Algorithms that find approximate solutions to Traveling Businessman Problem are handful (some just use Markov Chains, a rather easy topic). Finding the exact solution is a hell lot harder.
If your solution has an estimated error margin of 2% or less, it works just fine for basically any practical purpose.
Just 4 years ago I was on the verge of doing it. Today while still having the recurrent though, I'm doing a lot better. And every single suicidal/depressive people is a different world, so yes, people can have 180° change in less than 6 years.
There is no single rule of thumb to apply here, as much as ignorance may lead you yo believe so.