Daily Word Ladder

Doublets: The Daily Puzzle That Predates the Daily Puzzle by 145 Years

At Christmas 1877 Lewis Carroll was staying with friends and two girls complained of having nothing to do. He invented a game for them on the spot: take two words of the same length, change one letter at a time, and make sure every step along the way is itself a real word. He called it Doublets, published it in Vanity Fair the following March, and ran it as a competition with prizes for the shortest chains. His first example asked players to turn COLD into WARM, and his own solution took four steps.

Why it is a graph problem, and why that matters here

What Carroll had invented, decades before the language existed to describe it, is a shortest-path search. Picture every four-letter word as a dot, with a line drawn between any two words that differ in exactly one position. A ladder is a walk along those lines, and the best possible ladder is the shortest such walk. Donald Knuth used precisely this construction — a graph of five-letter English words — to teach the technique, and it remains one of the neatest illustrations of breadth-first search anyone has found.

Building the puzzle on that footing rather than on a list of hand-picked pairs buys two things a curated list cannot. The first is that the tool can state the shortest possible length as a fact: it searched for it. The second is more important and much less obvious.

So this one does not pick two words and check. It picks the first word, searches outward until it finds words at exactly the intended distance, and takes the second from those. The route is known to exist before the puzzle is ever displayed, because finding the route is how the puzzle was made. The word list helps too: it is restricted to the largest connected region of the graph, within which every word can reach every other. That restriction costs nothing, because a word one letter away from a reachable word would itself be reachable — so anything outside the region could never have been a legal move anyway.

The midnight problem

"Everyone gets the same puzzle today" sounds like a statement about a puzzle. It is really a statement about a calendar, and it is false unless the calendar is shared. If the puzzle followed each player's local date, then for most of any given 24 hours a player in Auckland would be on a different puzzle from one in Los Angeles, and a link shared between the two would open the wrong thing at the other end. The fix is to use UTC everywhere and to have no local-date path at all, which is why the countdown on this page is to midnight UTC rather than to yours.

Across ASEAN that offset is worth knowing. Singapore, Malaysia and the Philippines run at UTC+8, so the new puzzle arrives at eight in the morning; Jakarta and Bangkok see it at seven. That is a considerably friendlier moment than the local midnight most daily puzzles use, and it means the ladder is waiting with breakfast rather than expiring at bedtime.

What this deliberately is not

There is no hidden word here, no row of coloured tiles, and no picture made of green and yellow squares to paste into a group chat. Part of that is legal caution — the tile grid and its emoji share image are trade dress belonging to a specific, vigorously defended game — but most of it is that a clone of something everybody already plays is not worth anyone's afternoon. Doublets asks a different question. Both ends are visible from the start, so nothing depends on luck, and two players given the same puzzle are solving genuinely the same problem rather than reacting to different opening guesses.

Sharing is therefore a sentence: the puzzle number, the two words, how many steps you took and whether that matched the shortest possible. It has a practical advantage over the squares, too. A sentence survives SMS, a screen reader, a plain-text email and a search box; a block of coloured emoji arrives as a wall of question marks on anything that has not kept up with the emoji standard, and as nothing at all to someone using a screen reader.

Getting better at it

The single most useful habit is to work backwards as well as forwards. Write down what the target word could have been one step before the end, then two, and aim to meet your own chain in the middle — bidirectional search roughly halves the space you have to hold in your head, and it is the same trick the solver uses. The second habit is to notice junctions. Some words sit at the crossroads of the graph because many letters can be swapped in them; land on one and the whole map opens up. Others are near dead ends, and a ladder that wanders into one has to come back out.

And when the shortest possible is five and you got there in eight, that is still a solved puzzle. Carroll gave his prizes for the shortest chains, but he published the long ones too.