Reading the Look-and-Say Sequence Instead of Calculating It

You see 1, 11, 21, 1211, 111221 and your brain starts hunting for addition, doubling, or some trick with squares. None of it fits. The rule is about reading, not arithmetic. Once you see that, the whole thing opens up.

The core rule

Each term describes the one before it. Scan the previous term from left to right. Group each run of identical digits. Write the length of the run, then the digit. That is all. MathWorld calls this run-length encoding, which is a fancy name for counting blocks.

Take 1211. Read it aloud: one 1, one 2, two 1s. Write that down and you get 111221. The digits themselves do not carry the pattern forward. The rule does. Try it by hand on a few terms and it starts to feel automatic.

Here is a quick scorecard. Build the next term from each of these before you check the answers below.

1. Start with 1. What are the next four terms?
2. Read 111221 aloud. What term follows it?
3. Start with 2. What are the first four terms?

1. The terms run 1, 11, 21, 1211, 111221, as listed by MathWorld.
2. Three 1s, two 2s, one 1 gives 312211. That follows directly from the rule.
3. Starting from 2 gives 2, 12, 1112, 3112, which both MathWorld and OEIS list.

Changing the starting digit

The sequence depends on its first term. Start from 3 and you get 3, 13, 1113, 3113. Start from 2 and you get the line above. These are different sequences, but closely related, so a puzzle that begins elsewhere can look strange even when the rule is the same.

So check the seed first. A lot of confusion comes from assuming the puzzle starts at 1 when it does not. Also ask whether it is the standard version listed in the OEIS entry. A thirty-second check saves you from solving the wrong problem.

Patterns hidden in the digits

The standard sequence from 1 only ever uses the digits 1, 2 and 3. Every term ends in 1, the seed, and apart from the third term each one begins with 1 or 3, according to OEIS. That gives you a fast way to throw out bad answers. If a proposed next term contains a 4, it is wrong.

There is also no 333 anywhere. The argument is neat. If 333 first showed up in some term, it would have to be in the term before it as well, which contradicts it being the first appearance. A restriction like that is worth hunting for in any sequence puzzle.

Now look at length. The digit counts run 1, 2, 2, 4, 6, 6, 8, 10, 14, 20, 26, 34, 46, 62 (MathWorld). They grow, and the growth settles toward a fixed ratio called Conway's constant. MathWorld says the same limiting ratio applies to every nonempty starting string except 22. I can tell you that much. I cannot walk you through deriving the constant in plain language, and the sources do not either.

MathWorld also notes that Conway sorted strings into "compounds" that split into independent parts and "elements" that do not. Strings of 1s, 2s and 3s with no four identical digits in a row eventually decay into 92 such elements, named after chemical elements. That is a nice detail to know exists, even if you never need it at the puzzle table.

Attribution and checking your sources

The sequence goes by "Say What You See", and in French "suite de Conway". Many accounts credit John Conway with inventing it. A 2024 contributor note at OEIS says otherwise: the first mention appears to date to the 1977 International Mathematical Olympiad in Belgrade. Treat origin as an open question. A famous name on a puzzle does not prove who made it.

The same habit applies to everything above. Confirm the starting term. Confirm the version. Confirm any origin claim before you call it settled. And be honest about gaps. The sources here give the terms and the rule, but no worked puzzle answers beyond those lists and nothing on scoring in brain-training use. Say that plainly rather than filling in.