Joe has a problem with his car, it makes a terribly annoying sound. He goes to Mechanic A
who decides to replace his engine. Problem solved, cost 10000$.
John has exactly the same problem (with the same cause), he goes to mechanic B who finds out
that a small part of the engine needs replacement. Problem fixed, cost 65$.
Btw, software, unlike buildings or chairs, is extensible and the (technical) desing decisions
of the programmer are very important in how extensible it is or if it is extensible at all.
Those examples are just outliers. You can always be unlucky. But the majority of mechanics can fix your car. And the problem remains: how are you supposed to know beforehand which mechanic is good?
Extensible software: maybe it doesn't matter so much for a prototype. If the concept proves itself, you could also start writing it again.
Btw, software, unlike buildings or chairs, is extensible and the (technical) desing decisions of the programmer are very important in how extensible it is or if it is extensible at all.