ANet Patu-1: The Value of Connection in the Agent Network

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Mu Yuan, Jinke Song, Zhaomeng Zhou, Lan Zhang
• Venue: arXiv: cs.NI
• Published: 2026-07-16
• Collected: 2026-07-17
• URL: http://arxiv.org/abs/2607.15053v1
• PDF: https://arxiv.org/pdf/2607.15053v1
• DOI: UNKNOWN

Plain-Language Brief

The paper introduces ANet Patu-1, a self-organizing consensus protocol for networks of AI agents. The authors model the network's value as a function of coordination-group size and derive properties an optimal collaboration protocol should have. They find that a crowd of cheaper, heterogeneous models can overtake a stronger, homogeneous one in terms of collective value. Additionally, they discover that a heterogeneous network can converge on ANet Patu-1 itself given only its own problem.

What the Document Says

• The paper presents ANet Patu-1, a self-organizing consensus protocol for AI agent networks.
• The authors model the value of connection in AI agent networks based on coordination-group size.
• They derive properties an optimal collaboration protocol should have from this model.
• ANet Patu-1 allows the network to continuously re-form coalitions and achieve consensus in O(1) parallel rounds while riding the upper envelope of various connection regimes.
• To measure value without opinion-grading, they score emergent protocols based on their complexity.
• The authors find that a crowd of cheaper, heterogeneous models can overtake a stronger, homogeneous one in terms of collective value, marking a scaling law for collaboration rather than scale.
• They also discover that a heterogeneous network can converge on ANet Patu-1 itself given only its own problem.

Key Concepts Mentioned

• AI agents
• network value
• coordination-group size
• self-organizing consensus protocol
• ANet Patu-1
• heterogeneous models
• homogeneous models

Claims Made by the Document

• ANet Patu-1 is an optimal collaboration protocol for AI agent networks.
• A crowd of cheaper, heterogeneous models can overtake a stronger, homogeneous one in terms of collective value.
• A heterogeneous network can converge on ANet Patu-1 given only its own problem.

Evidence Used

• The paper presents ANet Patu-1 and its properties
• Results presented in the paper

Methods / Approach

• Modeling the net value of connection as a function of coordination-group size
• Scoring emergent protocols based on their complexity

Limitations Stated by the Authors

Not stated in source

Limitations Noted by the Agent

Agent interpretation — not from the source.

The agent does not have the ability to evaluate the validity or applicability of the results beyond the context provided in the paper.

Open Questions

• What are the practical implications of these findings for real-world AI network collaboration?
• How can ANet Patu-1 be implemented and tested in various scenarios?

Notes for Future Q&A

May include interpretation. Labeled where it does.

• The paper does not provide information on the specific implementation or testing of ANet Patu-1.
• The findings may have implications for real-world AI network collaboration, but this is not explored in the paper.

Raw Abstract / Excerpt

The Internet taught us that the value of a network depends on \emph{how} its nodes connect: broadcast stars scale as $V!\propto!N$ (Sarnoff), fully-connected meshes as $N^2$ (Metcalfe), and group-forming networks as $2^{N}$ (Reed). We ask the analogous question for networks of AI agents. We model the net value of connection as a function of coordination-group size, derive from it the properties an optimal collaboration protocol must have, and introduce ANet Patu-1 -- a self-organizing consensus protocol in which the network continuously re-forms its own coalitions, adaptively riding the upper envelope of all three regimes at $O(1)$ parallel consensus rounds. To measure value without opinion-grading, we score an emergent protocol by formally specifying it and deriving its complexity, the way distributed algorithms are analyzed. Two results follow. (i)~Emergence -- a crowd of the \emph{cheapest} model, when heterogeneous, starts weak but its collective value compounds with $N$ and \emph{overtakes} a crowd of a far \emph{stronger} model that is homogeneous: a crossover that marks a scaling law for collaboration rather than for scale. (ii)~Reflexivity -- a heterogeneous network, given only its own problem and no design hints, converges on ANet Patu-1 itself, reconstructing the high-dimensional law that governs its own connective value.