Does the statistical selection artifact proposal avoid faster-than-light signaling without hidden variables?

This is a sharp question about a specific loophole in the foundational puzzles of quantum mechanics.

The statistical selection artifact proposal—developed in detail across multiple papers in the archive—does attempt to avoid faster-than-light signaling *without* hidden variables by reframing Bell correlations as a kind of measurement bias [AG-2024.06-1630, AG-2024.04-2060]. The core idea is elegant: entanglement correlations arise from "collider bias" (a well-known statistical phenomenon where conditioning on a common effect can induce spurious correlations between independent causes) combined with a boundary constraint on the collider variable. Since no direct causal influence needs to cross a light cone, this sidesteps the usual tension between Bell nonlocality and relativity [AG-2024.06-1630].

Why this matters: If correct, it would dissolve a 50-year puzzle by showing Bell's correlations are *not* evidence of nonlocal influence at all—just statistical selection in disguise.

However, the archive also contains substantial criticism. One paper argues that all current Bell experiments fail to actually prove Bell's theorem due to subtle statistical issues in sample space definition [AG-2024.06-1550]—suggesting the loophole may run deeper than selection artifact alone. Another line of work [AG-2024.04-1862] tested whether transient deviations from QM (appearing only at short timescales) could explain correlations without signaling, and found no evidence over 24 meters.

The proposal explicitly claims to require *no hidden variables* [AG-2024.06-1630], relying instead on classical statistical reinterpretation. This distinguishes it from superluminal-hidden-variable models like Eberhard's [AG-2024.05-1868].

The archive does not contain empirical tests designed specifically to refute or confirm the selection artifact proposal itself.