So I think this concept isn't bad. I'm not saying I love it, but it's okay.
A few scientific innaccuraccies to nitpick at:
We can U-series date cave paintings back millions of years http://science.sciencemag.org/content/336/6087/1409 as seen in this article.
Secondly, arabic numerals definitely didn't exist in the paleolithic era, which I'm assuming this is supposed to date from. Welp, I'll give you that one, 2.5 billion works.
Smilodon was from the Pleistocene epoch, which includes the paleolithic period, but without a period attached to this, it's hard to determine the accuracy of this.
I'm not a fan of the idea of not being able to affect the paint in any way, as it seems like a cop out from the harder science of the peice, by covering it up with being unable to touch it.
Other than that, the tone of the description is a little off, with imagery being used, and phrases like "generalized version of things present during the era" which is honestly a two million year period that we don't particularly have a complete record, supposedly squashed in to a 160m^2 area.
The observation log feels off, tone wise as well, with things like describing objects in the abstract (berry farm, socialization "taking place" over a specific time)
As a general critique, this isn't a particularly grabbing article. There's some interesting images, in the log, but nothing unexpected.
I think this is fairly well thought out, but the execution needs work.
Edit: After reviewing some of the changes, my vote is changing from downvote to upvote. Thank you for addressing the inconsistencies. That one small change (dating to 2.5 billion years) really changed the tone of the article completely.