I'd even go so far as to say that wishes kinda
suck in cube, simply because the situations where wishes are
good hardly come up in draft formats.
Like, a wish is good when you're using it to pretend that you have seven copies of a key combo piece (3x in deck, 1x in board, 4x wishes), when you're using it to "tutor" up silver bullets against decks that are vulnerable to them, or in, like, ramp decks where they let you shove some of your payoffs into your sideboard and open up space for more ramp spells (which
helps solves the old "what if I draw the wrong half of my deck?" problem). All of these are things that can and will come up in constructed, where you've had the opportunity to fine-tune your 75 into an efficient game-winning
machine.
In comparison, a draft deck is the
rough equivalent of trying to build a car after a smash-and-grab at the local junkyard. When you take into account all of the unplayable
garbage that you were forced to pick up and how rough it is to get to your 23-24 playables as a result, all three of those scenarios seem pretty unlikely. Using wishes to stretch the number of combo pieces you have doesn't work if you only have one copy in the first place, draft decks tend to be
far more robust against silver bullets than their constructed counterparts, and your ramp archetype is going to be so comparatively jank that you wouldn't notice the difference.
This isn't really restricted to wishes (tutors
also kinda suck in less-curated decks), but wishes definitely run into these problems harder than most similar "find a single card" things.
---
If you wanted to make a cube where wishes were
useful and cool, you probably want an environment like japahn's
Smooth Twin cube or the like, since if you can have N copies of a given card in your deck, wishes let you pretend you have 2N-1 copies instead. And you also
probably want really good fixing so that more of your non-playables are valid wish targets.