What does Algorithm mean?

An algorithm is a clear, finite sequence of instructions that produces a specific output from a given input. Every single step is fixed, and the sequence ends after a finite number of steps with a defined result. An algorithm is thus the computational rule itself, not the program or the language in which it is later implemented.

Deterministic algorithms always return the same result for the same input, such as the Euclidean algorithm for finding the greatest common divisor. Probabilistic methods draw on random values and return a result with a stated probability. Learning algorithms such as decision trees or gradient boosting derive their rules from training data rather than receiving them ready-made. How strongly compute time and memory demand grow with the size of the input is described by its complexity.

A fixed, explicitly formulated algorithm pays off wherever the same decision comes up often and the criteria can be named in advance. Typical cases include allocating scarce places, planning routes for a vehicle fleet, and setting the order in which incoming cases are processed. Once the criteria can no longer be described unambiguously, the domain of learning methods begins.

The advantage of a fixed rule lies in its repeatability. Identical cases lead to identical results, regardless of who processes them and on which day. Because the rule exists as text, it can be checked, versioned, and produced in a dispute — something that is not possible with a decision based on experience alone.

An algorithm does not judge on its own; it applies the criteria that people have defined beforehand. Which criteria are included and how heavily they are weighted remains a substantive decision that must be documented before the first line of code is written. Where the outcome affects individual people, a legal basis is required in addition.