Word Search

Every word on the list is somewhere in the grid, spelled out in consecutive cells along a straight line. Find one and mark it: drag from its first letter to its last, or tap the two ends if you would rather not drag. Nothing is hidden diagonally-then-round-a-corner, and no word ever runs off one edge and back on the other. When all of them are struck through, the grid is finished and the clock stops.

• Words share cells. A letter can belong to two words at once, and on a 15 × 15 grid about twelve cells do.
• Words never bend. If the line is not straight, it is not a word.
• Nothing is upside-down or mirrored — only reversed, which is a different thing: RETAW is WATER written right to left.
• The list is the whole list. Only listed words score, and no listed word is ever hidden twice — the filler is checked to make sure of it.

Eight, at most: across either way, down either way, and the four diagonals. Which of them are in play is what the level actually controls — the grid getting bigger is the smaller half of the difficulty. On Hard the generator draws a direction first and a position second, so all eight are equally likely: half the words run diagonally and half run backwards.

The backwards diagonal is the hardest line in the game and it is worth knowing why: reading is trained left-to-right and top-to-bottom, so a word running up-and-left is the only one of the eight for which the eye has no habit at all. It is the last direction most players think to check and the first place an unfound word usually turns out to be.

Because the filler was not checked. This is the defining bug of cheap word searches: the generator places the words, sprays random letters into the gaps, and by chance one of those gaps completes a second copy of a word already on the list. The player finds that copy, the board refuses to mark it, and the game looks broken — when in fact the player was right and the puzzle was wrong.

It is rarer than it sounds and far commoner than it should be. A 15 × 15 grid holds 572 straight runs of five cells, each readable from either end, and most of any run's letters are already fixed by a word crossing it, so the filler only has to guess one or two letters correctly. Measured over 2,000 hard grids, unconstrained filler produced a duplicate in about one grid in 220. Here, every empty cell is handed the set of letters that would complete a listed word through it in any of the eight directions, and that set is excluded before a letter is chosen — zero duplicates in the same 2,000 grids.

The related trap is the palindrome. LEVEL and RADAR read the same in both directions, so a generator that counts hits by scanning all eight directions sees two copies of a single placement and rejects its own grid forever. Occurrences here are identified by their two endpoints rather than by a direction, which counts a palindrome once — and lets you drag it from whichever end you reach first.

Hunt letters, not words. Reading the grid looking for a whole word is slow, because you are matching seven things at once; picking the rarest letter in the target word and sweeping the grid for that one letter is a single-symbol search, and the eye is very good at it. JAVELIN has a J: there will be one or two on the board. PENGUIN has no rare letter at all, so leave it until later.

• Take the longest words first. A nine-letter word fits in far fewer places than a five-letter one, and it lights up more of the grid when you find it.
• Scan columns deliberately. Reading habits make horizontal words nearly free and vertical ones nearly invisible, so give the columns their own pass.
• Double letters are a gift — the SS in BLOSSOM, the LL in TRELLIS. A repeated pair is far more visible than any single letter.
• Do the diagonals last and do them one direction at a time. Sweeping for 'any diagonal' finds nothing; sweeping for down-right, then up-right, works.

Honestly: less than the marketing suggests. What a word search trains is visual search and pattern scanning, and it trains those well — it is essentially the same task psychologists use in attention experiments. What it does not do is exercise vocabulary or reasoning, because the answers are printed on the page next to the grid. There is good evidence that puzzle habits correlate with sharper attention in later life and very little that any one puzzle causes it. Play it because a grid of letters resolving into a word is a genuinely satisfying moment, which is reason enough. If you want the deduction as well, sudoku and minesweeper are next door.