This actually doesn't seem like such a bad idea. Just a business-level application of the s/w development principle, "build the first version to throw away" by Frederick Brooks.

I wouldn't say "design for failure" so much as "treat as a training exercise." Why waste a potential first-time-lucky shot and a few years? But you can basically go in with the mindset, "this is a dress rehearsal/trial run" and set "acclimatization" objectives to prepare for future success, like putting a great team together, building your hiring networks, getting to know the local startup culture, other CEOs, and get a practice round of pitching with VCs etc.

So long as you try this with a really dirt-cheap idea (under 50k), of the Y-combinator level, the learning costs may be well worth it.

So yes to the idea, but obviously "no" to the "high-profile" failure for the simple reason that you have to work pretty darn hard to get a "high profile" to fail with, which usually means raising so much money that it can't be your "trial run."