What problems can you catch with an MVP? Use RATs instead?

Interesting piece about minimum viable products that argues we should build “RATs” (Riskiest Assumption Tests) instead.

This has relevance to concept design; you could imagine a RAT for just one or a few concepts. But I wonder if the idea might be mistaken and MVPs are valuable for a more subtle reason. As I wrote in a comment there:

Nice idea nicely explained. But there’s one reason to consider that an MVP might address risks that a RAT will miss. The RAT assumes that you know what the risks are. But your product might fail because users don’t like it for a reason you hadn’t anticipated. Simple example: you might omit authentication for your app on the reasonable grounds that it’s well understood. But perhaps for this particular app users will think that authentication is too burdensome. My conclusion from this is that RATs and MVPs address different kinds of risks. An MVP simulates the experience of mainstream use of the app, so it has a chance of detecting a flaw in the very conception of the app. A RAT could test a particular interaction you’re concerned about, or might expose snags in implementing a particular feature.