Does anything break if the GM simply picks out one patron Arcanum and four targets?
I notice that if the GM simply determines targets randomly, there is a chance that only three, two, or even one world will be played in: which can be disappointing.
I also notice that if there are supposed to be obstacles equal to the total number of Approaches plus +33% to +50%, then it is possible for the PCs to simply solve obstacles using their Approaches and then use their Blessings for the rest, completely bypassing the Deck of Fate. Is this intentional?
Can the GM use the same shift more than once? Esther Marlowe's Predictable suggests that the GM can use the same shift more than once, but this becomes game-breaking with defensive shifts like Gloria Fitzroy's Financial Revenge or León Gallo's Game Face. Indeed, if the GM spams León Gallo's Game Face, he cannot actually be defeated.
To answer your questions:
1. No, nothing breaks if you just pick the targets.
2. Not all blessings are one-shot problem solvers. Some just open up opportunities that you wouldn't otherwise have, like The Devil for instance, which just tells you terrible secrets about people.
When I ran it +33% to +50% was enough obstacles, but if your group have all picked up easy-solve blessings then you might want to up that number a bit. Essentially, you want to force the players to think a bit about what they're doing rather than just spam win buttons.
3. My intent was that each shift should be used once -- a bunch of them don't make sense if they're used more than once -- but that you need room for reuse on some of them in case combat goes on for more than 4-5 rounds. In hindsight Financial Revenge and Game Face should definitely also come with the 'don't use this more than once' note that's attached to Predictable, just for clarity.