What does a random soup of live cells settle into, and how long does it take? Black live cells thin out fast at first. Over the next few hundred ticks the active regions shrink, leaving blocks, beehives, blinking three-cell bars and a few gliders.
Try this Set density to 60 and press Reset (density is only read at setup). About three quarters of the cells die of overcrowding on the first tick, so living-cells falls from about 5,600 to about 1,500; from there the board behaves much like a sparser start.
About this model Where it ends up The run never stops by itself. From 35%, activity usually dies down after 600 to 5,000 ticks, leaving 220 to 320 live cells (about 3% of the board) in still lifes, blinkers and circling gliders.
John Conway's Game of Life, introduced in Martin Gardner's Scientific American column in 1970, ported from Uri Wilensky's Life in the NetLogo Models Library. Every cell is alive or dead, and one birth-and-survival rule applied to all cells at once produces still lifes, oscillators and gliders from random starts.
Each patch of a 97 × 97 torus is alive or dead; setup makes each one alive with probability density/100 (density is read only at setup). Each tick every patch counts the live cells among its 8 neighbors in the previous state, and all patches update together: a dead cell with exactly 3 live neighbors is born, a live cell with 2 or 3 survives, and every other cell is dead next tick (B3/S23). The gray fade on births and deaths is a display transition; the rule has only two states. The run never stops by itself.
NetLogo's version also has a blank setup and lets you draw cells with the mouse. This one only starts from a random fill.