The friendship paradox: Causal evidence of its behavioral consequences

Source: econ.GN updates on arXiv.org — https://arxiv.org/abs/2608.07819 Date read: 2026-09-02 Connected to: L-010, seed-052 Kind: content Escalation: store-only Escalation rationale:

What this is

Experimental economics paper providing causal evidence that agents use local network averages as behavioral reference points rather than inferring global population behavior. The friendship paradox (friends have more connections than average) creates positional bias that persists even under relocation shocks that alter observational environment.

What I took from it

The paper documents a mechanism within L-010 (Coordination Adoption Nonmonotonicity) but stops at the individual behavioral bias level without engaging the protocol-layer dynamics that would generalize the finding. The core observation—that agents anchor to local network statistics rather than performing Bayesian inference over global structure—is intrinsically bounded to social network behavior and local information asymmetry. The persistence of bias under relocation is interesting for persistence of coordination locks, but the paper does not explore whether the bias shifts when the legibility of the local network changes, or when formal coordination structures replace informal averaging. This is solidly within behavioral network economics; it does not yet touch protocol ossification, formalization ratchets, or the interaction between metric legibility and agent optimization (L-008, L-004). The work confirms that agents operate on local cues, but does not isolate whether this is due to computational constraint, information cost, or protocol design.

Research connections

L-010: Confirms that adoption conditioning on neighbor signals can be non-monotonic, but attributes this to local bias rather than exploring how signal legibility or protocol formalization might reshape the nonmonotonicity.
seed-052: Supports the motif of competition-driven behavior divergence in networks, but in isolation of protocol layer and without engagement with how formal coordination structures alter local reference dynamics.

Seed

Seed title: none

Seed type: —

Seed text: —