Exactly! It tried to express that and a strategy for handling unknowns here - https://medium.com/@hfuecks/what-donald-rumsfeld-can-teach-u...
My theory is the reason why we’re not getting any better at estimating software projects rests on a flawed assumption; because computers are binary and deterministic, we assume that the process of creating software must also be deterministic and quantifiable.
Example: when Apple opened the App Store to 3rd party developers, would it be realistic to expert an accurate estimate from one of those developers, building their first app? There were so many unknowns; how does the build / publish process work? what will pass the review process? what about all these new APIs?