LexTower

The game

Build a crossword tower from today’s letters. Every run of letters must be a real word - and every word is also a beam. A word needs support: cross a letter, rest on tiles below, or lean flush against a standing tower. Nothing floats.

Each letter’s score is also its weight. Build heavy, high, or far off centre and the whole structure starts to tip - the bubble level tells you how close you are: green is safe, amber is strained, red collapses. The best vocabulary in the world will not save a tower with bad foundations.

Worth a closer look

• Two disciplines in one move: every placement is simultaneously a crossword play and an engineering decision, and neither skill substitutes for the other.
• The score is the peak, not the finish. It is the highest the tower ever stood, so a collapse costs you the climb ahead rather than the climb behind, and every turn asks the same question: climb, or brace.
• The forecast is public. The wind at each band and the quake that is coming are on screen before you commit, so a tower that falls fell to a decision rather than to a surprise.
• Failure is legible: the red squares mark the cause rather than the symptom, and the next tower starts in seconds.

How it plays

1. Start on a landing pad: tap one and type a word downward - the foundations.
2. Every placement needs support: cross an existing letter, rest on tiles, or lean flush against a tower. A word over open air wants two supports - or one it balances over.
3. Watch the level. Letter numbers are weight; heavy, high and off-centre tips the tower. Strain is visible before collapse, and an impossible placement tells you why.
4. Bridge the sky: across-words tied to towers on both sides - or two words meeting mid-air as a weaker truss - still count, and score the height they dare.

The daily

Today’s letters are the same for everyone, and the game is untimed - the pressure is structural, not chronological. Same material, very different towers: comparing skylines is the social loop.

How the tower is judged

Every word is laid in as a beam or a column, and the whole structure is re-analysed after each one. The engine reads the grid as runs and scores the joinery: a crossing where a real beam meets a real column is a joint, a beam capping two columns is a seam, a beam tying two columns is a truss, and two beams boxed in by two full-height columns close a frame. Each of those earns capacity.

Reaching out past your last support spends capacity, and spends more the further you reach. Three courses with a joint at the same column make a fracture line, which is the crack that staggered bond exists to avoid. Capacity is reckoned per footing, so a wider base genuinely carries more, and every row you climb moves the centre of mass and takes capacity away.

Demand is the overturning moment plus buckling, and buckling is the one that surprises people. It costs nothing at all until a shaft passes about ten times its own width, and then it climbs steeply: nothing at eight, nine or ten, six at eleven, fifteen at twelve. Stable, stable, stable, gone. The answer is not to brace the thin thing. It is to stop it being thin.

The weather

Wind and earthquakes are seeded from the date, so everybody meets the same forecast on the same day. A standing wind blows at each altitude band and gusts multiply it on a schedule you can see coming.

Earthquakes are the examination. Severity climbs band by band, and a word’s risk is how far it reaches past its supports, scaled by severity and by height, divided down by every distinct support holding it and by how well the tower is braced. Past the threshold, the part reaching past its support comes away and the rest stays put, at most three beams to a quake. Then the structure is re-solved.

Nothing is destroyed by decree. If losing that beam puts demand past capacity, the ordinary collapse rule fires on its own. And a properly tied beam has an arm of zero, which means it is untouched at any magnitude: the quake takes exactly what was not tied down, and nothing else.