What you're seeing
- Top — linear axis: the familiar picture. The hump is sharp, the rest of the universe looks empty. At your twin's position the curve appears pinned to the floor — apparent detachment.
- Bottom — logarithmic axis: the same numbers, plotted by their order of magnitude. Now the tail is a clean descending line (a Gaussian becomes a downward parabola in log-space) that crosses the whole universe and keeps going. It never reaches the floor because there is no floor.
- Follow the tail: the zoom marker rides out along the curve. Watch the readout: the exponent keeps growing more negative, the probability keeps shrinking — and stays resolutely above zero, at every distance you can name.
Honest caveats
Two, so the chapter can't be ambushed. First, a freely spreading position-space packet does fall off as a Gaussian and is genuinely nonzero everywhere — but a sharp-walled box would force exact zeros at its walls; "never zero" is the rule for smooth, physically realistic potentials, which is the relevant case. Second, "nonzero probability of being 600 light-years away" is a statement about the amplitude, not a claim that you can be detected there without the enormous energy that localisation costs — that cost is the subject of the Convergence page. Here the single, clean point stands: the amplitude is small, never zero, and the linear axis has been hiding it from every reader for a century.
Model notes: ψ(x) is a normalised Gaussian of adjustable width; |ψ|² is plotted linearly (top) and as log₁₀ (bottom). The readout reports the exact tail value at the marker using log-space arithmetic, so it stays accurate far below what floating point could represent directly.