The "M" in MVP has quietly come to mean "first version" — the smallest thing you can ship and then keep. That reading guts the entire idea. An MVP is not a small product. It is an experiment, and an experiment that cannot fail was never an experiment; it was a launch with extra steps.
The purpose of a minimum viable product is to buy a specific piece of learning at the lowest possible cost: does this solve a real problem for real people who will act on it. That framing has consequences founders dislike. It means the MVP is disposable by design — built to answer a question, not to be the foundation you pour the next two years of engineering onto. Treating it as a foundation is how teams scale the scaffolding and then wonder why the building keeps cracking.
It also means the healthiest outcome is sometimes death. An MVP that dies at the right time — that returns a clear "no" in six weeks — is a success, not a failure. It bought the most valuable thing a startup can own: the truth, early, while it was still cheap. The failure is the MVP that overstays, propped up because it exists, accruing users and code and sunk cost until killing it feels like killing the company.
So the discipline is to keep asking what the MVP is for. Discovery does not end when it ships; it ends when the question is answered, and then a new question begins. Build less, learn faster, and stay willing to throw away the thing that taught you — because the moment you cannot bring yourself to kill the experiment, you have stopped experimenting and started hoping. And hope is the most expensive way to be wrong.