oversampling

Oversampling is a specialized form of sampling that operates at rates far above Nyquist. Remove sampling from the picture and oversampling cannot operate — it needs the sampling framework as its foundation. The which-came-first test confirms sampling (0.14) predates oversampling (0.99).

Oversampling is a specific kind of sampling — sampling above the Nyquist rate. Law 9: specific→general. The which-came-first test (epoch: oversampling 0.17, sampling 0.14) confirms sampling is older/more-foundational, consistent with the kind-of hierarchy.

Oversampling as a technique was formulated after sampling theory was established; it extends sampling by taking more samples than Nyquist requires. Which-came-first test per Law 7: sampling predates oversampling.

Oversampling is a specific kind of signal-sampling: the practice of sampling a continuous signal at a rate substantially above the minimum (Nyquist) rate required for reconstruction. Pinned sense (Law 11d): elevated-rate sampling of a continuous-time signal — not statistical oversampling of a dataset, and not the 'sampling' entry's carve of subset-of-population selection (which is why oversampling→sampling fails). Specific→general per Law 9; nearest kind is signal-sampling.

Sigma-delta modulation requires oversampling to operate — it runs the quantization clock at many times the Nyquist rate so that shaped noise energy can be filtered out digitally. Remove oversampling (run at exactly Nyquist) and the noise-shaping benefit disappears: quantization noise folds back into the signal band. Removal test per Law 8: without oversampling, sigma-delta modulation stops being useful and ceases to operate as a high-resolution encoding technique. Epoch 0.5 > 0.2, direction confirmed.