Non-Uniqueness of Remote Outcomes

• Non-Uniqueness of Remote Outcomes is a phenomenon where identical systems produce diverse macroscopic results due to inherent properties of the governing equations and structures.
• It manifests in quantum mechanics as branching in measurement outcomes, in PDEs as multiple blow-up profiles, and in mean-field games through equilibrium bifurcations.
• The concept challenges uniqueness theorems across disciplines, from Bayesian inversion and statistical chains to organizational studies, revealing intrinsic solution multiplicities.

Non-uniqueness of remote outcomes refers to the phenomenon where, despite identical systems and initial data, multiple distinct macroscopic or statistical outcomes can arise in the future or in distant parts of a system. This non-uniqueness is not due to measurement error, lack of knowledge, or randomness in initial configuration, but rather emerges from the intrinsic mathematical structure of the equations, physical processes, or inferential procedures involved. It manifests across applied mathematics, quantum foundations, PDE theory, statistical mechanics, Bayesian inference, and organizational sciences. In the technical literature, “remote outcomes” commonly denote results that are separated either in time (the far future, e.g. tail events), in space (distant regions), or in configuration space (distinct branches or solutions).

1. Non-Uniqueness in Quantum Theory and Bell Inequalities

Quantum mechanics exemplifies non-uniqueness of remote outcomes in the foundational context of Bell-type experiments. Standard derivations of Bell inequalities assume that measurement results on a remote (spacelike separated) system are uniquely determinable; in hidden-variable models, this corresponds to the “outcome independence” (OI) assumption. However, in the Everett (many-worlds) interpretation, measurements in one region result in branching—there is no unique remote outcome, but rather a full measure-valued assignment to all possible results. Saunders (Saunders, 17 Jan 2026) formalizes this as a “non-uniqueness-of-remote-outcomes loophole,” showing that the locality condition (LOC), which requires local probabilities to be insensitive to spacelike-separated actions, no longer entails factorizability or Bell’s inequality when outcome uniqueness is dropped. Instead, each measurement produces a superposition (branching) of outcomes:

$$\Psi \;\to\; \alpha\,\bigl|\text{Bob up};\text{Alice?}\bigr\rangle + \beta\,\bigl|\text{Bob down};\text{Alice?}\bigr\rangle.$$

In this framework, the non-uniqueness of remote outcomes is not merely a loophole but is essential to reconciling strong quantum correlations and observed Bell inequality violations with a strictly local (no action at a distance) evolution. The empirical non-uniqueness is further corroborated by work showing that no classical hidden-variable assignment can underpin all observable quantum statistics over a continuum of remote measurement settings (Wallman et al., 2011). Observers—even without aligning measurement bases—are forced into contextuality: there is no consistent, unique “instruction set” for remote outcomes compatible with all observed quantum nonlocality.

2. Non-Uniqueness in Dynamical Systems and PDEs

A prototypical class where non-uniqueness of remote outcomes arises is in nonlinear partial differential equations governing dispersive or statistical systems. In the context of energy-critical wave maps, Engelstein and Mendelson constructed the first example of non-uniqueness of blow-up “bubbles” for a Hamiltonian (dispersive) equation (Engelstein et al., 2020). For certain target geometries—a warped product with infinitely winding geodesics—the weak limit extracted at the singularity can depend on the particular sequence of rescalings chosen, generating a continuum of distinct harmonic map bubbles. This non-uniqueness is not an artifact but an intrinsic obstruction: in the presence of strong “winding” around the target, no canonical (unique) bubble exists at the blow-up core.

These results challenge the soliton resolution conjecture and the hope for canonical decomposition into unique solitary profiles. They demonstrate that, at criticality, remote (singular) outcomes—identified via weak limits—may be fundamentally nonunique, depending on the path taken through configuration space.

3. Mean-Field Games and Macroscopic Branching

In evolutive mean-field games (MFGs), non-uniqueness of remote outcomes arises as multiplicity of equilibrium branches, even for identical initial data. Bardi and Fischer (Bardi et al., 2017) explicitly construct classes where, for non-smooth (or sufficiently smooth but at large time horizon) Hamiltonians, the coupled forward–backward PDEs—governing both density evolution and optimal control—admit at least two distinct solution pairs $(v_1,m_1)$ and $(v_2,m_2)$ with differing macroscopic means (“remote outcomes” in time). The branching mechanism is driven by symmetry-breaking in the feedback structure: two constant strategies (e.g., $u^+\equiv b$ or $u^-\equiv a$) yield self-consistent solutions with strictly positive or negative population mean drift.

For short time-horizons or under strict Lasry–Lions monotonicity, uniqueness is restored. Thus, non-uniqueness of remote outcomes is an intrinsic dynamic phase transition—a bifurcation in the possible large-time macroscopic states, not dictated by noise or individual initialization, but by the collective structure and nonlinearity in the game.

4. Statistical Chains with Complete Connections

Remote outcome non-uniqueness arises in statistical chains with complete connections, particularly when past configurations have lingering, nonlocal influence. Berger–Hoffman–Sidoravicius (BHS) models (Dias et al., 2013) formalize this by constructing $g$-functions where local updates depend on large random subsets of the infinite past via a majority rule. If the majority rule is discontinuous at the origin (“pure majority”), infinitesimal biases in the far past are amplified, yielding multiple stationary chains with distinct tail $\sigma$-algebra events—i.e., different remote-past outcomes, even under identical dynamics and noise. The mathematical criterion is sharp: as soon as interaction strength decays slower than a critical threshold, uniqueness fails. Smoothing the majority rule (making it Lipschitz at the origin) always restores uniqueness, regardless of decay rate. This defines a rigorous phase boundary for remote outcome uniqueness in chains with memory.

5. Bayesian Inversion and Practical Non-Uniqueness

Inferential procedures that attempt to reconstruct remote (unobservable) causes from observed data frequently encounter non-uniqueness. In the context of earth pressure inversion for tunnel lining health monitoring (Tian et al., 2024), multiple candidate pressure fields yield (nearly) identical deformation responses, especially under deformation-only measurement regimes. The essential degeneracy arises from the “uniform shift” mode: pressure fields differing by a constant produce deformation differences smaller than measurement noise, remaining unresolvable. Bayesian inversion exposes this non-uniqueness as a flat direction in the posterior, with large associated variance. Only by augmenting the data with a single measurement of internal normal force (which directly responds to the uniform shift mode) can the non-uniqueness be sharply mitigated.

6. Organizational and Social Systems: Non-Unique Outcomes in Remote Work

In organizational and labor-market settings, non-uniqueness of remote outcomes is realized as non-monotonic and heterogeneous relationships between remote work frequency and career-related endpoints such as turnover risk and promotion. Large-scale empirical work (Lu et al., 11 Dec 2025) shows that distinct remote work patterns (e.g., low vs. high frequency) yield dramatically different probability distributions for outcomes, with both U-shaped and inverted-U-shaped dose-response forms. Subgroup-specific parameterizations produce further non-uniqueness across demographic axes (gender, role, leadership status). The existence of multiple optima—e.g., both low and high remote-work frequencies associated with increased turnover depending on parameter region—embodies practical non-uniqueness of remote organizational outcomes, robust even after accounting for observable confounders. Underlying mechanisms include both the direct functional relationship and the mediating structures of time-allocation and flexibility, themselves non-monotonic in remote status.

7. Conceptual Synthesis and Implications

Non-uniqueness of remote outcomes is a pervasive and structurally diverse phenomenon, emerging under:

• Indeterminacy (quantum branching, non-contextual hidden variables),
• Path dependence (PDE weak limits, mean-field branching),
• Strong memory or long-range coupling (statistical chains),
• Inferential degeneracy (inverse problems underdetermined by data), and
• Heterogeneous optimization landscapes (complex sociotechnical systems).

It signals a generic limitation of uniqueness theorems—arising not from lack of information or random noise, but from the underlying space of solutions or the geometry of the mappings from causes to observables. In several cases, such as quantum nonlocality or energy-critical PDEs, non-uniqueness is the mechanism by which fundamental constraints (e.g., locality, conservation) are reconciled with observed phenomena. Understanding, quantifying, and mitigating non-uniqueness is thus central to the theory and practice of inference, prediction, and control in high-dimensional, strongly coupled, or path-dependent systems.