Gödel's incompleteness
Formal systems seem to provide the perfect framework for proving everything there is to prove within mathematics, but as it turns out, they come with limitations. Gödel's first incompleteness theorem says that within any formal system that's strong enough to express arithmetic, is free of contradiction and whose axioms can be recognised by a computer algorithm, there are statements, expressed in the system's own language, which you can neither prove nor refute. You can of course try to settle such an undecidable statement by adding whatever axioms are necessary to prove it, but according to Gödel's result, other undecidable statements will pop up elsewhere.
The most famous undecidable statement within the ZFC axioms is the continuum hypothesis. It's undecidable in a very profound way, but it's concerned with rather strange infinite sets of real numbers that working mathematicians rarely ever come across. "In the normal course of mathematics one considers much more regular sets of real numbers," explains Friedman. "In fact, these days, one often looks at things one can simulate on a computer, which are several layers of magnitude more concrete."
Over the years, Friedman and others have come up with examples of incompleteness that feel a lot more concrete than the continuum hypothesis. They still don't put ordinary maths into any immediate danger, but they are striking nevertheless, especially because you don't need to have studied advanced maths to get an idea of what they're about