Will an image model win Scott Alexander’s bet on compositionality, to Edwin Chen’s satisfaction, in 2023?
  • 376
  • 272
  • $591
Scott Alexander and Edwin Chen are conducting a bet to determine whether an image model has achieved compositionality, with Gary Marcus serving as a substitute arbiter. The resolution criteria is based on whether Edwin can access the top models of late 2023 or use public demos and his own judgment, with defaulting to false if no clear progress is made. A resolution of "likely true" would imply that current AI models have successfully demonstrated compositionality. Gary Marcus will be used as a substitute arbiter if neither Scott Alexander nor Edwin Chen can resolve the question, resulting in an unclear outcome. The bet's exclusive authority lies with Scott Alexander's judgment, which will govern the market's resolution.

Probability History

0102030405060708090100↑ ProbabilityJan2023AprJulOct

Source: brier.fyi

Detailed Scores

Criterion Grade Brier/QuadraticLogarithmicSpherical
At Market Midpoint
F
0.2711
-0.7353
0.6773
Time-Weighted Average
F
0.3567
-0.9094
0.5592
30 Days Before Close
F
0.4146
-1.0326
0.4839
7 Days Before Close
F
0.3650
-0.9266
0.5481
Relative Score
A+
-0.053
+0.136
+0.061
Criterion Grade Brier/QuadraticLogarithmicSpherical
At Market Midpoint
F
0.4417
-1.0924
0.4506
Time-Weighted Average
F
0.4605
-1.1352
0.4280
30 Days Before Close
F
0.7627
-2.0661
0.1436
7 Days Before Close
F
0.7708
-2.1031
0.1377
Relative Score
F
+0.053
-0.136
-0.061
  • Overall
  • 376
  • 272
  • $591
Criterion Grade Brier/QuadraticLogarithmicSpherical
At Market Midpoint
F
0.3564
-0.9139
0.5639
Time-Weighted Average
F
0.4086
-1.0223
0.4936
30 Days Before Close
F
0.5887
-1.5493
0.3137
7 Days Before Close
F
0.5679
-1.5149
0.3429

Similar Questions

  • 1,481
  • $522,273

Manifold
C-
+0.007
Polymarket
B
-0.007
  • 642
  • $2,888

Manifold
A
-0.025
Metaculus
D
+0.025
  • 434
  • $586

Manifold
C
+0.001
Metaculus
C+
-0.001
  • 415
  • $340

Manifold
A
-0.028
Metaculus
D-
+0.028