Grouprather

Grouprather is a term used in decision science to describe a method for aggregating individual judgments into a single group estimate through iterative weighting based on demonstrated accuracy. The core idea is to reward reliable contributors and to refine the group's estimate by reweighting inputs as participants' performance on calibration questions becomes evident.

In a typical grouprather process, participants submit estimates for a set of calibration questions with known answers. Each participant receives an initial equal weight. After initial estimates are scored against true values, weights are assigned proportional to accuracy, and a new round of estimates is collected. The process may repeat for several rounds, updating weights and the consensus until convergence. The final group estimate is a weighted combination of participants' predictions.

Grouprather has been discussed as a framework for crowdsourced forecasting, risk assessment, and decision support in settings where diverse expertise must be synthesized quickly without centralized control.

The term does not refer to a specific product or organization; in scholarly discussions it is described as an abstract ensemble method that can be implemented with different scoring rules and interfaces.

Limitations include sensitivity to calibration data, potential biases in the sample, and strategic input by participants seeking to influence the outcome. Effective use typically requires diverse participation and transparent weighting criteria.

Related concepts include wisdom of crowds, ensemble forecasting, prediction markets, and judgment aggregation.