Stonefury

Five mana and two of them red buys you removal whose damage you do not get to set: it equals your land count, which means the spell is weakest exactly when you cast it on curve and strongest only after you have spent enough turns flooding into the lands it counts. That inversion is the whole problem. By the point Stonefury reliably one-shots a sizable threat, you have made your fifth land drop and several before it, and you are paying a double-red premium to do work that cheaper fixed-damage spells handle without asking your manabase a question. The design conceit is that a lands-matter deck turns its mana into a removal payoff, the same logic that rewards a big top-end finisher in those shells, except the dividend here is a single kill rather than an accruing engine. Instant speed is the saving grace: you can hold it up and ambush an attacker, and because you can wait until you have already made your land drop for the turn before deciding to fire, the number is as high as your board will allow at the moment you choose to commit it. But the floor is grim and the ceiling arrives late. This is a spell for decks that want a critical mass of lands anyway, where killing a creature is incidental to a land count the deck was already accumulating. Outside that lane, the math rarely justifies five mana and the color commitment for one creature dead.