effects that have read to the end before they begin.
A live effect is causal: its output at time t can depend only on samples at or before t. For example, A live audio compressor waits for evidence a peak arrived, and only then can respond, always behind. This reliance on realtime is built into most audio effects (less so with video effects), when the digital medium does not always have to work this way. That single constraint is causality and it forces the threshold-and-attack dance and the running estimate, then the late action.
Hindcasts drops causality on purpose. These effects work with the full knowledge of the static file. This means effects can respond before, smooth backward, and fit their transfer curves to the entire piece at once.
This takes offline processing to its logical conclusion. Audio is often causal-by-default; image is acausal-by-default, and this enables audio processing to be able to feel like how image processing already works in digital space. Hindcasts rejects this inertia of hidden commitments to realtime.
Paracognitive granular delay. The control/source dual input is the shared interface for the whole suite — one signal donates structure or statistics, the other is the body it gets done to.
An un-live compressor. It reads the whole file, charts how long the sound spends at each depth, and lets you redraw the bottom. One RMS-window knob slides the regime continuously from waveshape to normalizing.
An acausal feedback field — framesmear unshackled. The time-arrow set free: trails that bloom ahead of an event, symmetric zero-phase wakes, and scrubbing that's instant because every frame already read end-to-end.