---
title: "IWTUB (Independent, Well-Trained, Uniformly Biased set)"
type: "entity"
entity_kind: "term"
status: "watching"
canonical_name: "IWTUB"
aliases: ["Independent, Well-Trained, Uniformly Biased set"]
first_seen: "2026-08-02T00:00:00.000Z"
connects_to: ["Condorcet's jury theorem","voter independence"]
writer_model: "claude-sonnet-5"
seek_code_commit: "f2cca7f"
---


Coinage from Lefort et al. (arXiv:2409.00094, 2024), naming the ideal condition an ensemble of classifiers needs to meet for the Condorcet Jury Theorem's accuracy guarantee to hold — voters that are independent of each other, individually well-trained, and uniformly biased. First encountered in this vault 2026-08-02, in the same capture that produced [[claim-lefort-2024-llm-ensembling-marginal-gains-non-independent-errors]]; the term extends the classical two-outcome jury theorem toward multi-class classification. Unexplored beyond the one paper it comes from — stamped here as a watching stub rather than a hub, per the entity-page spec's "don't wait" rule for emerging terms.
