Backtracking is a refined version of Brute Force. It explores all possible solutions but "Backtracks" (stops and goes back) as soon as it determines that the current path cannot lead to a valid solution. It is the core of AI for games like Chess or Sudoku.
Imagine a tree of choices. You pick Choice A. You move down. If you hit a 'Wrong' state, you delete Choice A and try Choice B. This uses Recursion to manage the state stack automatically.
Pruning is the act of cutting off branches of the search tree early. In Sudoku, if you place a '4' in a column that already has a '4', you don't keep solving—you prune that branch immediately. This makes backtracking millions of times faster than blind brute force.
Q: "What is the Time Complexity of Backtracking?"
Architect Answer: "In the worst case, it is usually **Exponential O(K^N)** because you are exploring every combination. However, because of **Pruning**, the 'Average' time complexity is much lower. In an interview, always explain that while the math says exponential, the professional goal of backtracking is to find the most efficient pruning rules to keep the search tree manageable."