A Bradley–Terry Model of the Produce Swap Table

Latent desirability from 2,148 take-or-leave decisions across eleven monthly swaps

A research poster from the School of Emergent Priorities fits a Bradley–Terry paired-comparison model to 2,148 take-or-leave decisions logged by trestle-side observers across eleven monthly community produce swaps, then splits the 34 catalogued item types between a run set out in the order the model ranks them and a run set out at random, holding the allocation for the series. Consecutive-refit rank agreement averages Kendall's τ = 0.87 under the ordered layout and 0.31 under randomisation, and the near end of the table carries a 1.42-logit advantage in the odds of an item being taken — wider than the fitted ability spread across 30 of the 34 item types. The poster reads the ordered run's stability as a property of the layout the ranking sets rather than of the preference it is taken to measure, and notes the fitted order's registration with the Indicator Commons as the swap's monthly desirability statement.

@misc{slop_unlxck,
  author       = {Petra Umbile and Osei Vandermeer and Ronja Oyelaran},
  title        = {A Bradley–Terry Model of the Produce Swap Table: Latent desirability from 2,148 take-or-leave decisions across eleven monthly swaps},
  year         = {2026},
  publisher    = {Slop University},
  doi          = {10.5555/slop.unlxck},
  url          = {https://slop.university/outputs/slop-poster-a-bradley-terry-model-unlxck/},
  version      = {1.0},
  note         = {Research poster},
}