threads-basketball-5x7-jumbo-jersey-box) and steelcitycollectibles "2017-18 Panini Threads Basketball 5x7 Jumbo Jersey Box" (https://www.steelcitycollectibles.com/i/2017-18-panini-threads-basketball-5x7-jumbo-jersey-box). cardboardconnection gives only "**Product Configuration:** 4 cards per pack" (https://www.cardboardconnection.com/2017-18-panini-threads-basketball-nba-cards) and steelcitycollectibles "3 cards + (1) 5x7 box topper memorabilia card per box." - consistent with each other and naming no SKU. Panini America calls it only "a unique packaging configuration" (https://blog.paniniamerica.net/panini-americas-2017-18-threads-basketball-delivers-games-biggest-memorabilia-cards/). Writing "Blaster Box" would rest on a single retailer's product title, and blaster is not a kind of retail in this schema, so the channel stays retail and the SKU column stays empty.
• G-012 MARKETPLACE CORROBORATION ATTEMPTED AND REJECTED FOR CONTAMINATION, recorded so the rejection is visible rather than silent. The eBay listing at https://www.ebay.com/itm/134542514879 would have independently corroborated the Box Topper Memorabilia Rookies image-variation tier. Its title reads, verbatim and in full, "2017-18 Panini Threads - Box Topper Rookie Memorabilia Jayson Tatum #BTR-JTT (MEM, RC)". The tcdb flag string "(MEM, RC)" appears verbatim in the title and again in the page meta-description, which is the EXTRACTION_SPEC 3.4 fingerprint; a seller does not write those. Rejected as circular. The tier still rests on tcdb plus sportscardchecklist.
• G-013 THE Inserts.cfm NOTES COLUMN WAS UNUSABLE ON THIS PARENT AND NO PRINT RUN CAME FROM IT. https://www.tcdb.com/Inserts.cfm/sid/176608/2017-18-Panini-Threads renders 9 rows and every single cell is an image-completion percentage - "36.0% of images added", "0.0% of images added", "4.0% of images added", "61.0% of images added", "7.0% of images added", "1.0% of images added", "8.0% of images added" - with zero serial statements in any format and zero blank cells, so the both-directions split test cannot be run at all. This is the CAV-036 worthless shape, now measured on a twelfth 2017-18 parent, and it fired in BOTH sections of the page (Insert Sets and Parallel Sets alike). Every print run in this release came from a per-sid row-level token sweep. The page did NOT truncate and did NOT interpolate: all 9 sids were confirmed against their own page headings, 9 of 9.
• G-014 COMPLETENESS BOUND, WITH ITS LIMIT NAMED. Both sid blocks were bounded in both directions by probes that returned foreign products or the tcdb home page: for the insert block 176610-176613, sid 176609 renders "1977 Sportscaster Series 3 Finnish", sid 176614 renders the tcdb HOME PAGE ("Total Cards: 35,403,916", the documented non-existent-sid signature) and sid 176615 renders "2016 Breygent Dexter Season 7 & 8"; for the parallel block 177656-177660, sid 177655 renders "2018 Panini Classics - Vintage Logo Stickers" and sid 177661 renders "2015 Topps Heritage - Flip Stock". Zero interleaving in either block. The parent sid sits 1,048 sids below its parallel block, so the scope memo 7.1 finding that the parent is nowhere near its children reproduces on a fourth parent. WHAT THIS DOES NOT PROVE: that tcdb has everything. sportscardchecklist carries two Image Variation tiers tcdb files no sid for, so a balanced sid count would have bounded the search without proving the composition.
• G-015 STRUCTURAL COMPARISON AGAINST THE ADJACENT SEASON, run per EXTRACTION_SPEC 6. data/threads-bkb-2018-19 sets base_set_size=100 against a parent of 220 and excludes a 120-card jersey-design SP block under 5.3. This release also sets base_set_size=100, but its parent header and its rendered enumeration both stop at 100, so there is no block to exclude and no convention conflict. The two seasons agree on base size and on base_print_run=unnumbered with basis=assumed. They DIFFER on structure in ways worth naming: 2018-19 has 22 sets.csv rows and 97 parallel rows across a Dazzle and a Premium family with a panini-direct Premium Edition box; 2017-18 has 3 sets and 9 parallels, no Premium Edition, no autograph set at all, and one box topper per box as the entire hit structure. The 2018-19 channel model also does not transfer: that season is split across three retail formats plus an online-only Premium Edition and its rows carry channel=blaster, channel=panini-direct and channel=UNKNOWN, while every row here is channel=retail on two independent exclusivity statements. Nothing was inherited.

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