Yamagishi Complexity Torus LabYamagishi Lab

[Paper 623] The Look-and-Say Sequence Grows by About 1.3036 Each Step, Whatever the Seed

Abstract

Consider the sequence whose next term is made by reading the previous one aloud as “how many of which digit”. Starting from $1$ it runs $1,11,21,1211,111221,\ldots$ (the look-and-say sequence). How its length grows is counted for several seeds (first terms). No new theorem or law is claimed.

First, the length grows by about $1.3036$ each step——from the seed $1$, term $10$ has $26$ digits, term $30$ has $5808$ and term $51$ has $1520986$. The ratio averaged over ten terms (the length of term $51$ divided by that of term $41$, to the power $1/10$) is $\mathbf{1.303610}$ (Section 2).

Second, and this is the core. The ratio does not depend on the seed——changing the seed to $3$, $123$ or $1113$, the ratio averaged over ten terms is $1.303747$, $1.303682$ and $1.303475$, all within $\mathbf{0.000170}$ of Conway’s constant $1.303577$. The lengths themselves differ by seed (from $1520986$ to $4070770$ at term $51$), but the growth is the same (Section 3).

Third, the ratio from one term to the next wobbles——the ratio of term $51$ to term $50$ ranges over the four seeds from $1.302897$ to $1.303918$, a spread of $0.001021$. Averaging over ten terms shrinks the spread to $0.000272$ (Section 3).

Fourth, only the seed $22$ never moves——$22$ reads as “two $2$s” and returns to $22$; its length stays $2$ for every term (Section 4).

Fifth, from the seed $1$ only $1$, $2$ and $3$ appear——checking up to term $40$, no digit $4$ or higher ever appeared, because four equal digits in a row never occur (Section 4).

Keywords: look-and-say sequence, Conway’s constant, growth rate, splitting theorem, integer sequence.

Scope (what is not claimed)

No new theorem or law is claimed——that the length of the look-and-say sequence grows at the same constant whatever the seed (Conway, $1987$) and the proof of the decomposition theorem behind it (Ekhad and Zeilberger, $1997$) are known. Where the constant comes from is not treated——Conway’s constant is known as a root of a polynomial of degree $71$, but neither that polynomial nor the division of the sequence into $92$ “elements” is treated here; only the known value $1.303577$ is compared with counted lengths. Relation to earlier papers: Paper 161 showed that “fixed point” theorems are different things according to what each demands and delivers——the seed $22$ is the fixed point of reading aloud, and no other seed goes there; they all keep growing at the same rate. Paper 622 showed that the repeated sorting and subtracting of digits stops or circles depending on the number of digits——repeated reading aloud never stops, and only the growth is shared regardless of the seed. What is added is counting the lengths up to term $51$ for four seeds, setting out in numbers the wobble of the term-to-term ratio and how the ten-term ratio lines up, and placing the separator on whether the seed is $22$.

1. Notation and the Question

Setting: reading a string of digits from the left, each run of equal digits is written out as “count, digit” to form the next term. Let $L_n$ be the number of digits of term $n$ (the seed being term $0$).

Question: how does the length grow, and what changes and what does not when the seed changes?

2. Counting Lengths

Computation 2.1: for five seeds, terms up to $51$ were made and their lengths counted.

seedterm $10$term $30$term $51$
$1$$26$$5808$$1520986$
$3$$30$$6784$$1766402$
$123$$74$$15570$$4070770$
$1113$$56$$11420$$3002354$
$22$$2$$2$$2$

Observation 2.2 the length depends on the seed. At term $51$ the seed $123$ is about $2.7$ times as long as the seed $1$——whether every seed grows in the same way must be seen in the ratio, not the length itself.

3. The Ratio

Computation 3.1: the term-to-term ratio $L_{51}/L_{50}$ and the ratio averaged over ten terms, $(L_{51}/L_{41})^{1/10}$, were found.

seed$L_{51}/L_{50}$$(L_{51}/L_{41})^{1/10}$
$1$$1.303730$$\mathbf{1.303610}$
$3$$1.303089$$1.303747$
$123$$1.303918$$1.303682$
$1113$$1.302897$$1.303475$

The spread (largest minus smallest) of the term-to-term ratio was $0.001021$, and that of the ten-term ratio $0.000272$. All four ten-term ratios were within $\mathbf{0.000170}$ of Conway’s constant $1.303577$.

Observation 3.2 this is the core: the ratio does not depend on the seed. Changing the seed among $1$, $3$, $123$ and $1113$ changes the length by up to about $2.7$ times, but the ten-term ratios all line up at the same constant——the seed decides the starting length, not the growth.

Observation 3.3 the term-to-term ratio wobbles. Reading aloud depends on the fine shape of the digit string, so the ratio of one term to the next moves up and down by about $0.001$——averaged over ten terms, the wobble shrinks to about a quarter and the constant comes into view.

4. The Separator

Observation 4.1 the separator is whether the seed is $22$. $22$ reads as “two $2$s” and returns to $22$; its length stays $2$ for every term——every other seed keeps growing at the same rate. Only one seed stands still.

Computation 4.2: checking the digits appearing up to term $40$ from the seed $1$, only $1$, $2$ and $3$ occurred.

Observation 4.3 digits of $4$ or more do not appear unless the seed has them. Reading aloud writes a $4$ only when four equal digits are in a row, which does not happen starting from $1$——the limited set of digits also supports the growth being independent of the seed.

Observation 4.4 beside Papers 161 and 622. Paper 161 showed that fixed-point theorems differ by what each demands and delivers, and Paper 622 that repeatedly sorting and subtracting digits stops or circles depending on the number of digits——in repeated reading aloud, the fixed point $22$ exists but attracts no other seed. The repetition never stops, and only the growth lines up.

Observation 4.5 what is not said. Where Conway’s constant comes from (the polynomial of degree $71$, the $92$ elements) is not treated——what is shown, from lengths counted for four seeds, is that the ratio lines up at the same constant, that it wobbles from term to term, and that only the seed $22$ does not move.

5. Roll Call and Closing

  1. term $51$ from the seed $1$ has $1520986$ digits (Computation 2.1): about $1.3036$ times per step.
  2. lengths differ by up to about $2.7$ times by seed (Observation 2.2): different starts.
  3. ten-term ratio $\mathbf{1.303610}$ (seed $1$) (Computation 3.1): close to the constant.
  4. all four seeds within $\mathbf{0.000170}$ of the constant (Computation 3.1): two routes.
  5. the ratio does not depend on the seed (Observation 3.2): the core.
  6. the term-to-term ratio spreads by $0.001021$ (Observation 3.3): $0.000272$ over ten terms.
  7. the seed $22$ never moves (Observation 4.1): the separator.
  8. only $1$, $2$, $3$ from the seed $1$ (Computation 4.2): up to term $40$.
  9. four equal digits never occur (Observation 4.3): a limited set of digits.
  10. a fixed point that attracts nothing (Observation 4.4): Papers 161 and 622.

Closing: Making each term by reading the previous one aloud, the length grows by about $1.3036$ each step. That ratio does not depend on the seed——changing the seed among $1$, $3$, $123$ and $1113$ changes the length by up to about $2.7$ times, but the ten-term ratios all lined up within $0.000170$ of Conway’s constant $1.303577$. The term-to-term ratio wobbles by about $0.001$. The separator is whether the seed is $22$——only $22$ returns to itself and stands still; every other seed keeps growing at the same rate. Placed among the earlier papers——beside the fixed points of Paper 161 and the digit repetition of Paper 622, here a fixed point exists yet attracts nothing, and only the growth lines up. To be honest——where the constant comes from is not entered; lengths were counted and compared with the known value.

References

  1. Conway, J. H. (1987). The weird and wonderful chemistry of audioactive decay. In T. M. Cover & B. Gopinath (Eds.), Open Problems in Communication and Computation. Springer.
  2. Ekhad, S. B., & Zeilberger, D. (1997). Proof of Conway’s lost cosmological theorem. Electronic Research Announcements of the American Mathematical Society.
  3. Yamagishi, Y. (2026). A Fixed Point Is Not One Theorem [Paper 161]. Zenodo. 10.5281/zenodo.22092530
  4. Yamagishi, Y. (2026). Sorting Digits and Subtracting Reaches 6174 for Four Digits, and Reaches No Fixed Point for Five [Paper 622]. Zenodo. 10.5281/zenodo.23014520

Reference

Yamagishi, Y. (2026). The Look-and-Say Sequence Grows by About 1.3036 Each Step, Whatever the Seed [Paper 623]. YOI: https://www.yamagishiyuuki.com/yoi/623
DOI: 10.5281/zenodo.23014522
ORCID: 0009-0006-8587-3450

PDF (English) · PDF (Japanese) · Date proof of the English PDF (OpenTimestamps .ots) · SHA256SUMS

The record page for this paper (in Japanese, with the PDFs in each language and the date proof): YOI 623. This paper is a preprint and has not been peer reviewed.