ExplorationsA lookup table of eight entries, one black cell, and a centre column nobody can predict

Rule 30: Order That Makes Chaos

Open problemcellular automatapseudorandomnesscomputationcurious

In an elementary cellular automaton, each cell of a new row is black or white according to a lookup of the three cells above it, eight cases in all, so there are 256 rules. Rule 30, started from one black cell, makes stripes down its left side and what looks like noise down its right; this page grows it row by row and counts the black cells in its centre column, which are an observation here and not a proof of randomness.

Current view: The rows

The rows themselves, grown one under another, with the centre column marked.

The rows · Rule 30 · one black cell64 rows

Rule 30 · 64 of 64 rows · centre column 35 black of 64 (54.7%)

Every row so far, with the centre column counted

The first 64 rows of Rule 30, from one black cell. Each row is 129 cells wide; the middle 33 cells are written out, 1 for black and 0 for white, with the centre cell in the middle of the string.
Row	Centre cell	Centre black so far	Black cells in the row	Middle 33 cells
0	1	1	1	`000000000000000010000000000000000`
1	1	2	3	`000000000000000111000000000000000`
2	0	2	3	`000000000000001100100000000000000`
3	1	3	6	`000000000000011011110000000000000`
4	1	4	4	`000000000000110010001000000000000`
5	1	5	9	`000000000001101111011100000000000`
6	0	5	5	`000000000011001000010010000000000`
7	0	5	12	`000000000110111100111111000000000`
8	1	6	7	`000000001100100011100000100000000`
9	1	7	12	`000000011011110110010001110000000`
10	0	7	11	`000000110010000101111011001000000`
11	0	7	14	`000001101111001101000010111100000`
12	0	7	12	`000011001000111001100110100010000`
13	1	8	19	`000110111101100111011100110111000`
14	0	8	13	`001100100001011100010011100100100`
15	1	9	22	`011011110011010010111110011111110`
16	1	10	15	`110010001110011110100001110000001`
17	0	10	19	`101111011001110000110011001000011`
18	0	10	20	`001000010111001001101110111100110`
19	1	11	24	`111100110100111111001000100011101`
20	0	11	21	`100011100111100000111101110110001`
21	0	11	23	`110110011100010001100001000101011`
22	1	12	23	`000101110010111011010011101101010`
23	1	13	28	`001101001110100010011110001001011`
24	1	14	26	`111001111000110111110001011111010`
25	0	14	27	`100111000101100100001011010000011`
26	1	15	26	`011100101101011110011010011000110`
27	0	15	33	`010011101001010001110011110101101`
28	1	16	30	`011110001111011011001110000101001`
29	1	17	34	`110001011000010010111001001101111`
30	1	18	31	`001011010100111110100111111001000`
31	0	18	39	`111010010111100000111100000111101`
32	0	18	26	`100011110100010001100010001100001`
33	1	19	39	`010110000110111011010111011010011`
34	1	20	29	`110101001100100010010100010011110`
35	1	21	46	`100101111011110111110110111110001`
36	0	21	32	`111101000010000100000100100001011`
37	1	22	44	`100001100111001110001111110011010`
38	0	22	38	`010011011100111001011000001110011`
39	1	23	45	`011110010011100111010100011001110`
40	0	23	47	`010001111110011100010110110111001`
41	1	24	41	`011011000001110010110100100100111`
42	1	25	45	`110010100011001110100111111111100`
43	0	25	49	`101110110110111000111100000000011`
44	0	25	38	`001000100100100101100010000000110`
45	0	25	55	`011101111111111101010111000001101`
46	0	25	42	`110001000000000001010100100011001`
47	1	26	51	`001011100000000011010111110110111`
48	1	27	44	`111010010000000110010100000100100`
49	0	27	53	`000011111000001101110110001111111`
50	0	27	43	`000110000100011001000101011000000`
51	1	28	59	`001101001110110111101101010100000`
52	0	28	52	`111001111000100100001001010110000`
53	1	29	60	`100111000101111110011111010101001`
54	0	29	49	`111100101101000001110000010101111`
55	1	30	65	`100011101001100011001000110101000`
56	1	31	57	`010110001111010110111101100101101`
57	0	31	60	`110101011000010100100001011101001`
58	1	32	56	`100101010100110111110011010001111`
59	0	32	69	`011101010111100100001110011011000`
60	1	33	61	`110001010100011110011001110010100`
61	0	33	70	`001011010110110001110111001110111`
62	1	34	59	`011010010100101011000100111000100`
63	1	35	78	`110011110111101010101111100101110`

a black cell: a cell the rule made 1
a black centre cell: a black cell in the centre column, the one the page counts
the centre column: the outlined column straight down from the starting cell
a white cell: a cell the rule made 0, left as paper

One square is one cell, and each row is the one above it passed through the rule. A black square is 1 and a pale one is 0. The centre column is outlined and its black cells carry a vermilion mark, so the column the page counts is the one you can see. The rule tiles show the lookup the rows were made with: each tile shows three cells above and the one cell the rule gives them. From one black cell, look at the left edge for stripes and at the right for noise; from a random row, look instead for what changes from one row to the next.

Rule 30 from one black cell, 64 rows. The left edge settles into stripes and the right side looks irregular. 35 of the first 64 cells of the centre column are black (54.7%). This is a count over these rows and not a statement about the column going on for ever.

Rows: 64 / 64
Centre cell: black
Centre black so far: 35 of 64
Share black: 54.7%

How this page is laid out

Make a prediction

Change the assumption

What this assumes: The disorder in the picture was made by the rule.
What breaks it: The first row is random. Even Rule 90, which only adds two neighbours mod 2, then draws a picture that looks as disordered as Rule 30 does, and the disorder was there from the first row.

Where it holds

Rule 30 from one black cell. The only thing put in is a single cell, so any irregularity in the rows below it was produced by the table.

The rule

Rule numberAny of the 256 rules, from 0 to 255.

00011110

The eight tiles are the rule: three cells above, and the cell they give. Read the outputs left to right as binary and you get the rule number. Press a tile to flip what it gives.

The start

First row

Nothing but the rule can make a mess from this.

Rows

Step by step

The Step, Run and Reset buttons are under the figure. Speed sets how fast Run goes.

Speed

Rows drawn

64 of 64 rows on the page. Every row is worked out before anything is drawn, so moving this changes only how much is shown.

Reproducible run

Advanced parameters

Random seedUsed only by the random first row. It is always written into the address bar.

Cases worth seeing

Try this

Start Rule 30 from one black cell and press Step three times. For each new row, find the three cells above each cell and check the answer against the matching tile.

1/3

What is going on

Each explanation begins with a worked example and follows the same operation through intuition, formal statements and proofs. Later sections distinguish what is established from questions that remain open, so you can follow the level of detail useful to you.

Worked by handRule 30 from one black cell, three rows by hand. Each row is nine cells with the starting cell in the middle, 1 for black and 0 for white. A new cell is read off the three cells above it: 111 gives 0, 110 gives 0, 101 gives 0, 100 gives 1, 011 gives 1, 010 gives 1, 001 gives 1, 000 gives 0.

000010000001 → 1, 010 → 1, 100 → 1, and every other neighbourhood here is 000 → 0000111000The one black cell is seen by three cells: the one to its left (which sees it on its right), itself, and the one to its right (which sees it on its left). All three turn black.
000111000001 → 1, 011 → 1, 111 → 0, 110 → 0, 100 → 1001100100The middle cell sees 111 and turns white, which is the first hole in the picture.
001100100001 → 1, 011 → 1, 110 → 0, 100 → 1, 001 → 1, 010 → 1, 100 → 1011011110Row 3, and the centre cell is black again. Reading the middle cell down the rows so far gives 1, 1, 0, 1, the start of the column the page keeps counting.

The laboratory does exactly this for every cell of every row, a few thousand times over, and draws the 1s as black squares instead of printing them.

Read the middle cell of each row down the page: 1, 1, 0, 1. Nothing was chosen at random, and that short column is already hard to describe in a sentence.

What you are seeing

A row of cells, a table of eight entries, and a new row underneath

A row of cells, each black or white. To make the next row, look at each cell and the two next to it, which is three cells in a row above it. There are only eight ways three black-or-white cells can be, so a rule is a table of eight entries saying what each one becomes. Write the eight answers as eight binary digits and you have a number between 0 and 255, which is the rule's name. That gives 256 rules, and the eight tiles in the control rail are the table for whichever one is on show.

Rule 30 is the table 0, 0, 0, 1, 1, 1, 1, 0, and the picture starts from one black cell. The left side settles into regular stripes, and the right side does not. The page marks the centre column and counts its black cells as the rows go down, because that column is the interesting one: it is the thing Wolfram proposed as a source of random numbers.

Why it starts to make sense

Nothing is chosen, so where does the disorder come from?

There is no dice here, and no hidden input. Run the same rule from the same start a thousand times and you get the same rows a thousand times. The page is the opposite of the chaos game: that one feeds randomness in and a fixed shape comes out, this one feeds in one black cell and a fixed table and what comes out looks, in places, random.

The tiles make the mechanism concrete. Press Step and watch one row become the next, and you can check every cell against a tile. What you cannot do is look at the table and say what row 1,000 will look like. The table is simple, the rows are the table applied over and over, and the second does not follow from the first by any shortcut anyone has found.

The key idea

The same machinery makes stripes, noise and a computer

Rule 90 uses the same eight-entry table format and gives a different kind of picture. Its table amounts to "add the left and right cells, mod 2", and from one black cell that draws nested triangles: Pascal's triangle with the even numbers left white. It is the same gasket the chaos game draws, made with no randomness at all.

Rule 110 gives something else again, shapes that travel and collide on a patterned background, and it has been proved that with the right background it can carry out any computation a computer can. So a table of eight entries is enough to make regular patterns, apparent noise, or a computer, and the rule numbers do not say which. That is the point of the side-by-side view.

The random first row is there to show a trap. Give Rule 90 a random first row and the picture looks as disordered as Rule 30 from one cell. But Rule 90 only adds, so the disorder was in the first row. Disorder in a picture is only a finding about the rule when the start was plain.

Formal statement

A local rule on a row of two-valued cells

Write a row as a sequence of cells $x_{i} \in {0, 1}$ , with the neighbourhood of cell $i$ being the three-cell word $x_{i - 1} x_{i} x_{i + 1}$ . A rule number $R$ with binary digits $b_{7} b_{6} \dots b_{0}$ gives the next row by

x_{i}^{'} = b_{4 x_{i - 1} + 2 x_{i} + x_{i + 1}},

so the neighbourhood read as a binary number from 0 to 7 picks which digit to use. There are $2^{8} = 256$ choices of $R$ . For Rule 30, $R = 0001111 0_{2}$ , and the table is the same as

x_{i}^{'} = x_{i - 1} \oplus (x_{i} \lor x_{i + 1}) .

For Rule 90, $R = 0101101 0_{2}$ and $x_{i}^{'} = x_{i - 1} \oplus x_{i + 1}$ , which is linear over the integers mod 2.

Proof

Why Rule 90 is Pascal's triangle, and why Rule 30 is not a shortcut away

The formula for Rule 30 and the table for Rule 90. Both are eight comparisons, listed in the ledger, and the unit tests make the same eight for the code.

Rule 90 and the binomial coefficients. From one black cell, row $0$ is a single $1$ . Every later cell is the sum, mod 2, of two cells above it. Pascal's rule says that the coefficient $(k n + 1)$ is $(k - 1 n) + (k n)$ , which is the same sum, so by induction on $n$ the cell $k$ places along row $n$ is $(k n) mod 2$ . A coefficient is odd exactly when $k$ and $n - k$ share no binary digit 1 (Kummer, Lucas), which is why the triangles are nested and why the same shape turns up in the chaos game's grid.

Rule 30 has no such proof. The centre column of Rule 30 has no known formula, and the page claims none of the properties it is famous for. The ledger gives the three open questions Wolfram posed in 2019: does the column ever become periodic, is it black exactly half the time on average, and can the nth cell be computed with much less work than running the rule. The first is checked here only for periods up to 2,000 within the first 20,000 rows. Wolfram's announcement says the column has been checked not to become periodic for a billion steps, and that is a report cited here, not a result of this page.

Limits and further questions

What a finite run shows and what it does not

The black count in the centre column is an observation. After 128 rows it sits close to half, and after 10,000 it is 5,032, and neither number is evidence that the share tends to a half. A fair coin would also wander about that far, and that is the honest sense in which the column looks random: the counts are consistent with a coin, not proof of it.

"Passes statistical tests" is the same kind of statement. Wolfram reports that the centre column passes standard tests and that Mathematica's random integer function was once built on it, using a ring a few hundred cells wide. The page repeats that as a report. A sequence that passes some tests might fail others, and being hard to predict from the rule is different from being safe to use where an adversary is looking for a pattern.

The cone snail Conus textile has a pigment pattern that resembles the patterns cellular automata such as Rule 30 make, and Wolfram has noted the likeness between shell patterns and cellular automata. This page shows no shell and does not claim the shell is made by this rule. A likeness is a reason to ask the question, and it has not been shown to be the answer.

What is not settled is almost everything about the middle column: whether it is ever periodic, whether it is balanced, whether a shortcut exists. Those are the three open questions, and the reason a rule with eight entries is a good place to meet them is that they can be stated to anyone who can read the tiles.

What is actually established

Every statement on this page, with its status, its exact scope, and the date that status was last checked.

Proved

A rule is a choice of output for each of the 8 three-cell neighbourhoods, so there are exactly 256 rules, and Rule 30 makes a new cell equal to left XOR (centre OR right).

Scope: Every one of the 8 neighbourhoods was checked, so this is a finite proof by cases. Rule 30 is the binary number 00011110, giving 0, 0, 0, 1, 1, 1, 1, 0 for the neighbourhoods 111, 110, 101, 100, 011, 010, 001, 000.
Why: Two choices for each of eight neighbourhoods is 2⁸ = 256. For the formula, list left, centre and right and compare. Left 0: the cell is black exactly when centre OR right is black, which matches 001, 010, 011 giving 1 and 000 giving 0. Left 1: the cell is black exactly when both centre and right are white, which matches 100 giving 1 and 101, 110, 111 giving 0. The test suite makes the same eight comparisons for the kernel.
Status checked: 24 October 2026
Sources: Statistical mechanics of cellular automata citation unresolved

Observed in this run

In the first 400 rows from one black cell, each of the eight diagonals nearest the left edge repeats with period at most 2 from row 8 onwards.

Scope: One run: Rule 30, one black cell, rows 0 to 399, diagonals 0 to 7 counted in from the left edge, and period two checked from row 8 to the end. It is a statement about those rows and those eight diagonals. The page makes no claim that the left side stays regular further out or further in.
Why: The left edge itself is proved to be black in every row, because a cell with a white left neighbour and a black right neighbour is black under Rule 30 (001 gives 1) and the left edge only ever has that neighbourhood. The stripes in the picture are the next few diagonals, and those are what the check above is about.
Status checked: 24 October 2026
Sources: A New Kind of Science citation unresolved

Observed in this run

The first 10,000 cells of the Rule 30 centre column contain 5,032 black cells, and the first 20,000 contain 10,120.

Scope: Rule 30 from one black cell, the starting cell included, counted exactly. Black shares of 50.32% and 50.6%. It is a count over a finite prefix and says nothing about the average over all rows.
Why: Close to half is what a fair coin would give, and the difference between 5,032 and 5,000 is about what a fair coin would be expected to be off by. That is the sense in which the column looks random. It does not show that the share tends to a half, and a column can be balanced for twenty thousand rows and not for all of them.
Status checked: 24 October 2026
Sources: Announcing the Rule 30 Prizes citation unresolved

Exhaustively checked

From row 2,000 to row 19,999 the Rule 30 centre column has no period of 2,000 or fewer.

Scope: Rule 30 from one black cell. For every period p from 1 to 2,000, the cells from row 2,000 to row 19,999 contain a pair c(i), c(i + p) that differ, so no period up to 2,000 fits that stretch. It says nothing about longer periods or later rows, and a period that began after row 20,000 is not excluded.
Why: A direct computation over the first 20,000 rows with whole-number arithmetic and nothing sampled. It is a much weaker statement than the first prize question, which asks about every period and every start. Wolfram’s announcement reports that the column has been checked not to become periodic in its first billion steps, which is a larger computation than this page runs and has not been repeated here.
Computation (exact): Rows 0 to 19,999 of the Rule 30 centre column from one black cell, periods 1 to 2,000, starting from row 2,000. Algorithm elementary-ca-cone-v1.
Status checked: 24 October 2026
Sources: Announcing the Rule 30 Prizes citation unresolved

Observed in this run

Wolfram reports that the centre column of Rule 30 passes standard statistical tests for randomness and that Mathematica’s random-integer function generated its bits from it.

Scope: A report by the rule’s inventor, in the 1986 paper and in A New Kind of Science. This page reruns no test suite and has not read the tests themselves. The generator in Mathematica ran the automaton on a ring a few hundred cells wide, not on an endless row, so it is a wrapped relative of the column drawn here. What the library uses today is not claimed.
Why: Passing tests means that a fixed list of statistical checks did not tell the column from fair coin flips. It is not a proof that no check could, and it is not a certificate that the sequence is safe for cryptography. The page uses it as a pointer to why the column is interesting, and its own evidence is the black-cell count in the claim above.
Status checked: 24 October 2026
Sources: Random sequence generation by cellular automata citation unresolved; A New Kind of Science citation unresolved

Open problem

Whether the centre column of Rule 30 ever becomes periodic is not known.

Scope: Wolfram’s first Rule 30 prize question, posed in 2019: does the centre column always remain non-periodic? No period has been found in the stretch computed, and no proof either way has been found.
Why: Computation can exclude a period up to a given length in a given stretch, as the check on this page does for periods up to 2,000, but it cannot exclude a period longer than anything run.
Status checked: 24 October 2026
Sources: Announcing the Rule 30 Prizes citation unresolved

Open problem

Whether black and white cells occur equally often, on average, down the centre column is not known.

Scope: Wolfram’s second Rule 30 prize question: does each colour occur on average equally often in the centre column? The share of black cells in the prefixes counted here is close to a half and that is all that is known.
Why: The question is about the limit of the share as the number of rows goes to infinity. Being near a half after twenty thousand rows is consistent with the limit being a half and with it being something else.
Status checked: 24 October 2026
Sources: Announcing the Rule 30 Prizes citation unresolved

Open problem

Whether the nth cell of the centre column can be computed with less than about n steps of work is not known.

Scope: Wolfram’s third Rule 30 prize question, worded as whether computing the nth cell of the centre column requires at least O(n) computational effort. Running the rule directly takes work growing like n² for the first n rows, as the kernel here does with its cone of live cells.
Why: A shortcut would mean the column is in some sense predictable without running it. Nobody has found one and nobody has shown that none exists.
Status checked: 24 October 2026
Sources: Announcing the Rule 30 Prizes citation unresolved

Proved

Rule 90 makes each new cell the sum, mod 2, of its left and right neighbours above, and it is linear: the rows from a sum of two first rows are the sums of the rows from each.

Scope: Every first row, every number of rows, on a ring or an endless row. The neighbourhoods 111, 101, 010 and 000 give 0 and 110, 100, 011 and 001 give 1, which is the same as left + right mod 2 with the centre ignored.
Why: Rule 90 is the binary number 01011010. The check is the eight cases again. Addition mod 2 is linear, so adding two first rows cell by cell adds their later rows cell by cell. A random first row therefore carries its randomness straight down, row after row, and the rule adds none of its own.
Status checked: 24 October 2026
Sources: Statistical mechanics of cellular automata citation unresolved

Proved

From one black cell, Rule 90 makes row n equal to the binomial coefficients C(n, k) mod 2, so the picture is Pascal’s triangle with the even numbers left white, the Sierpiński triangle.

Scope: Every row n of the endless row from one black cell. The cell k places in from the left of row n is black exactly when C(n, k) is odd. The ring on this page is wide enough that the wrap does not touch the rows it draws.
Why: The cell at position x in row n + 1 is the sum mod 2 of the cells at x − 1 and x + 1 in row n. Pascal’s rule C(n + 1, k) = C(n, k − 1) + C(n, k) is the same sum with the cells lined up differently, so by induction on n the two triangles agree mod 2. The unit test checks it against the standard test for odd binomials, that C(n, k) is odd exactly when k AND (n − k) = 0, up to row 255.
Status checked: 24 October 2026
Sources: Geometry of binomial coefficients citation unresolved

Proved

Rule 110 is computationally universal: set up suitably, it can carry out any computation a computer can.

Scope: Cook’s published theorem, cited here and not re-proved or checked by this page. The set-up is an endless row with a particular repeating pattern on each side, not a single black cell on a white row. Whether Rule 110 is universal from a finite pattern on a plain white background is a different question and the page makes no claim about it.
Why: The proof works by building the behaviour of a simpler machine (a cyclic tag system) out of the moving shapes of Rule 110. The slowdown involved is large, so this says nothing about it being a practical way to compute anything. For the laboratory it is the third kind of behaviour: not stripes, not noise, but structure that can interact.
Status checked: 24 October 2026
Sources: Universality in elementary cellular automata citation unresolved

Heuristic

The pigment pattern on the shell of the cone snail Conus textile resembles the patterns made by cellular automata such as Rule 30, and Wolfram has remarked on the likeness between shell patterns and cellular automata.

Scope: A visual resemblance noted by Wolfram. It is not a test: nobody here has shown that the shell is made by Rule 30 or by any cellular automaton, and shell patterns have been modelled in other ways. The page shows no photograph and makes no claim about how the animal produces its pigment.
Why: The pattern on the shell is laid down along the growing edge, one row of pigment cells at a time, which is why a row-by-row model is natural to try. Resemblance is a reason to look, not a result.
Status checked: 24 October 2026
Sources: A New Kind of Science citation unresolved

Sources

Review notes show which bibliographic details Mathomaly has checked and which remain unresolved. Checking a publication record does not independently verify its proof.

Stephen Wolfram, Statistical mechanics of cellular automata, Reviews of Modern Physics 55 (3), pp. 601–644, 1983. Link
Where the one-dimensional two-colour, three-neighbour automata were studied systematically, and where the numbering of the 256 rules by the bits of their lookup table comes from. Cited for the setting and the numbering, not for any number on this page.
Citation details remain unresolved after review by Mathomaly.
Bibliographic review: 2026-10-24
AI-assisted bibliographic audit. Evidence type: unresolved.
The Crossref record for this DOI was retrieved and gives the title, journal, volume 55, issue 3, pages 601–644 and July 1983. The paper itself was not read, so what is attributed to it is unchecked and the citation stays flagged.
- Review evidence 1 for Statistical mechanics of cellular automata
Stephen Wolfram, Random sequence generation by cellular automata, Advances in Applied Mathematics 7 (2), pp. 123–169, 1986. Link
Proposes the centre column of Rule 30 as a source of random bits and reports statistical tests on it. Cited for that report; the page reruns no test suite of its own.
Citation details remain unresolved after review by Mathomaly.
Bibliographic review: 2026-10-24
AI-assisted bibliographic audit. Evidence type: unresolved.
The Crossref record for this DOI was retrieved and gives the title, journal, volume 7, issue 2, pages 123–169 and June 1986. The tests described in the paper were not read, so the citation stays flagged.
- Review evidence 1 for Random sequence generation by cellular automata
Stephen Wolfram, A New Kind of Science, Wolfram Media, 2002. Link
The book describes Rule 30 and its centre column, says Mathematica’s Random[Integer] generated its bits from the centre cell of the Rule 30 automaton (wrapped round in a region a few hundred cells wide), and remarks on the likeness between such patterns and the pigment patterns of some shells. Cited for what Wolfram reports; none of it is tested here.
Citation details remain unresolved after review by Mathomaly.
Bibliographic review: 2026-10-24
AI-assisted bibliographic audit. Evidence type: unresolved.
The author’s online edition was fetched for the page on Random[Integer] and the sentences about the centre cell and the wrap-around were seen. The shell remark was found only through a search result and the book’s printed page numbers were not checked, so the citation stays flagged.
- Review evidence 1 for A New Kind of Science
Stephen Wolfram, Announcing the Rule 30 Prizes, Stephen Wolfram Writings, October 2019, 2019. Link
States the three prize questions about the centre column of Rule 30 and records that the column has been checked not to become periodic in its first billion steps. Cited for the wording of the open problems.
Citation details remain unresolved after review by Mathomaly.
Bibliographic review: 2026-10-24
AI-assisted bibliographic audit. Evidence type: unresolved.
The announcement was fetched and the three questions and the billion-step remark were read there. It is the author’s own post rather than a publisher or library record, and whether any question has been answered since was not checked beyond a web search that found none, so the citation stays flagged.
- Review evidence 1 for Announcing the Rule 30 Prizes
Matthew Cook, Universality in elementary cellular automata, Complex Systems 15 (1), pp. 1–40, 2004. Link
The published proof that Rule 110 can carry out any computation a computer can, as Wolfram had conjectured. The proof works on an infinite row with a particular repeating background, which the page says. Cited for the theorem; the proof was not checked here.
Citation details remain unresolved after review by Mathomaly.
Bibliographic review: 2026-10-24
AI-assisted bibliographic audit. Evidence type: unresolved.
The Crossref record for this DOI was retrieved and gives the title, Complex Systems volume 15, issue 1, pages 1–40, March 2004, Matthew Cook as author, and an abstract that states the aim of proving Wolfram’s conjecture that Rule 110 is universal. The proof itself was not opened, so the citation stays flagged.
- Review evidence 1 for Universality in elementary cellular automata
Stephen Wolfram, Geometry of binomial coefficients, American Mathematical Monthly 91 (9), pp. 566–571, 1984. Link
Pascal’s triangle reduced modulo 2, its nested triangles, and their production by a cellular automaton. The page proves the Rule 90 case itself in a line, so this is cited for the history only.
Citation details remain unresolved after review by Mathomaly.
Bibliographic review: 2026-10-24
AI-assisted bibliographic audit. Evidence type: unresolved.
A Crossref search returned this record: title, The American Mathematical Monthly, volume 91, pages 566–571, November 1984. (The JSTOR form of the DOI, 10.2307/2323743, only redirected, so the Taylor and Francis one is cited.) The article was not read, so the citation stays flagged.
- Review evidence 1 for Geometry of binomial coefficients

Connected by how they work, not by sharing a topic label.

Interactive laboratory

Make a prediction

Change the assumption

The rule

The start

Step by step

Reproducible run

Cases worth seeing

Try this

What is going on

What you are seeing

Why it starts to make sense

The key idea

Formal statement

Proof

Limits and further questions

What is actually established

Sources

Related by mechanism