You refer to a section on a Wikipedia page that talks about
''how to solve in a way that is faster than testing every possible answer''
and paraphrase this in your own story.
This is a very misleading way to popularize the problem. There are many NP-complete problems where there are plenty of techniques that solve the problem much faster than by testing every possible answer.
An example where I know all the details is the linear integer feasibility problem, asking for deciding whether a system of linear inequalities with integral coefficients has an integral solution. Branch and bound (and enhancements) is the prototypical method for solving these problems, and fairly large problems can in many cases of practical interest be solved quickly. It is also known that the problem can always be solved in finite time. But testing each possible answer takes an infinite amount of time.
Nevertheless, this has no effect on the validity of P=NP or its negation, which are effectively statements about the asymptotic worst time cost of a family of problems whose size grows beyond every bound.