Sure thing. Think of the recipe as a point in a multi‑dimensional ingredient space, each axis a measured amount. An experimental twist is just a small shift along a chosen direction. You can model it as adding a delta vector to the base vector. If you want to keep the overall volume the same, you normalize the resulting vector. The amount of surprise is essentially the angle between the original and new vectors; a large angle means a big flavor shift. You can compute that with a dot product: the closer the dot product to the product of the magnitudes, the more similar the flavors; the farther apart, the more unexpected the result. That’s the math behind turning a measured recipe into a canvas for creativity.