2017-18 Prizm

• G-001 PER-CARD VARIABLE PRINT RUN, INEXPRESSIBLE IN THE SCHEMA. tcdb sid 163563 renders 42 of its 142 rows and the rows disagree: flag cell "AU, SN10" on 40 rows and "AU, SN5" on 2. The tier is PROVABLY serial numbered, but print_run and serial_numbered are forced to UNKNOWN because EXTRACTION_SPEC 2.3 forbids the modal token and the schema cannot hold "serial numbered, run varies". tcdb's own Inserts.cfm Notes cell for this tier reads "AU, SN10 or less" - a ceiling, which is not a print run. Held fact: the run is 10 or less, with at least two cards at 5. https://www.tcdb.com/ViewSet.cfm/sid/163563/2017-18-Panini-Prizm and https://www.tcdb.com/Inserts.cfm/sid/157371/2017-18-Panini-Prizm
• G-002 PER-CARD VARIABLE PRINT RUN, INEXPRESSIBLE IN THE SCHEMA. 65 of the 90 rows enumerated across five differently-parameterised fetches; 64 read "MEM, SN10" and one - SW-SP Scottie Pippen, on https://www.tcdb.com/Gallery.cfm/sid/163594/2017-18-Panini-Prizm-Sensational-Swatches-Starburst-Prizms?PageIndex=8 - reads "MEM, SN5". tcdb's Inserts.cfm Notes cell independently says the same thing. print_run/serial_numbered written UNKNOWN; the modal token 10 was NOT written. This tier is the single reason the whole-tier enumeration discipline was worth its cost on this release: a three-row sample would have written a confident, sourced-looking 10. https://www.tcdb.com/ViewSet.cfm/sid/163594/2017-18-Panini-Prizm
• G-003 SINGLE-SOURCE SET. tcdb sid 163552 is the only source that carries this set. cardboardconnection (https://www.cardboardconnection.com/2017-18-panini-prizm-basketball-nba-cards) names no "Autographs" set at all, though it does state "each 12-pack hobby box contains two autographs" and "1 Autograph" per Fast Break box. The count 196 was not taken from the header alone: https://www.tcdb.com/Gallery.cfm/sid/163552/2017-18-Panini-Prizm-Autographs?PageIndex=20 renders exactly 6 tiles and no page 21 exists, so 19 pages of 10 plus 6 = 196. The set is base-MIRRORED - it reuses base card numbers (1,2,3,5,6,7,8,11 ... 299) with holes where a player did not sign - so its card numbers collide with base_cards.csv by design. No independent corroboration of the roster or the count was found.
• G-004 SERIAL-TOKEN ENUMERATION IS PARTIAL, NOT COMPLETE, ON EVERY BASE TIER. Each of the 25 base parallels has a 300-card roster. Each was read through three separately-parameterised Checklist.cfm windows (PageIndex 1, 2 and 3), which delivered between 186 and 231 rows per tier - roughly 62-77 percent - because the fetch layer truncates the tail of every 100-row page. Within every window read, each tier showed exactly ONE distinct serial token (or none at all), and those tokens agree with tcdb's Inserts.cfm Notes cells and with cardboardconnection's stated print runs on 17 of 17 tiers where cardboardconnection states one. Completing all 25 would have cost roughly 175 further Gallery.cfm fetches. The residual risk is real and is not hypothetical: two INSERT-scope tiers in this same release do carry per-card variable runs (G-001, G-002), though both were announced by their own Notes cell and no base tier's Notes cell carries such language. https://www.tcdb.com/Inserts.cfm/sid/157371/2017-18-Panini-Prizm
• G-005 ANOMALOUS FLAG. tcdb renders card 170 (Michael Malone, Denver Nuggets - a head coach) with the flag cell "CO, RC" on two different endpoints: https://www.tcdb.com/Checklist.cfm/sid/157371/2017-18-Panini-Prizm?PageIndex=2 (both under the default sort and under &sort=0) and https://www.tcdb.com/Gallery.cfm/sid/157371/2017-18-Panini-Prizm?PageIndex=17&sort=0. Every other coach card in the set (the tenth card of each of the 30 team blocks) carries CO alone. rookie_flag=yes is written because the source explicitly shows the RC designation, but a rookie designation on a coach is not a fact about the player and may be a tcdb data error. Not resolved.
• G-006 CROSS-SEASON NAMING DIVERGENCE, RECORDED NOT RESOLVED. tcdb prints this season's base tiers in the PLURAL - "Silver Prizms", "Blue Prizms", "Mojo Prizms" - and those are the names written here. The sibling release data/prizm-bkb-2018-19 carries the SINGULAR form ("Silver Prizm", "Blue Prizm", "Mojo Prizm") for the same visual tiers, and cardboardconnection prints a third form with no noun at all ("Silver", "Blue #/199"). A resolver keyed on the exact string will not match across the two seasons. Separately, insert-scope tiers are written here with the set-name prefix stripped (tcdb prints "Emergent Black Prizms"; this file carries parallel_name "Black Prizms" with scope "Emergent"), following the catalogue convention that prizm-bkb-2018-19 uses - but 2018-19 also strips the noun, writing "Black". https://www.tcdb.com/Inserts.cfm/sid/157371/2017-18-Panini-Prizm
• G-007 NAME DISAGREEMENT BETWEEN SOURCES. tcdb names the /25 base tier "Red Pulsar Prizms" (page heading of sid 163546, read back verbatim on https://www.tcdb.com/Checklist.cfm/sid/163546/2017-18-Panini-Prizm-Red-Pulsar-Prizms?PageIndex=1 ) while cardboardconnection lists it as "Red #/25" in a list that spells its two sibling Pulsar tiers out in full ("Pink Pulsar #/42", "Green Pulsar #/25"), so the omission is not obviously abbreviation. The print run agrees at 25 on both. The tcdb form is written. Unresolved which name the card itself carries.
• G-008 NO CHANNEL SOURCE FOR MOST BASE TIERS, AND THIS IS A REGRESSION AGAINST THE SIBLING SEASON. For 2018-19, cardboardconnection annotates base parallels with their channel ("Green (Retail)", "Pink Ice (Walmart Mega)"), and prizm-bkb-2018-19 carries channel values with channel_basis=sourced-presence on six rows. The 2017-18 cardboardconnection page carries no such annotations - its parallel list is "Silver, Green, Hyper, Red/White/Blue, Ruby Wave, White Sparkle (online promo packs), Blue #/199, ..." with channel information only for White Sparkle and Red Shimmer (FOTL). Channel is therefore UNKNOWN with an EMPTY channel_basis on Silver, Green, Hyper, Red White and Blue, Ruby Wave, Blue, Blue Ice, Purple, Orange, Pink Pulsar, Green Pulsar, Red Pulsar, Mojo, Gold and Black. Nothing was inherited from 2018-19. https://www.cardboardconnection.com/2017-18-panini-prizm-basketball-nba-cards
• G-009 PARALLEL_FAMILY THE ONE REPO DEFINITION DOES NOT CARRY. "Starburst" is a distinct visual treatment and is not in the ordered FAMILIES list in pipeline/fix_schema.py, so family_of("Starburst Prizms") falls through to the "Prizm" pattern and returns "Prizm (plain)" - which is what these two rows carry. The value was NOT invented and fix_schema.py was NOT edited. Raised for a maintainer to decide whether a Starburst family should be minted. https://www.tcdb.com/Inserts.cfm/sid/157371/2017-18-Panini-Prizm
• G-010 SAME-SEASON PREFIX COLLISION, CONFIRMED ON BOTH SIDES OF THIS RELEASE. Prizm's Sensational Signatures uses the SS- prefix (https://www.tcdb.com/ViewSet.cfm/sid/163591/2017-18-Panini-Prizm - card numbers SS-DF, SS-JI, SS-ZC). 2017-18 Panini Select uses SS- for Select Swatches. Any resolver keyed on card_number alone will cross-hit; the key is (set_name, card_number). Recorded for the later chase-checklist pass, which will write these codes.
• G-011 NO VARIATION SET ESTABLISHED, AND ITS ABSENCE IS NOT EVIDENCED. https://www.tcdb.com/Errors.cfm/sid/157371/2017-18-Panini-Prizm is explicitly headed "Errors / Variations" and returns "No errors or variations have been added for this set yet." Per caveats.csv CAV-020 that silence is worthless - the endpoint returns that identical sentence on sets that provably carry UER rows. What does survive is independent: the base run was enumerated contiguously 1-300 with no b-suffixed row seen on Checklist.cfm (which does render suffixes) and no number rendered twice. So: no variation set found, absence NOT established.
• G-012 PARALLEL ROSTER SIZES DIVERGE FROM THE PARENT SET AND FROM EACH OTHER. The Autographs set has 196 cards, but its three Fast Break tiers have "Total Cards" of 136 (Fast Break Prizms, sid 163561), 146 (Fast Break Black Prizms, sid 163562) and 142 (Fast Break Gold Prizms, sid 163563). Three different rosters under one parent. No source explains the differences and they were not reconciled; they are recorded here because a consumer comparing a tier roster to its parent set count will otherwise read the gap as missing coverage. https://www.tcdb.com/Inserts.cfm/sid/157371/2017-18-Panini-Prizm
• G-013 ONE OF THE TWO USUAL EDITORIAL LINEAGES WAS NEVER READ. No checklistcenter.com page for 2017-18 Panini Prizm was surfaced by a site-scoped search. Per caveats.csv CAV-037 no URL was constructed and no absence is claimed - the page may exist under an irregular slug. Consequence: every editorial fact in this release rests on cardboardconnection alone, with tcdb as the independent structural source and one marketplace listing as a third lineage on exactly one print run (see G-014). beckett.com was not attempted; the 2017-18 scope memo records it as unreachable for this season.
• G-014 MARKETPLACE CORROBORATION ATTEMPTED ON TWO TIERS; ONE CLEAN, ONE REJECTED FOR CONTAMINATION. Clean: https://www.ebay.com/itm/206240837113 carries the free-text seller title "Donovan Mitchell 2017-18 Panini Prizm Fast Break Blue Rookie RC 025/175 BGS 9.5" - a serial off a card in hand, corroborating Fast Break Blue Prizms at 175 independently of tcdb and cardboardconnection. The fetch layer reported the string "(RC" as present on that page while quoting a title containing only a bare "RC"; that is the documented false positive of the contamination-fingerprint test and the listing was NOT discarded. Rejected: https://www.ebay.com/itm/274598858316 was reached from a search snippet reading "2017-18 Panini Prizm DONOVAN MITCHELL #87/199 Blue Prizm Rookie BGS 9 Mint!" and the page itself renders "2017-18 Panini Prizm - Donovan Mitchell #117 Blue Prizm /199 (RC)" - the eBay-catalogue retitling shape, with a parenthesised tcdb flag string and a different card number from the snippet. Not used. No other tier has marketplace corroboration.
• G-015 BASE PRINT RUN IS ASSUMED, NOT SOURCED. base_print_run=unnumbered with base_print_run_basis=assumed. The evidence is strong but indirect: tcdb renders a serial token on every row of 18 of the 25 base parallel sids and renders none on any of the 300 base rows across ten separate fetches. No source states in words that the base cards carry no serial number. Per EXTRACTION_SPEC 2.1 that is "assumed", not "sourced". https://www.tcdb.com/Checklist.cfm/sid/157371/2017-18-Panini-Prizm
• G-016 SET FILED AS insert BECAUSE THE ENUM HAS NO RELIC VALUE. tcdb sid 163593 renders 67 of its 90 rows and every one is flagged MEM with none flagged AU, so this is an unsigned memorabilia (relic) set. EXTRACTION_SPEC 2.2 states there is no relic value in set_type and that an unsigned memorabilia set is an insert; that is what is written, with the relic fact carried in the set notes. Recorded so a consumer filtering set_type=insert knows a relic set is inside it. 23 of the 90 rows were never rendered, so the "none flagged AU" observation covers 67 of 90. https://www.tcdb.com/ViewSet.cfm/sid/163593/2017-18-Panini-Prizm
• G-017 STRUCTURE REVERSES AGAINST THE ADJACENT SEASON AND THE DIVERGENCE IS RECORDED RATHER THAN SMOOTHED. Against data/prizm-bkb-2018-19: base_set_size agrees at 300 and the exclusion convention agrees (no block past the base run in either). Everything else differs. 2017-18 has 25 base parallels against 2018-19's 37, and the ladders are not nested - 2017-18 has no plain Red /299, no Purple Wave, no Pink Ice, no Choice family at all, no Purple Pulsar, and no Black Gold, while it DOES have a Red Pulsar /25 and a Ruby Wave that 2018-19 files differently. Insert sets: 9 here against 16 in 2018-19, and only Emergent, Get Hyped! and Luck of the Lottery appear in both. The 2017-18 base run also includes 30 head coach cards (every tenth number), which the 2018-19 base run does not. Nothing was carried across the season boundary.

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