I'm most surprised the MVP effort can be so consistently estimated. It might be that the pool of "apps" is extremely shallow, so they're just reimplemention #1231 of "XYZ" therefore estimation is simple based on extensive past experience reimplenting #1230, #1229, etc. "fart app #2935315" takes just as long to MVP as doggcatcher or evernote?
Also a major problem I've experienced is trying to force what I consider the technical requirements for a MVP past the non-technical people barrier, I didn't see that in the discussion, maybe it was assumed. For example, in my opinion, at least some minimal backup strategy is part of the MVP, whereas the non-technical types just reply "well, if THE hard drive crashes, we'll just sue the hosting provider, my buddy is a lawyer so he'll work cheap" or "if it blows up, you'll just work 36 hours straight on salary to fix it, right, so don't waste money/time preventing it from blowing up or adding debugging code or monitoring code, just work on features for the PR checkboxes".
Also a major problem I've experienced is trying to force what I consider the technical requirements for a MVP past the non-technical people barrier, I didn't see that in the discussion, maybe it was assumed. For example, in my opinion, at least some minimal backup strategy is part of the MVP, whereas the non-technical types just reply "well, if THE hard drive crashes, we'll just sue the hosting provider, my buddy is a lawyer so he'll work cheap" or "if it blows up, you'll just work 36 hours straight on salary to fix it, right, so don't waste money/time preventing it from blowing up or adding debugging code or monitoring code, just work on features for the PR checkboxes".