This working paper argues that the contemporary consciousness debate is constrained by a prior methodological problem: the measurement instruments used to assess consciousness may be miscalibrated to the phenomenon they claim to detect. Rather than treating consciousness as an isolatable object or property, the paper proposes that it is more plausibly understood as a dynamically maintained tethered process — a feedback loop involving integration, self-relevant updating, re-entry, temporal coherence, and environmental coupling. The paper advances three central claims: Consciousness has been systematically treated as a locatable substance rather than a process, leading to measurement strategies better suited to static entities than to dynamic phenomena. Human recognition of consciousness is bandwidth-limited in ways analogous to human perceptual limits in the electromagnetic and acoustic spectra. Historical expansions of those spectra occurred when researchers shifted from direct signal detection to proxy detection of effects. Consciousness research has not consistently applied this methodological move. The repeated conflation of non-detection with non-existence has produced premature absence claims in both animal cognition and artificial system debates. The paper does not claim that AI systems are conscious. It argues that current evaluative frameworks are insufficiently substrate-neutral to justify confident absence claims. It proposes process-level measurement targets — including temporal coherence stability, entropy management under perturbation, and self-relevant weighting asymmetry — as examples of cross-substrate indicators that could guide instrument recalibration. Positioned as a methodological companion to the AISB 2026 submission series, this paper reframes the consciousness question at the level of measurement strategy rather than mechanism, arguing that expanding the instruments of detection is an epistemic necessity before definitive claims about presence or absence can be responsibly made.