The 4-up evolution works great for images, less great for text.

Branching text interfaces are less satisfying than branching image generation.

Because text has to be serially scanned to be absorbed (whether being read or being heard).

But images are absorbed as a field, all at once.

You can perceive the image as a whole, including in your peripheral vision, allowing your eyes to dart to any part of the image that it wants more detail on without delay.

Wittgenstein pointed this out in his Picture Theory of Language.