Kangaroo Puzzle #140

WINDOW - → - - TAME FRONT - - -

NO SCAN

STATUS - ACTIVE

STEPS/SEC-

waiting for data-

HERDS-

LIVE HERD POSITIONS

TAME START MID-

WILD START Q-

TAME FRONT-

WILD FRONT-

TRAIL TO TAME-

AVG/HERD-

P(SUCCESS) approx-

RECENT DISTINGUISHED POINTS (DP)

MASK p=- · - · total -

STEPS - / - ETA - ELAPSED 00h 00m 00s 0 Nodes Connected

KEY POSITION vs UNIFORM

MEAN / MEDIAN0.504 / 0.500

SD uniform 0.2890.271

CHI-SQUARE, 9 df6.76 · crit 16.92

KS D0.075 · crit 0.149

Keys #1–#135 land where a uniform draw would put them.

PUBKEY PREFIX 02 / 03

KEYS39 / 44

MEAN POSITION0.481 / 0.524

WELCH t−0.73 · 5% = 1.99

Even or odd y says nothing about where the key sits.

DRIFT WITH PUZZLE NUMBER

OLS SLOPE / PUZZLE9.7e−4

95% CI[−8.6e−4, 2.8e−3]

SPEARMAN / MANN-KENDALL0.121 · z 1.18

NEWEST 13, #75–#135mean 0.484 · MK z 2.01

ITS #140 95% PREDICTION[0.228, 1.000]

z 2.01 is one hit in ~18 tests, not a trend — and its band still spans most of the range.

WHAT NARROWING COSTS

WIDTH 1.00work x1.00 · best

WIDTH 0.70 / 0.50x1.20 / x1.41

WIDTH 0.25 / 0.10x2.00 / x3.16

A band hands back the sqrt saving while throwing away 1−f of the keys.

WINDOW - AT - CLEARED 0 OF RANGE - WINDOWS/HOUR - P(MISS) - EXPECTED TO HIT -

Each cell is a window of the same width, walked by the same engine in this tab, one after another from the start key — nothing jumps. CLEARED means the planned steps for that window were spent with no collision; a kangaroo walk cannot prove a window empty, so the per-window miss risk is printed instead.