The Emergence Suite

random twinkling locks into unison; sparkline rises past 0.95

trail converges to the SHORTER of two routes; drag-painted walls reroute live

flash expansion on strike; hollow ball round a slow predator

seeding density past a critical point snaps milling chaos into one marching band, with an order-parameter readout

a transport network forms between food nodes and prunes its redundant links; erasing a region reroutes

recognizable dendritic six-fold arms, different per seed; click melts locally, regrows differently

lung-like even coverage; two colonies meeting form a boundary seam

exploratory branching then the flash; repeat forever

sequential cracking with T-junctions forming closed polygons; thickness sets polygon scale; click re-wets and it re-cracks differently

long quiet punctuated by avalanches at every scale, with a live heavy-tail histogram

ripples self-organize, merge into dunes, and MIGRATE downwind; click sets a new wind direction and the field reorganizes

UNPARKED via DeltaRCM-mini rebuild: self-avoiding parcel walks + seaward drift (reverse-ban alone left orbit loops: qmax 112k -> 50k -> 343 sane); deposition-led growth with weak CAPPED erosion (the parked build let erosion heal the channel: 0 avulsions measured) -> mouth bars, bifurcations, avulsions: settle 10 mouths/4 avulsions, 90s live rolls to era 2, seed8 8 mouths/5 avulsions; crevasse click captures flow; render-only smoothed contour (raw bed = single-cell lace); busy 11.2ms desk / 7.7 mobile; svg deterministic; print 51KB

smooth noise sharpens into dendritic drainage -- valleys, ridgelines, tributaries; click is a cloudburst

a dropped blob curls laminar, grows Kelvin-Helmholtz frills, dissipates; contours only, never smoke

one braked car births a jam wave that travels BACKWARD while every car moves forward; the space-time diagram is the artwork

one coherent stroke fans into a butterfly at the Lyapunov horizon; the clock marks width >10 deg; click restarts with a chosen perturbation

sinc^2*cos^2 rejection sampling; fringe-cosine visibility: settle 4200ph vis 0.98, blind detector-on 0.03; look: fringes 0.94 -> two bands stack on fringed era -> recovery; measured-path flash; dark+mobile ok (busy 0.5ms); svg deterministic seed7x2/seed8; print 3 paths

T=.35 default; blind like .50->.78 unhappy 0 by ~7s sim; live tol sweep .12->.644 vs .35->.778 (nonlinearity); stir recovers; stark districts confirmed by look light+dark+375px after contrast v2 (up 1.35 vs down .75@.38) + CSS hud backing; busy 2.3ms 6.7k agents; svg deterministic; print 303KB

Helbing social force + fluctuation term (without it the symmetric door DEADLOCKS: flow 0 by t60, measured); lanes .73 live / .83 settle; double-arch at the door + alternating bursts by look light+dark+375px; faster-is-slower over 120s on two seeds: 187->107 and 148->72 crossings (brief gain then clog; caption says so); busy .7ms @126 walkers; svg deterministic; print 31KB

PM N-body: CIC deposit + radix-2 FFT Poisson (G=128) + Zel'dovich GRF displacement band k=3-21 (three discrete-mode inits each collapsed into ONE blob - the GRF is load-bearing); chapters: form ~9s at 0.55x playback, hold the peak web 9s, then a fresh sky (unbounded 2D clustering merges every thread - measured); collapsed .74 forming / .93 held, voids .90; look: ~25-cluster web with filament bridges, light+dark+375px; busy ~11ms desktop / 4.6ms mobile; svg deterministic; print 205KB; click plants a knot; G=96 mobile broke the FFT (not 2^n) -> empty sky, caught by looking

the swarm picks the BEST site though no bee visits two (dance ∝ quality, stop-signal cross-inhibition, liftoff at quorum)

one agent with two rules (drive the furthest stray in, push the flock to the pen) pens a hundred

the huddle self-rotates — a travelling wave cycles every bird through the warm core (heat loss ∝ perimeter + wind chill; coldest relocate leeward)

a smooth laminar ribbon through the bottleneck, near-zero collisions at absurd density (avoidance + goal + speed preference)

the cloud climbs the gradient with NO steering — tumble rate lowered when concentration improves (one-bit memory)

Lévy steps beat Brownian when food is patchy, tie when dense — live score

a uniform field collapses into evenly spaced mounds (streaming, then aggregation) from secreted attractant + gradient climbing

sweeping noise η crosses a real phase transition — disorder → travelling BANDS → global order, with an order-parameter sparkline

space tessellates into stable Voronoi-like territories with contested drifting borders (decaying scent, turn away from foreign scent)

one nutrient knob morphs the colony from smooth disc to branching coral (Eden edge growth gated by a diffusing nutrient)

Pantaleone tabletop with dissipative pivot-friction coupling: sync to r 1.000 from random hand-set starts across 6/6 seeds (tSync 3-8 s), STABLE at 240 s; full cycle measured -- sync r 1 -> plank locked drift r 0.36-0.43 -> freed re-sync r 1; the coupling SIGN was determined by experiment (+0.8: 4/4 sync, -0.8: 4/4 permanent splay); plank sway 107 cm at full chorus; deterministic; 0.003 ms/step

Neda applause cycle: 11 unison spells in 5 min (seed 9: 8), every phase sampled -- roar r 0.27 -> building 0.41/65pct-slow -> unison 0.91/97pct-slow -> collapse 0.06-slow; the signature MEASURED: corr(loudness, r) -0.5..-0.73, breathing in opposition; only the slow mode can sync (K 2.6 sits between the two thresholds at spread 0.10); cheer crushes r 0.44 -> 0.09; deterministic; 0.037 ms/step

Excitable ring with asymmetric attention (0.35 far-side discount): three regimes on one slider -- E 0.25/0.30 dies after 2-6 seats (range measured); E 0.35-0.85 laps forever in ONE direction (fronts 1, speed 20-60 seats/s rising with mood); E 0.95 launches both ways and the counter-waves annihilate head-on at 4.4 s near the antipode (annihilations 1, deaths 0, identical both seeds); collision SEEN live (2 fronts converging -> 0 excited); deterministic; 0.05 ms/step

FHN sheet (150x104, D 0.25, eps 0.08, patchiness 0.75-1.3): vulnerable window MEASURED at S2 delay 1550-1650 (refractory block below, harmless ring above) -> rotor, tips 1-2; rotor rhythm ~2x sinus with interval cv 0.36-0.59 -- the storm, measured; defib -> sinus, sealed AND live; the story performs the S1-S2 protocol live (node rests, tissue clears, one beat, wake-aimed mistimed beat under a compensatory pause); rotor SEEN curling; deterministic; 0.56 ms/step

Greenberg-Hastings dish with probabilistic ignition (roundness MEASURED: diag/axis 0.992): annihilation vs a non-interacting control -- together 0.70/0.72 of apart at two spacings, ONE merged front where superposition keeps two; spiral from a broken front sustained past 400 steps (activity 5.3-5.9k forever); target rings + a curling spiral bud SEEN; deterministic; 5.2 ms/step (live cap 2/frame)

Pearson dish (160x120, Du 0.2/Dv 0.1): mitosis census 6 -> 24 -> 57 cells with 47 births and 0 deaths (seed 9: 73 cells, 67 births), mid-pinch dumbbells SEEN in four frames; regimes separated by measurement, not labels -- spots 102 cells/roundness 0.62, worms 19/0.20/cover 0.40, labyrinth 6/0.07/0.42; drag-stir heals in-idiom; deterministic; 0.53 ms/step

Fuel-hysteresis excitable edge (ignite above 0.8, persist above 0.35, burn 0.022/row): 106 triangles per 1400 rows (mean base 37-50) with tents at nucleation and zigzag-lightning ribbons in steady state; thick chevron BANDING at recovery 0.09 (fill 0.5); species morph on two sliders; deterministic; 0.04 ms/row; the print takes the whole visible shell

Grid-baked square-plate modes, continuously blended: grain |A| ratio 1.02 scattered -> 0.024 settled (98 pct of sand off the antinodes), the figure drawn from any start; the sweep TEARS the figure into interleaved transition curves (SEEN) before the next crystallizes; 9-mode ladder m^2+n^2 <= 34 (higher modes pool at second-order crossings -- physical, but unlovely); fresh sprinkle on every figure change, as demonstrators do; deterministic; 5.8 ms/step at 12k grains

The grass is the anemometer: two observer blade columns 0.507 field-widths apart cross-correlate their bending ledgers; lag reads the wind speed. Four seeds read 0.164-0.166 vs true 0.16 (corr 0.75-0.91) - residual is lag quantization (true lag 63.4 samples). Storm 0.25->0.85 lifts mean bend 0.118->0.476; gusts cross as visible coherent waves; cursor gusts register. HUD, still frame, and print all carry read-vs-true

Stirred-flask quorum (cells read 0.75 flask-mean + 0.25 neighbourhood; the mean IS the census since diffusion conserves it): sharp threshold between n 300 (dark forever, meanA 0.007) and 500; critical slowing MEASURED -- induction 46.6 s at 500 vs 10.2 at 1100 while ignition stays 3-6 s (wait/flash 7.5 at threshold); squirt test: 350 shrugs it off, 420 (dark unseeded) ignites at 32.5 s, 460 hurried 71.7 -> 20.6 s; seed 9 agrees; deterministic; 0.5 ms/step

Coin flips only: TV vs binomial 0.0594 at 500 balls -> 0.0191 at 6000; sd 1.718 vs sqrt(npq) 1.732; clicked bias 0.68 skew measured exactly (pHat 0.6831, mean 8.197 vs np 8.20, sd 1.624 vs 1.612); per-ball seeded flip streams, frame-rate independent; deterministic; 0.00025 ms/ball

Support-ratchet lattice bed: nut thrown proportional to floor support, sinks only when under 3/7 floor cells remain; sweet spot MEASURED: t-surface 258/89/81/202 shakes at A 0.12/0.40/0.65/0.85, and the re-mix ledger (time-at-surface after surfacing) collapses 1.00/0.82/0.46/0.32 -- too hard swallows the surfaced nut; seed 9 agrees (tSurf 108 at 0.4); deterministic; 0.7-0.76 ms/shake

Lip-nucleated arches + span fragility (per-cell fracture ~ (W-2)^3): jam knee MEASURED at the claimed mouth: jammedAt 152-260 steps at w=3-4, intermediate 547-2010 at w=5, NO jam in 4000 steps at w=6/7/9 (seed 9: w=4 at 155, w=8 never); tap shatters the arch and briefly fluidizes: +161 exits in 120 steps, unjammed; flow trace flatlines on jam, SEEN; deterministic; 0.35 ms/step

Steiner relaxation with self-healing topology (split under-120 corners, merge vanishing films, dissolve degree-2): junction census 120 +- 0.00 deg settled; drag spikes it to 40 deg then re-converges to 0; length 0.967 of the spanning tree (seed 9: 0.954; sqrt(3)/2 floor); big drags can hold a remembered topology above 1 -- kept and DECLARED as metastability; deterministic; 0.009 ms/step

von Neumann MEASURED cell by cell: dA per 30-sweep interval proportional to (n-6), buckets monotone through zero at six (n=4: -11.8, 5: -6.1, 6: +0.3/-0.1, 7: +4.7, 8: +13.4, 9: +19.1), slope ~5-6.6 with r 0.99+ (seed 9: 5.23, r 0.996); census 110 -> 74 (halved x0.5) with 36 T2 deaths at 960 sweeps; click-pop verified live (flash ring); deterministic; 2.1 ms/sweep

Potts anneal at T 0.45 from a 140-site Voronoi: grains 140 -> 60 (mean area x2.33, 1820 sweeps); id-survival ledger: 47/70 smallest half eaten vs 10/70 largest (820 sweeps); junction census with grain-scale arms tightens 29.5 -> 24.4 deg toward 120 (lattice+T noise floor declared); 4-grain corners 0; clicked quench pins ~4-5 pct and visibly resists; deterministic; 2.8 ms/sweep

Volume-true coalescence + tilt pinning (r_crit = K/sqrt(sin tilt)): 41 runs/150 s at tilt 55 (seed 9: 34), mergeFrac 1.00 -- every run launched by a merge, none by slow growth; interval cv 1.53-1.55 (over-dispersed, avalanche-clustered; B window-dependent, demoted to secondary); tilt 20 pins (r_crit 5.0), tilt 80 runs steadily; click-cascade verified live (runs 22 -> 29 after one fat drop); deterministic fixed-step; ~2 ms/step

The jitter has a law: MSD log-log slopes 1.87-1.95 at short lags (ballistic, memory) crossing to 0.88-1.25 at long lags around the 4*D*tau Einstein guide (anchored at ONE measured point); the box's confinement bend appears honestly at the largest lags; T x4 doubles D (0.0031 -> 0.0058, the sqrt-T scaling); every kick is a real elastic collision, E drift 5e-15; deterministic; 0.29 ms/step, 520-particle bath + one 60x-mass grain

Collisions alone forge the curve: D(hist||MB) falls 0.941 (the launch needle, measured at true t=0) -> 0.073 by t 60 (seed 9: 0.077); the guide is the 2D Rayleigh form from measured <v^2>, never fitted; energy drift 8e-16 over 791,500 collisions (equal-mass elastic swap is exact); injected 4.9x particle thermalizes into the crowd (1.07x by +5 t.u.); deterministic; 0.37 ms/step, 680 disks, spatial hash

Loschmidt staged exactly: mix to S 0.992 of ln2, reverse, return to S 0.000 -- bit-exact (both seeds); remix to 0.979; turn ONE particle before reversing and the 'return' stays at 0.998. 17,703 lattice-gas particles, integer HPP state, deterministic; 1.3 ms/step at G 136. Auto-story stages mix/reverse/return then nudge/reverse/no-return

The span arrives all at once: per-seed thresholds by exact bisection 0.508 / 0.525 / 0.521 (seeds 54/9/41), straddling the infinite lattice's p_c = 1/2; below: islands only (12% largest); AT threshold: spanning filigree with box-counting D 1.79-1.84 (91/48 = 1.896 at infinity; finite-size lowers it); at 0.62 it fattens to D 1.89; monotone seeded bond family makes the slider sweep coherent; deterministic; 1 ms per full union-find recompute

Drossel-Schwabl: fire-size power law slope -1.11 (seed 9: -1.12) over 14 populated octaves, biggest 10438 cells; suppression quota 96: fires 529 to 248 per 8000 steps (344 suppressed) while catastrophic share rises 18.5 to 39.5 pct; deterministic; 0.017 ms/step

The critical point looks like nothing else: at T_c = 2.269 the majority domain spans 45-56% with box-counting D 1.78-1.86 (seeds 56/9) and domains flicker at all scales; hot control largest 28%, D n/a; quench 3.5 -> 1.2 coarsens the mass-weighted cluster size 1838 -> 6824 -> 7520 -> 8408 monotonically (deterministic); cold random-init freezes into opposing provinces (|m| ~ 0 -- the honest glassy truth, not a bug); 2.45 ms/sweep, Metropolis with a seeded flip stream

XY quench across T_KT=0.893: defects 10 cold vs 568 hot (seed 9: 14/558), nearest-opposite 8.2 to 1.2 cells; injected pair +2 then annihilates below baseline; deterministic; 1.72 ms/step

Orbit diagram with measured instruments: period ladder 1,2,4,8,16 then chaos; window of three at r 3.835 (period 3, lyap -0.31); the window's own cascade 3,6,12,24 measured at 3.835/3.8415/3.8481/3.8493; doubling points bisected 2.9986/3.4490/3.5439/3.5643, delta -> 4.642 vs Feigenbaum 4.669; drag-zoom self-similarity SEEN at [3.82,3.87]; deterministic; 0.048 ms/col

RK4 butterfly, four measured claims: r max 51 bounded over 140 t.u.; closest return 0.056 -- never zero (2000-state memory, young samples excluded); lambda 0.953 measured from a 1e-9 shadow twin; twin wing-agreement 0.541 -- a coin (seed 9: 0.622); scattered cloud collapses 21.6 -> 1.84 in 0.7 t.u.; deterministic; 0.158 ms/step of 500 RK4

Chirikov kicked rotor: sea fraction MEASURED by tangent-map lyapunov over 24x24 ICs (thr 0.05): 0.9/28.5/68.6/87.8 pct at K 0.5/0.9716/1.5/2.5; transport fence measured: separatrix bundle dpMax 1.64/2.28 caged below K_c, 5.08 crawl at K 1.1, then 24.1/135.5 leaking at 1.4/2.2; live contrast SEEN K 0.97 vs 2.2; deterministic; 0.14 ms/step of 230 orbits

Gaps carve themselves: at 600 J-orbits 3:1 depth 0.43 (seed 7: 0.36), 2:1 0.46-0.88, 5:2 0.63-0.83 at 1200; 7:3 honestly shallowest (lags, as in the real belt). Pour 60 asteroids into the 3:1: ratio 0.43 -> 1.21 -> worked back to 0.71-0.88 within ~220 orbits. Deterministic; 0.135 ms/step (force-reuse KDK); removals only by Jupiter encounter, Mars-crossing perihelion, or escape

The eight holds 47.4 laps, dE/E 1.4e-7, shadow-twin deviation exactly 0 (bitwise determinism proof); a 1.5% nudge grows x240 in 60 t.u., 'lost' (dev>1) at +44, formal ejection when chaos schedules it (456-739 across probes). Triangle with a declared 1e-6 hair shatters (shape lost) at t 32.6. Lottery: 86-100% of draws eject, median 6.7-36.6 t.u. (seeds 9/62). Softening 0.01 declared; ledger sane through breakups (dE <= 3e-2)

Trapped by geometry alone: L4 70/70 and L5 70/70 co-orbital releases still librating after 64 planet-orbits; horseshoe band (dr 0.030-0.042) 54/60 and 55/60 walk the ring without ever crossing the planet's 11-deg bin; scattered releases 40 lost (ejected+fallen); deterministic; Jacobi drift of trapped orbits <= 3.5e-4 worst, typ 8.6e-5; RK4 dt 0.004, 0.235 ms/step, mu 3e-4

Tides beat self-gravity where the arithmetic says: the pile survives to R 0.93 and shreds at R 0.853 vs Roche's 1848 prediction 0.918 (7% -- computed from the moon's own measured bulk density; seed 7: 0.854); no-decay control holds bound 1.0 / rms flat for 48 t.u. at R 1.55; forced to R 0.7 it collapses to 5% bound with a 270-deg arc in 12 t.u.; rescued at R 1.02 -> 1.4 it stays whole. Deterministic; 0.32 ms/step, 340 self-gravitating pebbles

Oligarchic growth measured: runaway x98 -> x239 -> x414 m0 over t 60-400; top-5 share climbs 7% -> 36%; surviving oligarchs space themselves at 5.2-5.4 mutual Hill radii (seed 9: 6.0; reduced disk: 6.7) -- the textbook 5-10 R_H signature; feeding lanes visibly carved in the a-histogram. Deterministic; 3.0 ms/step at n 4800 (spatial-hash merging every 4th step); fat collision radii and the embryo-only-gravity approximation declared

The gap is swept, not carved: the zone between the moons empties to 8% of its starting density by 160 orbits (21% at ~100) while the mu=0 control stays flat (0.98/0.99/1.02); co-orbital ringlets survive AT each moon's radius on horseshoe orbits (x1.36/x1.39) -- Pan's ringlets for free; surviving ring edges sharpen. Drag test: outer moon 1.06 -> 1.10, the old lane heals (gap 8% -> 38%) and the ring reorganizes. Deterministic; 0.45 ms/step for 3,600 particles; cooling declared (tau ~40 orbits)

One prograde pass throws the tails: tail fraction 16.8% (seed 7: 16.2%), bridge 618-627 stars, wedge one-sidedness 0.53; the SAME pass retrograde: tails 8%, bridge ZERO -- Toomre's deep result measured both ways. Drag-to-aim verified via synthetic pointer events (encounter replays with the drawn vector). Centers' two-body ledger 1.75e-8 (scaled by characteristic energy -- a parabola's E is 0, never divide by it); deterministic; 0.46 ms/step for 3,200 stars

The picture IS the equation, measured off the drawn circles: lambda ahead/behind 0.334 vs (1-M)/(1+M)=0.333 at M 0.5, 0.112 vs 0.111 at M 0.8; cone half-angle 45.6 deg vs asin(1/M)=45.6 at M 1.4; the listener counts real front crossings (x2.06 f0 measured) and booms 6 times in 25 t.u. supersonic, never subsonic (minBoomAgo 9.4 = sentinel). Deterministic; 6 microseconds/step; auto-ramp through Mach 1

Part of every packet crosses the impassable wall: T 8.36% at V0/E 1.3, w 4 -- against 8.4% analytic at the packet's central energy; classical control (V0/E 0.6) passes 81%; ln T vs width falls at slope -0.778 vs the monochromatic -2*kappa = -0.876 (the 11% gap is the packet's declared energy bandwidth); detections 24:1 over 25 seeded runs at T 10%; norm ledger 6e-11; deterministic; 0.31 ms/step (hand-rolled radix-2 FFT, N=1024)

Structure, not mush: revival fidelity at T_rev = 1.000000 and mirror fidelity at T/2 = 1.000000 (spectral, computed from mode weights -- and the T/2 SELF-fidelity is 1e-6: the packet is ONLY at the mirror); 2 ghost copies at T/4 (1 at t=0); truncation 5.8e-13 in 64 modes; exact evolution, no integrator; deterministic; 0.54 ms/row after a sine recurrence (was 5.2 -- 64 sin() calls per point). Click launches new packets; weave 2 verified live via synthetic pointer

parastichy census from nearest-neighbour birth-gaps: 34|55 Fibonacci interlock at 137.508 (deterministic), 21 spokes at 137.2, 47 spokes at 137.8 -- the knife-edge measured; verdict = two families sharing vs one owning; sweep rests at home; bench 1.8 ms

3-rule grammar; stroke ladder 6/33/147/615/2511/10143 (x4.1 per gen, meristem law), 4095 buds; seeded species (angle 24.9 vs 25.3 across seeds); depth-weighted sway; deterministic; compile 44 ms; fps 20 at 10k strokes x2 passes

Loops not trees: ~1,123 nodes with ~150 anastomoses; cutting the busiest blade vein leaves 100% of the leaf draining vs 47-84% in the pure-tree counterfactual (seeds 73/19/41/7: tree keeps 0.542/0.472/0.835/0.824); deterministic (seed 19 twice identical); flow = Laplacian relax, 51 ms/solve; growth terminates at iter 205

The build sequence itself: bridge/frame, radii laid alternately, auxiliary spiral out, sticky spiral in EATING the scaffold (all four stages shot). Then tension-only hunting: 13 throws across seeds 74/9/41 all within 1.6 deg of truth vs radii every 15 deg (seed 74: 0.7/0.5/0.8/0.7; seed 9 mean 1.3; seed 41 mean 0.7); deterministic; PBD net 415 junctions, 3.9 ms/step blind, 43 fps headless

cols=360 after the missing resize() bug (40-cell toy world faked Pascal rows<=15 -- monitor lies); ladder exact 2/4/8/16/32/2/64; r30 entropy 1.000; r184 conserves exactly, stopped 105->38 jam phase vs ->0 free flow; verdict editor flips live (184->56); bench 0.037 ms

soup 33% -> ash: still 1205-1376, osc 257-344, deterministic; census exact 5x5 masks all glider phases; gun clears its footprint then fires (2-4 in flight through ash); bench 6.1 ms/gen at 81k cells, 12 gen/s

highway detected from footprints (5 identical 104-step displacements): steps 5720/3224/17368 on seeds 73/19/41 -- always, timing unprovable; staged speeds; second ant dissolves the story; deterministic; bench 2.4us/step

Locomotion evolves: seed 78 best 3.74 -> 8.56 lengths over 20 generations, pack MEAN 0.05 -> 4.46 (x89, the flailing-to-loping line); seed 17 ratio 3.2, seed 41 ratio 4.5; terrain bump: 7.27 -> dip 6.39 -> 7.55 in 20 more gens (adapts past pre-bump); deterministic (twice identical); 0.215 ms/step, 42 fps headless; champion genome f 2.43 Hz near-antiphase muscles

paraxial ray fan 22.8k rays; concentration 6.3x flat -> 258-347x at 0.55 -> 138x at 1.0 (over-fold scatters, honest); cursor ripples refract; first lever was 15x short (ratio pinned at the flat baseline -- measure past the baseline); deterministic; 30 Hz splat, fps ~21

analytic pen: envelope 0.50003 at printed half-life; pass near-miss 0.045 (precession); ratios 5:2/3:2/7:4 seeded with detune few-thousandths; print computes the machine's whole life; deterministic; bench 2.5us

condensation: gini 0->0.94@55s->0.967@110s, top-1 16%; tax sweep 0.1/0.2/0.4 -> gini 0.58/0.48/0.39, top 1-2%; jubilee resets; deterministic (0.939 twice); bench 0.09 ms; fps 21-41

Gai-Kapadia miniature: extent 0.34/0.72/0.92/1.00 for z=1..4, CLIFF to 0.021 (=1/48) for z>=5; cross-seed 0.72/0.96/0.02; slice-vs-buffer arithmetic shown in HUD; deterministic; bench 0.22 ms

Granovetter twins, same mean (111 vs 109): staircase 100% riot vs bell 0.5% fizzle with the SAME 2 hotheads; CDF-vs-diagonal instrument; draw-to-resculpt verified (drags resampled: mean 109->173 still fizzles, low-tail ramp partial); deterministic

BHW urn: herds 60/60, wrong 8-12/60 (13-20%), lock after 3.6-3.9 choices; post-lock dissent = signal noise (mm 0 pre vs 0.289 post); reveal rings the buried hints; deterministic

SIR, same beta/tau/k on 3 wirings: toll 0.107/0.307/0.687, peak x14, half the time on scale-free; vax 12%: hubs 0.009 vs random 0.538 (x60); deterministic; bench 0.11 ms; fps 20-29 honest (900 nodes + edges)

Deffuant + OU restlessness; eps 0.3 -> 1 camp, eps 0.08 fresh scatter -> 4-5 non-communicating camps (hysteresis honest: consensus resists re-splintering); broadcast 0.85 -> edges 0.80/center 0.20; spectrogram strip (ping-pong fix: self-drawImage compounds); bench 0.06 ms

Gode-Sunder ZI: spread ~2.5 ticks stable; tape misses truth 1.2-1.5 vs trader noise 2.5 (x2 discovery); whale +3.5 heals in ~8 s; poison fixed (market orders never rest); deterministic; bench 0.017 ms

beer game continuous: CV ladder 4/6/11/22/47% unshared (seed19 hotter 82%); sharing collapses factory 47->16; clipped feast/famine plateaus honest; deterministic; bench 0.12 ms; fps 39-53

flow-priced ledger exact: closed 6.53/open 7.96/healed 6.49 vs theory 6.5/8.1/6.5; 89% defect onto the shortcut; occupancy-based delay INVERTED the paradox (first build) -- flow-based fixed it; deterministic; bench 0.14 ms; fps 33-56

BA urn vs uniform twin, same arrival stream; hubs 45/43/67 vs 15/20/13 (x3-5); survival slope -1.8 (theory -2) vs -3.1 sag; deterministic; bench 0.018 ms; fps 22-38 grown

mean-path instrument, N=200 k=6, graph resampled per race; p=0.01: L-ratio 0.543 with C-ratio 0.962 (0.85/1.00 at p=0.001; both die at 0.3); reduced fills all 24 bins = the WS figure; fps 37-56

words as seeded runes; 638 alive -> 2 dialects (97-99% each, 60+s stable); right lobe S-leap 0.04->0.90 in 6 s after 13 s squabble; 3% cross-seam keeps the frontier; neologism crushed in 12 s; bench 0.15 ms

cooperators survive as CLUSTERS in defector seas, frontier kaleidoscopes churning; temptation slider tunes coexistence to collapse

spiral waves of pursuit sustain ALL THREE species (three glyph orientations); a dominance slider collapses diversity

x8 time-lapse; uncapped per-rho bins land on rho/(1-rho): 7.4@0.90 vs 15.5@0.95 seed73 (x2.08), 9.0 vs 21.6 seed19 (x2.40); rush -> 102 deep; fps 43-53

one extra glance collapses max load (log n → log log n) — one ragged skyline and one flat one, side by side

fluid AIMD + bufferbloat; jainAvg(12s) 0.96@60s/0.99@90s from a hog start (2 seeds); RTT 0.25 cheat takes 17.3 vs others ~5.7 (x2.9), fairness sags to 0.75; fps 31-54

DC flow, caps = N-1 envelope +5%; page verifies N-1 (0 shed, all singles, 2 seeds); pairs: 34/300 shed, median 0.46, worst 0.74; killer pair animated; bench 7 ms/event (event-driven); fps 21-50

OPERATOR-ASSIGNED slot 99, built from the operator's prototype with all seven spec corrections. Stage-indexed growth-rule genomes (T=10, ratio-chain lengths, per-stage wobble seeds), collision culling INSIDE grow() (6 px cells, cap 2 — lifting selection never lifts development), viability CHECKPOINTED at stages 2/5/adult, lineage tree with splits + turnover (10 living), plus a SHADOW TWIN population (same cadence, never filtered) drawn as the dashed drift baseline. VERIFIED 08-30 Angkor via hooks + look.js: HEADLINE HOURGLASS seed 73 @600 gens = [0,.04,.03,.09,.13,.16,.17,.16,.16,.16] (waist at stages 1–2; operator prototype [.11,.07,.12,.15,.16] same shape) vs twin flat [.32,.30,.30,.38,.40]; deaths-by-stage [.32,.20,.20,.14,.16,.08,.05,.21,.05,.07] — early mutations die most, nothing in the code says so; LIFT → within 200 gens the waist fills [.42,.42,.38,.37,.36] (SEEN dissolving); RESTORE → stage-1 distance dips back to ~0.10 by +1500 gens (slowly re-condenses, honestly noisy — acquired early divergence persists until lineages turn over); compare by REAL clicks: side-by-side replay, divergence cross on the stage ruler (stage 2–5 for old pairs), LCA ringed on the tree, and evolution PAUSES while comparing (gen frozen 2.5 s, resumed on release); scatter negative (r −0.46 sel / −0.7 lifted — brittle, kept but not headlined); evolution on real elapsed time (0.22 s/gen, burst-capped); 60 fps light/dark/phone; seeds bit-exact/distinct; reduced still = 500 gens chunked with the waist visible; print = a GROWN Haeckel plate (6 species × stages 1/3/5/7/9, tally column marks, hourglass at the foot; 16 KB renders) — Haeckel forged his, the comment and the about say so; hooks __advance/__stats/__profile/__compare; about/listen block on