Preserve creative differences in federated aggregation

Determine whether current federated learning aggregation methods can preserve the differences among contributors in small creative communities rather than smoothing those differences away through averaging.

Background

The paper argues that creative federations may contain only a small number of highly heterogeneous contributors, such as studios or artists. In this setting, heterogeneity is not merely statistical noise: it may be the generative reason for federating. The authors therefore identify whether aggregation can retain meaningful creative differences as an unresolved technical and social question requiring collaboration with the relevant communities.

References

Whether current aggregation methods can preserve that difference rather than smooth it is, to our knowledge, an open question.

— Govern the Model, Not Only the Data: Storage, Circulation, and Learning in Creative AI  (2609.03800 - Perry et al., 3 Sep 2026) in Section 4, “The research space this offers”

Four questions raised by the framework are, to our knowledge, open. (i) Experience gradient: can representation scope, behavioral trust tier, and human-approval gate be unified as a single gradient on one ordered structure, making the authority formula of \S\ref{sec:trust} a theorem rather than an engineering rule? (ii) Order-freedom: commitment semantics are history-free in principle~\citep{singh2000, yolum2002}; if so, temporal weight is an implementation leak rather than an inherent difficulty, and one should be able to prove that sealed answers restore the history-free semantics exactly. (iii) Federated reconciliation: pairwise commitment alignment and epidemic convergence are well studied; convergence whose target is ``agreement plus explicitly preserved disagreement'' among many sovereigns lacks a definition of termination.

— The Civilization Framework: Sovereign-Anchored Communication Between Personal Multi-Agent Systems  (2609.03425 - Liu, 3 Sep 2026) in Section 6, paragraph “Open formalizations” (following Section 5.1)