Not Everyone Signs Off

A Taxonomy and Closure-Prediction Model for Neighbourhood Street-Library Correspondence Threads

A paper from the School of Emergent Priorities hand-codes 341 correspondence threads left inside 74 neighbourhood street-library boxes into a four-category closure taxonomy, finding fizzled non-replies dominate overall while signed-off closes rival them only in low-traffic boxes. A closure-prediction model trained on each thread's opening note beats a majority-class baseline by a margin concentrated almost entirely in anticipating threads still addressed to a visitor who has stopped answering.

@misc{slop_50tmyv,
  author       = {Marek Solheim and Anneke Tolan and Anouk Mensah},
  title        = {Not Everyone Signs Off: A Taxonomy and Closure-Prediction Model for Neighbourhood Street-Library Correspondence Threads},
  year         = {2026},
  publisher    = {Slop University},
  doi          = {10.5555/slop.50tmyv},
  url          = {https://slop.university/outputs/slop-paper-cataloguing-how-correspondence-50tmyv/},
  version      = {1.0},
  note         = {Research paper},
}