A minimum viable product is not "selling the least effort product you can get away with", it's spending the least amount of time you can to build something that you can use to gather more data than you can use to focus your product on what the customer wants rather than what you think they want. Building what you think the customer wants will very likely be wrong - customers lie, they're deluded, they don't understand their own problems, and they can't see value in solutions. But they do start to understand the problem when they see a solution to it, even if it's the minimum viable solution. And then they start having ideas about what else they can pay for to make it better...

Source: 

I think the trouble with MVP is if you look at early days of Twitter for example (both its UI and the frequent 'fail whale' outages), it wouldn't pass muster today but at the time was enough to become one of the main platforms of the web. Nowadays, you'd need a much more polished MVP in order to compete because the standard of what end-users expect has risen a lot (both in terms of functionality & NFRs, like security, robustness, interop, etc.).