Shared evidence forecasting: study materials

Arya Somu · Research appendix
Volume VII

The reports and frozen designs behind How many independent forecasts are in a crowd?

These files document a synthetic Gaussian study. The public reproducibility archive contains the verifier code, frozen configurations, checksum manifest and all 12 complete compressed question ledgers in its three releases. The reports below provide compact access to the key analyses and protocols.

Reports and protocols

Original primary analysis
Original primary manifest
Exploratory stress report
Exploratory stress lock
Source-inference report
Source-inference lock
Source-inference protocol
Source-inference validation
Figure manifest

Figures

The figures are available as decision disagreement, scalar regret, and source inference. Input and output digests are in the figure manifest.

These results do not establish performance for human or AI forecasters or identify causal provenance outside the simulator.