2017-18 Donruss Optic Premium Box Set

• G-001 BASE-CARD IDENTITY COLLISION between TWO releases, per EXTRACTION_SPEC 5.2, and this is how all seven built opticpremium seasons are modelled. This release (opticpremium-bkb-2017-18) and data/optic-bkb-2017-18 describe THE SAME 200 physical base card slots, so the catalogue now carries 200 duplicate base cards for this season on top of the duplicates the later seasons already carry. MEASURED, NOT ESTIMATED: all 200 rows were read independently off this product's own tcdb sid 171625 (see G-017 for the exact eleven fetches) and then diffed positionally in code against data/optic-bkb-2017-18/base_cards.csv. THE OVERLAP IS 200 OF 200 ON card_number, 200 OF 200 ON team_or_school, and 197 OF 200 ON player_name - see G-002 for the three exceptions, none of which is a different player. Nothing was deduplicated and nothing was double-counted; the full base run was built here rather than referenced out to the other directory. The EXTRACTION_SPEC 4.2 copy check therefore fires, legitimately: this IS a repackaging of the parent base run on premium stock with serial numbering added, which is what every later season of this line is - opticpremium-bkb-2018-19 G-001 200/200, -2019-20 G-001 200/200, -2020-21 G-001 200/200, -2021-22 G-001 200/200, -2022-23 G-001 250/250, -2023-24 G-001 250/250, -2024-25 G-001 300/300. THIS IS NOW THE EARLIEST SEASON OF THE LINE IN THE CATALOGUE, one below the floor opticpremium-bkb-2018-19 G-012 recorded as the next thing to build. A complete overlap does NOT by itself settle product-versus-parallel; what settles it here is the exclusive 10-card autograph set, the separate sealed box, the separate sale channel (Panini: "The sets are available only at iCollectPanini.com while supplies last.", https://blog.paniniamerica.net/2017-18-donruss-optic-basketball-premium-box-set-arrives-on-icollectpanini-com/) and the separate release date 2018-04-23 against the parent's 2018-04-04.
• G-002 CONTRADICTION FOUND AGAINST EXISTING CATALOGUE DATA, AND IT IS A REAL tcdb DIVERGENCE RATHER THAN A TRANSCRIPTION ERROR. The task prompt asked whether this product's base checklist is identical to the parent's or differs anywhere. It differs in exactly three player-name STRINGS out of 200, and in no card number and no team. Under this product's sid 171625 the rows render "J.J. Redick" at 113, "C.J. McCollum" at 123 and "Wayne Selden Jr." at 153; data/optic-bkb-2017-18/base_cards.csv carries "JJ Redick", "CJ McCollum" and "Wayne Selden" at the same three numbers. Each of the three was confirmed on this sid on a second differently-parameterised fetch - https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=12&sort=0 ("J.J." with periods after each initial), https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=13&sort=0 ("C.J.") and https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=16&sort=0 ("Wayne Selden Jr." with the suffix included) - and each was then checked against the PARENT sid, whose own rows render "JJ Redick", "CJ McCollum" and "Wayne Selden" (https://www.tcdb.com/Checklist.cfm/sid/165179/2017-18-Donruss-Optic?PageIndex=2&sort=0). So tcdb spells three names differently on its two sids for the same physical cards, and neither build mis-transcribed its own page. Note also that this sid renders "TJ Warren" at 120 WITHOUT periods on the same Gallery page that renders "J.J. Redick" WITH them, which rules out the fetch layer uniformly inserting punctuation. WHAT WAS WRITTEN: this file carries what this product's own sid renders, per EXTRACTION_SPEC 2.4 (never normalize what the source printed). CONSUMER IMPACT: a resolver joining these two releases on (card_number, player_name) will miss three of 200 rows, and a resolver joining on card_number alone will not. No repair is proposed here for the other release; the divergence is recorded so it can be repaired deliberately rather than smoothed over.
• G-003 parallels.csv is deliberately EMPTY - header only, zero data rows - per EXTRACTION_SPEC 2.3, and the absence is EVIDENCED rather than unresolved. No parallel OF THE PREMIUM BOX SET base run exists on any source fetched. The complete subset inventory of the 2017-18 Donruss Optic parent was enumerated row by row in one untruncated fetch: "Parallel Sets (29)" and "Insert Sets (123)", 152 rows in all, every row's name and sid listed, and the two section headings' own counts agree exactly with the counts of the rows actually enumerated (29 and 123), which is the EXTRACTION_SPEC 3.2 guard against taking a fetch's own tally (https://www.tcdb.com/Inserts.cfm/sid/165179/2017-18-Donruss-Optic?sort=2; the same page under https://www.tcdb.com/Inserts.cfm/sid/165179/2017-18-Donruss-Optic gave the same two Premium rows and the same Notes cells). "Premium Box Set" (sid 171625) is itself one of the 29 Parallel Sets rows of the PARENT, and no set name anywhere on the page other than that row and "Rated Rookies Signatures Premium Box Set" (sid 171641) contains the word "Premium" - so the Premium Box Set has no children of its own. cardboardconnection lists no parallel of it either, and Panini's post describes one format, not a ladder. This matches all seven sibling seasons: opticpremium-bkb-2018-19 through -2022-23 also carry zero parallel rows; -2023-24 and -2024-25 carry three each. RESIDUAL RISK, stated rather than hidden: EXTRACTION_SPEC 3.2 records that Inserts.cfm can omit real sets and that the Optic parent in particular has truncated before the Parallel Sets heading in three consecutive seasons. It did not truncate here - both sections rendered in full in a single response - but the numeric gaps inside the sid listing were NOT probed for an omitted Premium child, because doing so would have meant sweeping the parent's sid neighbourhood, which belongs to the optic-bkb-2017-18 slice and not to this one.
• G-004 BRAND-IDENTITY WORD-ORDER DIVERGENCE, REPRODUCED AS PREDICTED, AND NOTHING WAS MINTED. specs/2017-18_SCOPE_MEMO.md 2b records that cardboardconnection writes this product with "Basketball" interposed. It does: its H1 is "2017-18 Donruss Optic Basketball Premium Box Set" (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set), while tcdb renders "2017-18 Donruss Optic - Premium Box Set" (https://www.tcdb.com/ViewSet.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set) and Panini America writes "Donruss Optic Premium Box Sets" and "2017-18 Donruss Optic Basketball Premium Box Set" in its own post title (https://blog.paniniamerica.net/2017-18-donruss-optic-basketball-premium-box-set-arrives-on-icollectpanini-com/) - so Panini itself uses BOTH orders, which the memo did not know. WHAT WAS WRITTEN AND WHY: release.csv product_name is "Donruss Optic Premium Box Set Basketball", byte-identical to reference/brands.csv product_name for line_prefix "opticpremium" (row brd_01M2B8M6BJKF4BVCZ1A77FRZ24), asserted equal in code at build time. brands.csv is authoritative for product_name, a brand id is minted once and never changes, and all seven sibling seasons carry that same string - so the alternative would have split the resolver for a word order. Same product, no minting question, per EXTRACTION_SPEC 5.1 reported rather than decided. THE CONSUMER CONSEQUENCE IS REAL AND IS THE POINT OF THIS ROW: a byte-for-byte matcher keyed on our product_name MISSES cardboardconnection's string entirely, because "Donruss Optic Basketball Premium Box Set" and "Donruss Optic Premium Box Set Basketball" share every word and no ordering. Any crosswalk to cardboardconnection for this line needs a token-set match, not a string match. Note separately that reference/caveats.csv CAV-035 records pipeline/brands.py as keeping its own map; "opticpremium" was one of the twelve prefixes backfilled there, so snapshot.py should resolve this line, but that was not run from this slice.
• G-005 base_set_size is written as 200 while the most quotable headline figure on BOTH dedicated product sources is 201. cardboardconnection's "Product Configuration:" line reads "201-card factory set" (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set) and Panini states "a 201-card set" and "Every box set contains 201 cards" (https://blog.paniniamerica.net/2017-18-donruss-optic-basketball-premium-box-set-arrives-on-icollectpanini-com/). 200 was written because both sources decompose their own 201 in the same place they state it: cardboardconnection's box-contents list reads "Full base set (200 cards)" then "1 Rated Rookie Autograph", and Panini's sentence reads "Every box set contains 201 cards, including 150 veteran players, 50 Rated Rookies and one exclusive Rated Rookie Signature." Two independent lineages splitting the same number the same way is the EXTRACTION_SPEC 5.4 confirmation to look for, and it agrees with tcdb's own header "Total Cards: 200" over a contiguous 1-200 (https://www.tcdb.com/ViewSet.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set). Per EXTRACTION_SPEC 5.3 the autograph is excluded from base_set_size. THE HAZARD: anyone reading only a headline figure, or comparing our 200 against either product page's 201, will read a one-card coverage hole where there is none. All seven sibling seasons made the same call against the same "201-card factory set" phrasing.
• G-006 THE NUMBER OF SETS PRODUCED IS SOURCED FOR THIS SEASON, WHICH RESOLVES AN OPEN QUESTION IN THE SIBLING RELEASE, AND THE TWO UNITS MUST NOT BE CONFLATED. opticpremium-bkb-2018-19 G-006 records that no source fetched stated how many boxes or sets were produced and that nothing was inferred from the per-card figure. For 2017-18 a source does state it, and it is first-party: Panini America writes "Just 249 Donruss Optic Premium Box Sets have been produced." (https://blog.paniniamerica.net/2017-18-donruss-optic-basketball-premium-box-set-arrives-on-icollectpanini-com/). That is the SET unit. The PER-CARD unit is stated separately and by a different lineage: cardboardconnection's "That means all 150 veterans and 50 Rated Rookie cards are included in every set, and each is limited to just 249 copies." (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set), corroborated by tcdb's literal "SN249" on all 200 rows and by Panini's own "Serial numbering will be mixed within each set, so every set has a chance for any sequential number, including jersey numbers and 1/249." THE TWO FIGURES COINCIDE AT 249 AND THAT COINCIDENCE IS ARITHMETICALLY NECESSARY, NOT A SECOND MEASUREMENT: 249 sets each holding all 200 cards yields exactly 249 copies of each card. So the agreement between the two sentences is not independent corroboration of either, and neither sentence was used to support the other. base_print_run=249 is written on the per-card sentence, which is the unit the column means. WHAT THE SCHEMA CANNOT HOLD: there is no column anywhere in this release for "249 sets produced", so the first-party set-count fact lives only in release.csv notes and in this row. A consumer asking "how many of these box sets exist" cannot get the answer from a structured column.
• G-007 rookie_flag BASIS IS THE LITERAL TOKEN "RR" ON THIS SID PLUS THE LITERAL TOKEN "RC" ON THE PARENT SID, AND THE SAME PHYSICAL CHECKLIST CARRIES DIFFERENT TOKENS UNDER THE TWO SIDS. Both were checked and both are quoted literally here with no expansion of either token, per EXTRACTION_SPEC 3.4 (the fetch layer invents glosses for flag tokens - see G-015, where it did). UNDER THIS PRODUCT'S SID 171625, rows 151-200 render the flag string "RR, SN249" and rows 1-150 render "SN249" with no "RR" and no "RC" on any of them - enumerated over all 200 rows, not sampled (https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=2&sort=1, https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=16&sort=0, https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=17&sort=0, https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=18&sort=0, https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=19&sort=0, https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=20&sort=0 for the upper block; https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=1, https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=1&sort=0, https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=8&sort=0, https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=9&sort=0, https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=10&sort=0, https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=2&sort=0 for the lower). UNDER THE PARENT SID 165179 the same rows render "RR, RC" and the rows below 151 render no token at all (https://www.tcdb.com/Checklist.cfm/sid/165179/2017-18-Donruss-Optic?PageIndex=2&sort=0). EXTRACTION_SPEC 2.4 asks for an explicit RC designation before writing yes or no; this sid does not carry one and the parent sid does, for the same cards, which is why rookie_flag=yes was written on 151-200 and no on 1-150 rather than UNKNOWN on all 200. The no side rests on the same within-parent discriminative test in the negative direction: a renderer that prints "RC" on 50 rows of the run and on none of the other 150. cardboardconnection describes the split in the same place: "all 150 veterans and 50 Rated Rookie cards" (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set), and Panini "150 veteran players, 50 Rated Rookies" (https://blog.paniniamerica.net/2017-18-donruss-optic-basketball-premium-box-set-arrives-on-icollectpanini-com/). NOT WRITTEN: card_tier is empty on all 200 rows, matching all seven sibling seasons, even though data/optic-bkb-2017-18 files the identical split as card_tier "Base" and "Rated Rookies" with a base_card_tiers.csv. That divergence is deliberate - base_print_run here is a flat sourced 249 rather than varies-by-tier, so EXTRACTION_SPEC 2.5 does not call for a tier file - but it means the two releases describing one checklist answer "which block is this card in?" differently.
• G-008 THE AUTOGRAPH SET'S CARD NUMBERS DUPLICATE BASE CARD NUMBERS WITHIN THIS RELEASE. Rated Rookies Signatures Premium Box Set renders 10 rows at numbers 151, 159, 174, 188, 192, 193, 196, 198, 199 and 200 (https://www.tcdb.com/ViewSet.cfm/sid/171641/zzz-probe-slug), every one of which is also a base card number in this release's base_cards.csv. Consequences, both recorded rather than resolved: (a) EXTRACTION_SPEC 5.3 has nothing to exclude by subtraction, because there is no numbered block PAST base_set_size - the autograph block sits inside 1-200, which is why base_set_size is 200 and not something smaller; (b) any consumer keyed on card_number alone within this release will collide these 10 autographs with 10 base cards, and only (set_name, card_number) separates them. This is the same shape as opticpremium-bkb-2018-19 G-008, which measured 25 such numbers for that season. NOT VERIFIED: whether the autograph card physically carries that number or whether tcdb has filed it against the base card's number; no source fetched states which, and no card was seen in hand.
• G-009 SCHEMA EXPRESSIVENESS: the autograph set's print run cannot be held in sets.csv. Every row of Rated Rookies Signatures Premium Box Set is serial numbered to 25 - tcdb renders the literal tokens "AU, SN25" on all 10 rows and the set-level "Notes: Serial numbered to 25" (https://www.tcdb.com/ViewSet.cfm/sid/171641/zzz-probe-slug), the parent's Inserts.cfm row for the sid carries the same Notes cell (https://www.tcdb.com/Inserts.cfm/sid/165179/2017-18-Donruss-Optic), and cardboardconnection states the cards "are numbered to just 25 copies apiece" (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set). EXTRACTION_SPEC 2.2 is explicit that sets.csv has no print_run column and that the figure must not be put in card_count, so 25 is carried in that row's notes in words and here. No parallels.csv row was written for it either: 25 is a property of the whole set, not a tier of it, and inventing a set-scope parallel row to hold a number would create an object no source describes. The distinct token set was enumerated across all 10 rows and is exactly {SN25} - there is no per-card variation here, unlike the 6-of-27 and 10-of-69 variable-tier rates measured elsewhere this season, so the figure is a flat run and the refusal is purely a column-availability problem. Same finding as opticpremium-bkb-2018-19 G-009, which recorded it for a 25-card /10 set.
• G-010 A VARIATION OR b-SUFFIXED LAYER CANNOT BE RULED OUT ON 57 OF THE 200 CARDS, AND THIS IS A REAL LIMIT ON THE ENUMERATION RATHER THAN A FORMALITY. EXTRACTION_SPEC 3.2 records that Gallery.cfm STRIPS the "b" suffix, so a variation row renders as a second tile on the same number with its token invisible, and that b rows are a Checklist.cfm job. Cards 1-75 and 101-168 were read off Checklist.cfm and no row in either range carries a letter suffix; two further Checklist parameterisations were asked the suffix question directly and both answered that no row on the page carries one (https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=1&sort=2 for 1-67, https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=2&sort=2 for 101-167). BUT CARDS 76-100 AND 169-200 REST ON GALLERY ALONE. Checklist.cfm page 1 was fetched under three different sorts and truncated at rows 75, 67 and 40 of 100; page 2 truncated at 168 and 167; and ViewSet.cfm on this sid truncated at row 74 (https://www.tcdb.com/ViewSet.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?sort=0), so no endpoint available here delivered the tails. Those 57 numbers therefore have no suffix-bearing rendering behind them. Errors.cfm CANNOT close this: reference/caveats.csv CAV-020 records that its silence discriminates nothing - it returns the same "no errors or variations" sentence on a set that provably has UER rows - so it was not fetched, because a clean response would have been no evidence and would have looked like some. WHAT SURVIVES: base_set_size 200, the contiguous 1-200 enumeration with no missing or duplicated number, and the absence of any numbered block past the base run, all of which rest on the enumeration itself plus the boundary probe (https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=21&sort=1). WHAT DOES NOT SURVIVE: any claim that this product has no variation layer.
• G-011 CONTRADICTION AGAINST THE TASK PROMPT AND AGAINST specs/2017-18_SCOPE_MEMO.md 5a, MEASURED TWICE. Both state that sid 171625 renders a "Parent Set: 2017-18 Donruss Optic" line, and the task prompt named it as one of three things to read back on the first fetch. The other two reproduced exactly - the heading "2017-18 Donruss Optic - Premium Box Set" and "Total Cards: 200" - but there is no "Parent Set" line on the page. Two fetches were each asked to quote back character for character every metadata line rendered above the card table, one under the canonical slug (https://www.tcdb.com/ViewSet.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set) and one under a deliberately wrong slug (https://www.tcdb.com/ViewSet.cfm/sid/171625/zzz-probe-slug), and each reported the heading, "Total Cards: 200" and "Notes: Serial numbered to 249" and no Parent Set line. No release date line renders on the page either. PARENTAGE IS ESTABLISHED ANYWAY, and structurally rather than from a header: the 2017-18 Donruss Optic parent's own Inserts.cfm lists sid 171625 in its "Parallel Sets" section and sid 171641 in its "Insert Sets" section (https://www.tcdb.com/Inserts.cfm/sid/165179/2017-18-Donruss-Optic, https://www.tcdb.com/Inserts.cfm/sid/165179/2017-18-Donruss-Optic?sort=2), which is the direction of evidence that matters. Recorded because the memo claim was made under a heading read-back and a later agent reading the memo against the page would find a mismatch and not know which to trust; it may be that tcdb renders that line only in some contexts, and it may be that the field was read off a different page.
• G-012 SINGLE-SOURCE AND SINGLE-LINEAGE ITEMS, LISTED SO THEY ARE NOT MISTAKEN FOR SETTLED VALUES. (a) release_date 2018-04-23 rests on ONE statement of the value: cardboardconnection's label "Release Date: 04/23/2018", re-fetched under a distinct URL and confirmed not to contain the word "Estimated" (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set). Panini's post announcing that the product "Arrives on iCollectPanini.com" carries the publication byline "April 23, 2018" (https://blog.paniniamerica.net/2017-18-donruss-optic-basketball-premium-box-set-arrives-on-icollectpanini-com/) and a narrow re-fetch established that the date appears only as that byline and nowhere in the body - so it is off-lineage support for the same day but it is a post date, not a stated release date, and it was not counted as a second statement. (b) THE BASE ROSTER IS SINGLE-SOURCE TO tcdb. No aggregator publishes a card-by-card Premium Box Set checklist for 2017-18; cardboardconnection gives counts and print runs but no roster. The roster's internal corroboration is that it matches data/optic-bkb-2017-18 197-200 of 200 per field (G-002), which is a second tcdb sid rather than a second source. (c) checklistcenter APPEARS NOT TO CARRY A 2017-18 DONRUSS OPTIC PAGE AT ALL, and per reference/caveats.csv CAV-037 that is not a finding: its slugs are irregular, a constructed URL was therefore never built, and a domain-scoped search for its 2017-18 Optic page returned its 2018-19 through 2024-25 Optic pages and nothing earlier. Absence from a search result is evidence of nothing. (d) BECKETT REMAINS UNREACHABLE - now the eleventh consecutive pass to record it. https://www.beckett.com/news/2017-18-donruss-optic-basketball/ was surfaced by web search and, fetched, returns HTTP 302 to https://maintenance.beckett.com/, so the usual third structural lineage is gone and any "two sources" in this build means cardboardconnection plus either Panini or tcdb. (e) THE ONE GENUINELY INDEPENDENT LEG THIS BUILD GAINED that its siblings did not have is Panini America's own blog, which is first-party and off the cardboardconnection lineage entirely. It is why base_print_run here is corroborated across lineages rather than resting on cardboardconnection alone as it does in opticpremium-bkb-2018-19.
• G-013 THE PARENT'S OWN AUTOGRAPH SET AND THIS ONE ARE DIFFERENT OBJECTS WITH OVERLAPPING NAMES, AND THE NAMES ARE THE ONLY THING KEEPING THEM APART. tcdb carries "Rated Rookies Signatures" (sid 171628) and "Rated Rookies Signatures Premium Box Set" (sid 171641) as two rows of the same Insert Sets section of the same parent page (https://www.tcdb.com/Inserts.cfm/sid/165179/2017-18-Donruss-Optic?sort=2), with a further eleven colour parallels of the former (Black 171629, Blue 171630, Checkerboard 171631, Fast Break 171632, Fast Break Black 171633, Fast Break Pink 171634, Gold 171635, Gold Vinyl 171636, Green 171637, Holo 171638, Pink 171639, Purple 171640). data/optic-bkb-2017-18/sets.csv carries the parent one at card_count 50; this release carries the Premium one at 10 (https://www.tcdb.com/ViewSet.cfm/sid/171641/zzz-probe-slug). The two are not a set and its parallel - the Premium one has a different roster size, a different print run (SN25 against the parent set's own runs) and, per cardboardconnection, is "exclusive to the format" (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set). Recorded because a name-keyed resolver that drops the "Premium Box Set" qualifier will merge a 10-card /25 set into a 50-card one across two releases.
• G-014 THIS PRODUCT IS INVISIBLE TO THE SEASON INDEX AND THAT IS A COVERAGE HAZARD FOR ANY FUTURE 2017-18 SWEEP, not a defect in this build. It has NO PARENT SID OF ITS OWN: tcdb files it as subset sid 171625 under the 2017-18 Donruss Optic parent (sid 165179), so ViewAll.cfm/sp/Basketball/year/2017 does not list it and nothing product-level points at it. It was reached only through Inserts.cfm on the parent (https://www.tcdb.com/Inserts.cfm/sid/165179/2017-18-Donruss-Optic). specs/2017-18_SCOPE_MEMO.md 5a records this; it reproduced exactly. The same is true of the autograph set (sid 171641). Anyone bounding the season from the index alone will under-count 2017-18 by at least this release, and the same shape should be expected for the Hoops Premium Box Set, whose modelling the memo leaves open in 8.2.
• G-015 FETCH-LAYER DEFECTS OBSERVED IN THIS PASS, recorded because EXTRACTION_SPEC 3.4 predicts them and two of them fired. (a) THE LAYER INVENTED A GLOSS FOR A FLAG TOKEN. Asked to list the tiles on https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=18&sort=0 and told not to explain any token, an earlier formulation of the same request returned the ten correct tiles and tokens and then added, unprompted, that "RR, SN249" indicates "Rookie Relic cards with a serial number of 249". No such expansion appears on the page, "Relic" is not what RR denotes on a card with no memorabilia, and the rows carry no MEM token. The gloss was discarded; only the literal token strings were recorded anywhere in this release. Subsequent asks carried an explicit instruction not to explain any token and none of them produced a gloss. (b) A FETCH REPORTED ITS OWN TRUNCATION AS THE END OF THE PAGE. https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=1&sort=1 was asked for cards 76-100 and answered "The page stopped at card 75", which is the documented shape of presenting a truncation as a boundary; the real page paginates at 100 and rows 76-100 exist, as https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=9&sort=0 and https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=10&sort=0 show. (c) NOT OBSERVED: no constant-shift renumbering was detected. Three named-player checks were run against enumerated tile lists rather than against direct questions (Redick 113, McCollum 123, Selden 153) and all three agreed with the Checklist transcription, and the two page overlaps (71-75, 161-168) agreed row for row across endpoints.
• G-016 LICENCE STATUS IS UNSOURCED AS TO GRANTOR, and the value was given by the task prompt rather than found. license_status "licensed (NBA)" was written as instructed, with licensed=yes and league=NBA. No source fetched in this pass states a licensing grantor for this product: neither cardboardconnection (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set) nor Panini's own post (https://blog.paniniamerica.net/2017-18-donruss-optic-basketball-premium-box-set-arrives-on-icollectpanini-com/) names the NBA or the NBPA as granting anything, and tcdb carries no licence field. The weaker sourceable form was used deliberately per the round brief; "licensed (NBA/NBPA)" was not written because nobody has ever sourced a grantor on this project. All seven sibling seasons carry the same value on the same basis.
• G-017 ENUMERATION METHOD AND TRUNCATION MEASUREMENTS, offered for EXTRACTION_SPEC 3.2 because this parent adds five more data points to a figure that has been revised downward four times. All 200 base rows were read off sid 171625 in these fetches, and the card ranges each one supplied: https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=1 gave 1-40; https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=1&sort=0 gave 41-75; https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=8&sort=0 gave 71-80; https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=9&sort=0 gave 81-90; https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=10&sort=0 gave 91-100; https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=2&sort=0 gave 101-140; https://www.tcdb.com/Checklist.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=2&sort=1 gave 141-168; https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=16&sort=0 gave 151-160; https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=17&sort=0 gave 161-170; https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=18&sort=0 gave 171-180; https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=19&sort=0 gave 181-190; https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=20&sort=0 gave 191-200; plus https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=12&sort=0 and https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=13&sort=0 for the named-name re-checks and https://www.tcdb.com/Gallery.cfm/sid/171625/2017-18-Donruss-Optic-Premium-Box-Set?PageIndex=21&sort=1 for the upper boundary. TRUNCATION MEASURED ON Checklist.cfm, 100-row pages: page 1 cut at row 75 under ?sort=0 and again under ?sort=1 (that fetch, asked for cards 76-100, answered that the page stopped at card 75), and at row 67 under ?sort=2; page 2 cut at row 68 of 100 under ?sort=1 (card 168, with card 169 cut mid-row) and at row 67 under ?sort=2 (card 167). ViewSet.cfm on the same subset sid cut at row 74. So five Checklist observations on one parent: 67, 67, 68, 75, 75 of 100 - all inside the band specs/2017-18_SCOPE_MEMO.md 7.2 measured this season (40, 44, 50, 50, 50, 61, 76, 80, 80, 82, 86) and none of them at its floor. Gallery.cfm delivered a full 10 tiles on every one of the nine pages fetched, with no dropped tile and no dropped flag token; every Gallery URL carried &sort=0 or &sort=1 so no two were the same URL, per the 15-minute per-URL error cache warning. Total fetch cost of the base run: 12 tcdb fetches for 200 cards, against 20 for Gallery alone.
• G-018 MARKETPLACE CORROBORATION WAS RUN, PASSED THE CONTAMINATION TEST, AND CARRIES ONE RESERVATION THAT IS STATED RATHER THAN BURIED. An eBay single-card listing title reads "2017-18 Panini Donruss Optic - Rated Rookie Dillon Brooks #152 Premium Box Set Prizm /249 (RC)" (https://www.ebay.com/itm/358024215759). It corroborates two things independently of tcdb: that card 152 of this product is Dillon Brooks, which matches this file, and the /249 per-card run. EXTRACTION_SPEC 3.4 contamination check: the title was re-fetched and asked, character for character, whether it contains any of "SN249", "RR,", "AU,", "MEM" or "EXCH", and it contains none of them - so no tcdb flag string appears verbatim and the documented fingerprint does not fire. THE RESERVATION: the title is shaped like an eBay CATALOGUE title rather than free seller text ("Product - description #number"), which is one of the four documented contamination shapes, and its "(RC)" is a token tcdb also uses on the parent sid - though RC is universal hobby shorthand and the serial flag SN249 is absent. The listing is therefore recorded as supporting corroboration and base_print_run does NOT rest on it: it rests on cardboardconnection's per-card sentence, Panini's first-party set-count sentence and a 200-row tcdb token enumeration. No copy number off a card in hand was found for this product.
• G-019 NO PACK ODDS WERE FOUND, AND THE ONE ODDS VALUE WRITTEN IS A BOX GUARANTEE RATHER THAN A PACK ODD. scarcity_tier, stated_odds, odds_evidence_url and odds_evidence_quote are empty on the Base Set row. On the autograph row stated_odds is "1 per box set", sourced to Panini's "Every box set contains 201 cards, including 150 veteran players, 50 Rated Rookies and one exclusive Rated Rookie Signature." (https://blog.paniniamerica.net/2017-18-donruss-optic-basketball-premium-box-set-arrives-on-icollectpanini-com/), corroborated by cardboardconnection's box-contents list "Full base set (200 cards)" then "1 Rated Rookie Autograph" (https://www.cardboardconnection.com/2017-18-donruss-optic-basketball-premium-box-set). That is a guaranteed contents statement about a sealed set, not a pull rate, and it should not be read as an odds ratio. scarcity_tier on that row is UNKNOWN because no source places the set on a short-print or SSP ladder - the product has no packs and therefore no insertion ladder at all. Nothing was written into scarcity_tier from the print run, which would have been a derivation.

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