Marco Arment: raw computed results
The raw ranking from the production similarity calculation (face-embedding centroid cosine → gender-cohort z-normalization). The web-opinion crown, curation, and fusion re-rank are removed.
- 1
77.8%z 3.99 - 2
63.7%z 3.84 - 3
56.9%z 3.76 - 4
54.2%z 3.73 - 5
47.6%z 3.64 - 6
46.1%z 3.54 - 7
46%z 3.54 - 8
45.9%z 3.53 - 9
45.8%z 3.52 - 10
45.4%z 3.50 - 11
45.3%z 3.49 - 12
45.2%z 3.49 - 13
44.8%z 3.46 - 14
44.2%z 3.42 - 15
43.1%z 3.35 - 16
42.4%z 3.31 - 17
41.9%z 3.27 - 18
41.6%z 3.25 - 19
41.4%z 3.24 - 20
40.9%z 3.21 - 21
40.9%z 3.20 - 22
40.7%z 3.19 - 23
40.5%z 3.18 - 24
40.3%z 3.17 - 25
39.9%z 3.14 - 26
39.7%z 3.13 - 27
39.5%z 3.12 - 28
39.3%z 3.10 - 29
39.1%z 3.09 - 30
39.1%z 3.09 - 31
38.9%z 3.08 - 32
38.9%z 3.07 - 33
38.6%z 3.06 - 34
38.5%z 3.05 - 35
38.4%z 3.05 - 36
38.4%z 3.04 - 37
38.1%z 3.03 - 38
37.8%z 3.00 - 39
37.7%z 3.00 - 40
37.4%z 2.98 - 41
36.8%z 2.94 - 42
36.6%z 2.93 - 43
36.4%z 2.91 - 44
36%z 2.89 - 45
35.9%z 2.88 - 46
35.6%z 2.87 - 47
35.6%z 2.86 - 48
35.6%z 2.86 - 49
35.1%z 2.83 - 50
34.9%z 2.82