Yes, there must have been a long pre-history of writing proper, like, leaving marks on trees or soil as a guide, a warning, etc. But it takes a civilization to start writing on a scale comparable to that of the Sumerians, and such a civilization cannot be lost to time so completely as to leave behind some evidence if its writing - especially given that they would likely use different media, including clay and stone.
It depends on the material. Incas uses knots for record keeping. Those are not as durable as burned clay. In our current global civilization with advanced computing tech, magnetic tapes, cdroms, and hard drives won't last very long. Maybe SSDs have greater durability, but they will also require computing tech to dicipher. I do not think the existence of surviving writing materials is a strong argument for evidence of civilization, and speaks more to the limitations of the technology.
Leaving marks on trees and such is not writing. It is not the transcription of spoken language. We have examples of this kind of symbol-placement on ancient cave art stretching far back into the ice age. It is not considered writing.
Written language is rarely just a transcription. It usually has a number of distinct differences in both lexicon and syntax. There are languages that are only written.
Art is not writing, true. Distinct symbols and syntax are required for writing. But it need not be transcription of anything spoken.