We use an ELO algorithm to dynamically assess the difficulty of problems. The gist is that the more often people accurately, and quickly can solve a problem, the less difficult that question must be on average, and vice versa.
ELOs don't work without being logged on with an account. So, if you don't have an account yet, REGISTER HERE or just login.
ELO is a rating system invented for chess, and player vs player scenarios that we have adopted for solving puzzles. There are variations on the algorithm, but essentially the idea is to generate an objective, relative, self-correcting score that can be used to predict the likelihood of victory between two opponents. In Puzzla's case, the opponents are the individual puzzlee, and the puzzle itself. It won't always be true because people have varied experiences, but the idea is that, on average, if two people have similar ELOs in the same competency sub-category (think algorithms, security, Java, et cetera), they should have a similar success on a puzzle.
There are many different competencies, and the goals is for a person to generate enough answers that for any given sub-category of competencies that they get a reasonable understanding of the their grasp of it.