In startup terms, generally it's necessary to start with a MVP, from which you build on. The iterative approach is effective at ensuring you make progress toward the problem you're solving, but I agree, the risk is that you might end up carrying forward or "inheriting" problematic assumptions which can impede your ability to effectively solve the larger problem in the long run.
On the flipside, the iterative approach at the very least will help you understand the problem domain thoroughly from multiple perspectives (iterations), even if that means you periodically need to throw out all of your work and start again from scratch.
In startup terms, generally it's necessary to start with a MVP, from which you build on. The iterative approach is effective at ensuring you make progress toward the problem you're solving, but I agree, the risk is that you might end up carrying forward or "inheriting" problematic assumptions which can impede your ability to effectively solve the larger problem in the long run.
On the flipside, the iterative approach at the very least will help you understand the problem domain thoroughly from multiple perspectives (iterations), even if that means you periodically need to throw out all of your work and start again from scratch.