Hmm? I don't think an algorithm has to give "correct" answers, it just has to be precisely defined. For example, one could say "For this problem, a greedy algorithm yields decent but suboptimal answers."

Merriam-Webster online says: "a procedure for solving a mathematical problem (as of finding the greatest common divisor) in a finite number of steps that frequently involves repetition of an operation" and "broadly : a step-by-step procedure for solving a problem or accomplishing some end".

Algorithms solve problems. Wrong answers are not solutions to (i.e. do not solve) a given problem. Hence, algorithm implies that it provides correct answers within the parameters of the problem.

> Hence, algorithm implies that it provides correct answers within the parameters of the problem.

And so they always do. The problem that an algorithm solves is defined by the algorithm itself. The purpose of a system is what it does[0].

Algorithms are useful to solve problems for us - the trick is always in making sure that the problem the algorithm solves is the one we want.

--

[0] - https://en.wikipedia.org/wiki/The_purpose_of_a_system_is_wha...

That's a bingo.

Disagree. We see this phenomena all the time: Right solution, bad input data. Right solution, wrong problem. Worlds turned to grey goo by replicators working to some technically correct algorithm.

These will never give the right solution, only something close.

The definition of algorithm is wider than you think. As your parent poster noted, "greedy algorithms" exist (as do many other algorithms which provide suboptimal answers). You can easily verify this by googling.