What you describe is the reason I prefer hypothesis driven development (HDD) or Design-Thinking-Lean-UX-Agile - because the challenges you describe are common to many industries.

1. Capability uncertainty? I used to work in edtech as a Customer Insights researcher (UX+market research). The question wasn't "how many students are using mobile phones" but "how many students WILL be using mobile phones?"

That's why I love Roger Martin's "what would have to be true to make this a winning idea?"

2. Output complexity

This is a common challenge in video game development - you have to design for virtually limitless configurations in user hardware and software AND virtually limitless configurations in user behaviour. This is why people complain about action-RPGs like Skyrim being "buggy" - they are being asked to do the almost-impossible, and have been doing it for about 25 years.

I'd definitely recommend Alex Cowan's work on how to approach these kinds of uncertainties: you can't entirely de-risk development, but you can improve your guesses.