Toward an Organizational Science of Multi-Agent LLM Systems: Decoupling Who, How, and Which Algorithm

Summary

The closest existing analogue to the paper's core design move: it argues LLM multi-agent frameworks conflate three logically distinct concerns - who is on the team (organization), how members align (coordination), and which algorithm fuses their work (collaboration protocol) - and builds IMACS to make them orthogonal, independently swappable layers. It runs controlled comparisons where organizational assignments vary while the collaboration protocol is held fixed, structurally the same identification logic as crossing organizational structure with aggregation rule. It also reports the winning organizational placement flips across model families, so organizational design cannot be hard-coded but must be revalidated per model binding.

Contribution

Queued — not yet read.

Method

Queued — not yet read.

Relevance to RISE

Surfaced by the 2026-08-29 literature sweep as a high-priority candidate for the Architecture as Epistemology (ISR) paper (angle: MAS topology / structure vs. aggregation rule; organization theory applied to AI agents). Verified against: Opened https://arxiv.org/abs/2607.25446 (arXiv abstract page, v1 28 Jul 2026)

Critique / open questions

Queued — not yet assessed.

Key quotes

Multi-agent frameworks built on large language models (LLMs) routinely entangle three logically distinct concerns: who is on the team (organization), how members align (coordination), and which algorithm fuses their work (collaboration protocol). (abstract)

IMACS (Intelligent Multi-Agent Collaboration System) separates the three into orthogonal, independently swappable layers. (abstract)