Backtracking Algorithms — ELI5

Imagine you are in a maze. You reach a fork and go left. Then another fork — you go left again. Dead end. So you back up to the last fork and try right this time. You keep doing this — try a path, hit a dead end, back up, try another — until you find the exit.

That is backtracking. It is the computer’s version of solving a maze by trying every possible route, but being smart about backing up the moment it knows a path will not work.

Here is what makes it powerful: the computer does not just wander randomly. At each step, it checks: “Does this choice still make sense?” If not, it immediately backs up instead of wasting time going further down a bad path.

Backtracking solves all kinds of puzzles:

• Sudoku: Try a number in a cell. Does it conflict with the row, column, or box? If yes, try the next number. If no number works, back up and change an earlier cell.
• Crossword puzzles: Try a word for a slot. Does it conflict with crossing words? If yes, try another word.
• Arranging things: How many ways can you arrange 4 books on a shelf? Backtracking tries every arrangement systematically.

The key idea is “try, check, undo.” Make a choice, see if it leads somewhere good, and if it does not, undo that choice and try a different one.

Without backtracking, the computer would have to list every possible combination first and then check each one. Backtracking is smarter because it prunes bad options early, often solving problems millions of times faster.

One thing to remember: Backtracking is organized trial and error. It is not guessing — it is systematically exploring every possibility while skipping dead ends as early as possible.