Archaeologists refer to this "gibberish" as pseudo-glyphs: symbols that replicate the basic units of real glyphs and appear in the same location but follow no linguistic rules, either of the known language or a hypothetical other one. At no point, though, does the scenario you've described occur. Our only surviving codices were written in the Post-classic period, possibly as late as the 14th or 15th centuries.

Rather, pseudo-glyphs are contemporary with the real script and are frequently limited to a certain community or social group. One study mentions the prevalence of pseudo-glyphs at Xunantunich. The fine wares with quality writing are believed to have been largely imported from regional super power Naranjo. In the late 9th century, Naranjo's control declined, and provincial centers like Xunantunich expressed greater individuality. At this point, vessels decorated with pseudo-glyphs are no longer manufactured at the site. It's likely that the style was a conscious imitation of Naranjo's more learned scribes and they wished to sever that connection.

Kevin Johnston's study at Itzan suggests a more class-based distinction. "Inexpensive" dishes with rough psuedo-glyphs indicated "an illiterate craftsperson." Other evidence furthers the conclusion that the structures where it was found were inhabited by "persons of modest socioeconomic means." Markus Eberl's recent book corroborates this; he discusses utilitarian vessels from low status households decorated with repeated, simplified versions of common glyphs.

How do we know they're not actually writing? Sometimes it's easy. Look at the vessel in fig. 10b (pg. 108) in this article. There's a top band of glyphs and two vertical bands in the right picture. That's where writing would usually be, but if you notice, it's all just the same symbol being repeated across. That'd hardly be writing, even if the symbols in the top band actually meant something. Compare that to the vessel in fig. 9. You see glyphs in the same spot, but they show the variance of an actual language.

Sometimes it's harder. Take a look at the vessel in fig. 5d in the Xunantunich article. The symbols are almost repeated, but not quite. How do we know it's just some decorative crap? Well, thankfully for those who have worked to translate the language, the writing on most ceramics is pretty formulaic. We call this "labeling" the Primary Standard Sequence, which I explained more here. Even in a language with ~12 different signs for the vowel/3rd person singular pronoun u, the topics and signs used in writing are fairly regular. It's like finding random Greek letters mixed in with a Latin text. (Assuming, of course, you don't know it's a Greek letter.) It's pretty obvious when something doesn't fit, and if the entire thing is illegible, odds are none of it is legit. Could it be another language? There's no evidence to believe as such. The signs are distinctly approximations of real glyphs, terribly misused "cut-and-pasted" glyphs, and/or lack and recognizable pattern.

I have read that later Mayan writing is gibberish and that these later scribes were imitating their predecessors but did not know what the symbols meant.

Do you remember where you read that? When the Spanish arrived there were still people who could read and write Mayan. In fact, there are even claims that a few Spanish missionaries were able to learn the Mayan writing system from them.