Rule 30: a deterministic rule that keeps looking random

The rule is small; the trace is not

Rule 30 uses the same elementary setup as the other 255 rules: each new cell is computed from the three cells above it. Its rule table is encoded by 30, or 00011110 in binary. That means the eight neighborhoods 111, 110, 101, 100, 011, 010, 001, and 000 map to 0, 0, 0, 1, 1, 1, 1, and 0.

Start with one black cell and the rule draws a widening triangle. The first few rows are easy to reproduce by hand. After a few dozen rows, hand prediction stops being a reasonable way to read the system.

It is not random in the ordinary sense

Nothing random happens after the initial row is chosen. If you use the same first row and the same boundary condition, the same diagram appears every time. That is the point of the example: the output can look noisy without the update rule consulting a random source.

MathWorld describes Rule 30 as chaotic and notes its use in Wolfram Language random-number generation for large integers. That does not mean the center column has been proved random in every mathematical sense. The stronger claim would need proofs, not screenshots.

The center column is the hard part

In 2019, Wolfram announced three Rule 30 prize problems centered on the apparent randomness of the center column. The questions are concrete: does the center column always avoid eventual periodicity, do black and white occur equally often in the limit, and is there a shortcut for the nth center-column bit that beats direct simulation in a strong way?

That is a useful boundary for this article. The demo can show why the problem is tempting. It cannot settle the problem. A running canvas gives evidence of behavior; it is not a proof of randomness.

Why it belongs next to Conway Life

Conway Life teaches a two-dimensional version of the same warning: a rule can be short while the future remains hard to compress. Rule 30 strips that warning down to one dimension. There are no gliders, eaters, or guns to name. There is just a row, a rule table, and the trace it leaves.

That makes Rule 30 a good first stop after Life. It keeps the local-rule idea, removes the pattern-catalog vocabulary, and forces the reader to look at prediction itself.

What to watch in the demo

• Run the single-cell seed for at least 120 rows. The left boundary forms a visible nested pattern, but the middle does not settle into the same kind of simple triangle.
• Turn on a random seed and compare it with the single-cell run. The rule is unchanged; the trace changes because the initial row gives the rule different local neighborhoods to consume.
• Watch the center column marker. The apparent randomness of that column is the part Wolfram later turned into concrete prize problems.
• Switch briefly to Rule 90. That contrast is the fastest way to see why Rule 30 is not merely a triangle rule.