To have the best chance of succeeding in a radical project, you should instead choose just a few related dimensions on which to make radical choices, and then make conservative conventional choices on all the other dimensions. This strategy minimizes the chance that some other project dimension will go badly wrong and take down your central radical idea with it.
“Doubt everything, but one at a time, not all at once.”
Interestingly, Robin Hanson has an existing post on this subject: