9.2 KiB
Card data: real stat distributions (council-data)
Status: final for tonight, 236 fully enriched cards, generated 2026-11-16 23:20 EST. Sampling stopped early on night-lead's request because Wikimedia was rate-limiting (429) our shared egress IP. 330 cards were picked; the first 236 finished enrichment.
Method: scripts/night/card_sample.py draws random ns0 articles in batches of 20 and keeps those that pass the server/wiki.ts filter: a thumbnail from upload/thumb.wikimedia.org and an intro extract of at least 40 characters. About 44% of random pages pass. Each card is enriched with the same API parameters as server/enrichment.ts:
- six completed months, Jun–Oct 2026; attention uses October
- monthly views summed from the daily REST pageviews
On top of that it adds Wikidata P31 and a 30-day drift (views 10-17..11-15 vs 09-17..10-16).
Politeness: the sampler runs serially at 1 request a second or slower, honours Retry-After, and caches every response to disk. scripts/night/card_analyze.py produces everything below, plus the percentiles (99 cut points each, for DESIGN §1.1) in card-sample.json.
Files
card-sample.json:cards[]with thePageEnrichmentshape, pluscategory(legacycategory()),instanceOf(P31),faction(the P31 rules below),tiersanddrift30. Top level:percentiles,tierBreakpoints,factionRulesandnote.starter-cards.json: the loaner pool, 20 real frozen snapshots (aMintStatsSnapshotinsnapshot). It also holdsstarterDeckPageIds: the 48 DESIGN §2.1 starter titles with their real pageIds andcardIds.
Headline findings (236 cards)
-
Raw scores are compressed.
score median p10–p90 max attention 32 20–53 77 depth 63 54–73 87 reach 20 0–45 70 No honest card reaches 100. Use
percentiles(§1.1) or the 1–10 tiers, never the raw 0–100 values. -
Correlations (Spearman):
- attention and depth: 0.61, so these two double-count fame
- reach is nearly independent of both (0.17 with attention, 0.02 with depth)
-
Reach has a hard floor: 18% of cards (42) have 0 other languages, so reach is 0. Under the §1.1 counting, all of them get pct_reach of about 8.
-
Raw values are long-tailed:
value median p90 max monthly views 82 1,576 42k pageBytes 6.1 KB 24 KB languageCount 3 22 -
Trend is mostly down:
- 65% of cards have a negative trendPercent, 34% positive; median −9.5%
- 16% swing by more than 50% either way
- null: 1%
-
Live drift is small in score terms. The 30-day median |pageview drift| is 22% (p90 60%). That moves attention by only 1.7 points (p90 6.3), so at-mint and live values differ little unless someone inflates them.
-
The legacy
category()split is very skewed:category share Culture 56% People 26% Nature 8% History 8% Science 1% Space 0.4% -
Candidate factions from P31:
faction share People 29% Places 20% Nature 8% Structures 7% Works 5% Events 5% Organisations 4% Ideas & Tech 3% Sport 3% unclassified 14% Suggested 6 for v2: People, Places, Nature, Works & Culture (Works + Organisations + Sport), Structures, Events & Ideas, plus a neutral fallback.
Full tables
Sample: 236 cards; ~44% (149/340 logged during sampling) of random pages pass the illustrated filter. Pageviews available for 236.
| stat | n | min | p10 | p25 | median | p75 | p90 | max | mean | skew |
|---|---|---|---|---|---|---|---|---|---|---|
| attention | 236 | 13 | 20 | 25 | 32 | 41 | 53 | 77 | 35 | 0.78 |
| depth | 236 | 48 | 54 | 58 | 63 | 68 | 73 | 87 | 64 | 0.42 |
| reach | 236 | 0 | 0 | 10 | 20 | 33 | 45 | 70 | 23 | 0.36 |
| pageBytes | 236 | 772 | 1847 | 3084 | 6110 | 12805 | 24412 | 168332 | 11691 | 4.61 |
| languageCount | 236 | 0 | 0 | 1 | 3 | 9 | 22 | 124 | 8 | 4.14 |
| monthlyViews | 236 | 5 | 16 | 30 | 82 | 300 | 1576 | 42025 | 812 | 9.82 |
| dailyAverage | 236 | 0.2 | 0.5 | 1.0 | 2.6 | 9.7 | 50.8 | 1355.6 | 26.2 | 9.82 |
| trendPercent | 234 | -86.6 | -49.3 | -26.3 | -9.5 | 5.3 | 31.2 | 388.9 | -3.6 | 4.11 |
| drift30Percent | 235 | -94.7 | -58.2 | -33.8 | -12.9 | 1.0 | 27.7 | 2268.7 | -4.2 | 14.07 |
Spearman rank correlations (cards with pageviews):
| attention | depth | reach | trendPercent | drift30Percent | |
|---|---|---|---|---|---|
| attention | 1.00 | 0.61 | 0.17 | 0.35 | 0.33 |
| depth | 0.61 | 1.00 | 0.02 | 0.15 | 0.14 |
| reach | 0.17 | 0.02 | 1.00 | 0.00 | 0.09 |
| trendPercent | 0.35 | 0.15 | 0.00 | 1.00 | 0.31 |
| drift30Percent | 0.33 | 0.14 | 0.09 | 0.31 | 1.00 |
(n=234; attentionmonthlyViews, depthpageBytes, reach~languageCount are monotonic so their rank correlations are identical.)
trendPercent sign: positive 79 (34%), negative 153 (65%), zero 2; null (prior month 0 or no views) 2. |trend| > 50%: 38; > 100%: 7.
30-day drift (views 10-17..11-15 vs 09-17..10-16): median |drift| 22%, p90 |drift| 60%. Attention score change implied: median 1.7 points, p90 6.3 points.
attention histogram (deciles 0-9,10-19..90-100): 0 19 74 74 36 20 12 1 0 0 depth histogram (deciles 0-9,10-19..90-100): 0 0 0 0 2 72 117 38 7 0 reach histogram (deciles 0-9,10-19..90-100): 42 60 53 43 21 15 1 1 0 0
Reach = 0 (no other languages): 42; reach 100 (>=999 langs) impossible in practice, max langs 124. Attention null: 0; attention 0: 0.
Sum of three stats: min 73, p10 89, p25 102, median 116, p75 135, p90 156, max 222
Legacy category() (server/wiki.ts): Culture 133, People 62, Nature 19, History 18, Science 3, Space 1
Top Wikidata P31 classes: human 64, taxon 16, village 7, commune of France 6, river 5, human settlement 4, musical group 3, building 3, gene 3, village of Poland 2, civil parish 2, protest 2, legislative election 2, castle 2, events in a specific year or time period 2, township of Pennsylvania 2, film 2, company 2, island 2, rural municipality of Austria 1, municipality of Austria 1, municipality of Brazil 1, television film 1, nor'easter 1, rail company (Japan) 1, municipality 1, association football match 1, final 1, local government area of Victoria 1, plant disease 1, basketball team season 1, mosque 1, functional programming language 1, purely functional programming language 1, dependently typed programming language 1, programming language 1, free software 1, proof assistant 1, municipality of Croatia 1, municipality of Bangladesh 1
No Wikidata item: 0; no P31: 13
Most frequent category words (cards containing): stubs 81, births 55, people 50, from 50, century 46, living 33, geography 30, american 29, deaths 28, united 27, county 27, states 23, established 21, establishments 20, alumni 18, places 17, university 16, buildings 16, state 15, national 14, structures 14, male 13, players 12, district 11, villages 11, sportsmen 11, members 11, described 10, football 10, sports 10, city 10, film 10, british 9, companies 9, populated 9, sportspeople 9, biography 9, york 9, women 8, taxa 8, germany 8, politicians 8, television 7, writers 7, french 7, genera 7, based 7, communes 7, language 7, expatriate 7, named 7, olympics 7, completed 7, historic 7, register 7, faculty 6, pennsylvania 6, civil 6, history 6, california 6
Rank-normalised tiers (1–10)
Tier = 1 + number of breakpoints the score is >=. Breakpoints are the sample deciles p10..p90 of the 0–100 score. Ties make some tiers thin (scores are integers).
| stat | p10 | p20 | p30 | p40 | p50 | p60 | p70 | p80 | p90 |
|---|---|---|---|---|---|---|---|---|---|
| attention | 20 | 24 | 27 | 30 | 32 | 35 | 39 | 44 | 53 |
| depth | 54 | 57 | 59 | 61 | 63 | 65 | 67 | 69 | 73 |
| reach | 0 | 10 | 10 | 16 | 20 | 26 | 30 | 36 | 45 |
Resulting tier counts — attention: None:0 1:19 2:27 3:24 4:23 5:23 6:17 7:28 8:25 9:22 10:28 ; depth: 1:14 2:23 3:26 4:21 5:25 6:26 7:22 8:21 9:30 10:28 ; reach: 1:0 2:42 3:0 4:30 5:30 6:32 7:21 8:27 9:27 10:27
Candidate factions (Wikidata P31 rules)
| faction | n | share | median attention | median depth | median reach | example |
|---|---|---|---|---|---|---|
| People | 68 | 29% | 31 | 65 | 22 | Salahuddin of Aceh |
| Places | 48 | 20% | 28 | 61 | 32 | St. Georgen am Längsee |
| Other | 34 | 14% | 40 | 65 | 16 | Village (Pennsylvania) |
| Nature | 20 | 8% | 30 | 56 | 29 | Dichromorpha viridis |
| Structures | 17 | 7% | 26 | 57 | 10 | Pathrail Mosque |
| Works | 12 | 5% | 36 | 63 | 13 | The Best of Men |
| Events | 12 | 5% | 40 | 72 | 13 | April 2021 nor'easter |
| Organisations | 10 | 4% | 48 | 68 | 10 | Hiroshima Electric Railway |
| Ideas & Tech | 8 | 3% | 30 | 63 | 16 | Agda (programming language) |
| Sport | 7 | 3% | 27 | 70 | 16 | 1912 Argentine Primera División final |
Unclassified P31 labels: typeface 1, annexation 1, position 1, family 1, historic site 1, automobile manufacturer 1, historical society 1, stock character 1, pigeon breed 1, shopping center 1, Wikimedia list article 1, clinical sign 1, tram route 1, rubber-tyred tram 1, oil refinery 1, waterfall 1, orisha 1, water deity 1, Loa 1, arch 1, award 1, sex scandal 1, military base 1, artillery battery 1