Indexing the archive…
Your Universe of Digital Possibilities
A truth you cannot see, only noisy blips of it. The filter keeps a belief — a guess and how unsure it is, the ellipse — and every tick it predicts (the ellipse swells) then corrects against the latest blip (it snaps shut). How hard it jumps is the Kalman gain: the exact, optimal trade between what the model says and what the sensor saw, decided by nothing but their uncertainties. Signal pulled from noise, in real time.
The model pushes the belief forward — here a constant-velocity guess — and inflates the covariance by the process noise Q. With no correction this step alone is a forecast: the ellipse only grows.
How wrong the prediction was, and how much that surprise could be noise. S is the innovation’s covariance — the prediction’s uncertainty plus the sensor’s.
The heart of it: the gain is the prediction’s share of the total uncertainty. Confident prediction (small P⁻) or noisy sensor (large R) → K small, trust the model. Loose model (large Q) → K → 1, trust the blip.
The new estimate is the prediction nudged toward the blip by K of the innovation, and the covariance shrinks — the ellipse closing. Information only ever sharpens the belief.
Why it’s optimal: for Gaussians, posterior ∝ prior × likelihood is again Gaussian, and its mean is precisely the Kalman update. The filter is Bayes’ rule, closed-form, once per tick.
Run predict without correct, over and over, and P grows without bound: the covariance ellipse becomes a forecast cone. The Lens, blinded, is The Oracle.
This is the rack’s inference instrument — signal pulled from noise. It is the working half of the Perception Engine’s explain the present, and the literal mechanism behind a claim he keeps returning to: that perception itself is a prediction machine — the brain runs a model of the world and corrects it against the senses, trusting each in proportion to its reliability (predictive coding). It is the twin of The Oracle(INST·24): the Oracle is this filter’s predict-step run to the horizon with no measurements; the Lens is the same engine kept honest by a stream of them. And it shares a question with The Slit (INST·11) — what does a measurement do to a state? There, observing destroys (coherence collapses); here, observing creates (uncertainty collapses into knowledge). The same √t-growing uncertainty as The Walk(INST·19) underlies both — but a measurement is the thing that beats it back.