Online Social Choice and Welfare Seminar: Gustav Alexandrie, Tuesday 26 May
7 views
Skip to first unread message
Marcus Pivato
May 18, 2026, 10:47:33 AM (22 hours ago) May 18
to social-choice-a...@googlegroups.com, com...@duke.edu
[with apologies for cross-posting]
Dear all,
The next presentation in the Online Social Choice and Welfare Seminar will be next Tuesday (26 May). Here are the details.
Time: 5PM GMT (10AM Vancouver, 1PM Toronto/Montréal, 2PM Rio de Janeiro, 6PM London, 7PM Zürich, 8PM Istanbul, 10:30PM New Delhi)
Speaker: Gustav Alexandrie (Universität Zürich)
Title: "Probability Aggregation Under Equal Expected Accuracy"
Abstract: A (binary) pooling function takes the probabilistic forecasts of two experts and returns an aggregated probability. We introduce the property of Equal-Expected-Accuracy (EEA) explainability, which is satisfied by pooling functions that treat the two experts as having equal expected accuracy under some scoring rule. EEA-explainable pooling functions provide a natural way of (1) formalizing the concept of epistemic peerhood from the philosophical literature on disagreement, and (2) designing scoring rules for paying forecasters fairly given the aggregated probability.
We prove two characterization theorems showing that EEA-explainable pooling functions can be expressed in terms of their associated scoring rules. These theorems imply that, under EEA-explainability, forecast aggregation among epistemic peers reduces to the problem of finding the scoring rules that best capture forecast accuracy in the relevant setting. We also establish a dual result showing how scoring rules can be expressed in terms of their associated pooling functions. This theorem yields a recipe for constructing scoring rules that pay forecasters fairly given the aggregated probability.
Finally, we explore the implications of EEA-explainability for the most commonly used pooling functions. In particular, we show that (a) linear pooling is uniquely EEA-explainable with respect to quadratic scoring rules, (b) geometric pooling is uniquely EEA-explainable with respect to a novel class of scoring rules based on the square root of the odds, and (c) multiplicative pooling is not EEA-explainable. Moreover, we show that EEA-explainability with respect to the standard logarithmic and spherical scoring rules yields novel pooling functions that have not been discussed in prior literature.
Reminder: On the seminar website you can find the video recordings, slides and supplementary materials for all past presentations, as well as information about future presentations.
Dear all,
The next presentation in the Online Social Choice and Welfare Seminar will be next Tuesday (26 May). Here are the details.
Time: 5PM GMT (10AM Vancouver, 1PM Toronto/Montréal, 2PM Rio de Janeiro, 6PM London, 7PM Zürich, 8PM Istanbul, 10:30PM New Delhi)
Speaker: Gustav Alexandrie (Universität Zürich)
Title: "Probability Aggregation Under Equal Expected Accuracy"
Abstract: A (binary) pooling function takes the probabilistic forecasts of two experts and returns an aggregated probability. We introduce the property of Equal-Expected-Accuracy (EEA) explainability, which is satisfied by pooling functions that treat the two experts as having equal expected accuracy under some scoring rule. EEA-explainable pooling functions provide a natural way of (1) formalizing the concept of epistemic peerhood from the philosophical literature on disagreement, and (2) designing scoring rules for paying forecasters fairly given the aggregated probability.
We prove two characterization theorems showing that EEA-explainable pooling functions can be expressed in terms of their associated scoring rules. These theorems imply that, under EEA-explainability, forecast aggregation among epistemic peers reduces to the problem of finding the scoring rules that best capture forecast accuracy in the relevant setting. We also establish a dual result showing how scoring rules can be expressed in terms of their associated pooling functions. This theorem yields a recipe for constructing scoring rules that pay forecasters fairly given the aggregated probability.
Finally, we explore the implications of EEA-explainability for the most commonly used pooling functions. In particular, we show that (a) linear pooling is uniquely EEA-explainable with respect to quadratic scoring rules, (b) geometric pooling is uniquely EEA-explainable with respect to a novel class of scoring rules based on the square root of the odds, and (c) multiplicative pooling is not EEA-explainable. Moreover, we show that EEA-explainability with respect to the standard logarithmic and spherical scoring rules yields novel pooling functions that have not been discussed in prior literature.
To obtain the Zoom link, please subscribe to the Seminar Mailing List, or contact one of the organisers.
Reminder: On the seminar website you can find the video recordings, slides and supplementary materials for all past presentations, as well as information about future presentations.
--
Marcus Pivato
Centre d'Économie de la Sorbonne
Centre d'Économie de la Sorbonne
Université Paris 1 Panthéon-Sorbonne
Reply all
Reply to author
Forward
0 new messages