For example, Aristotle's writings point to Homer, Socratic dialogues, and other works by Aristotle, all of which survived to the present day at least partially. But shouldn't we be seeing far more references to texts we've never heard of in ancient writings? Am I wrong in assuming we don't?
Perhaps others might have input here but I'm not sure this is so much a historical question as a statistical one or social scientific one as it strikes me as an example of the Lindy effect. Which is to say that in order for a text to have been so initially widespread as to be written by Homer (c. 700BC) and referenced by Aristotle (c. 350BC) we're talking about a text that had already by virtue of having survived that long in the first place established itself as being unlikely to be lost or destroyed.
That said, within the class of all ancient texts that we have I would be incredibly skeptical that it is more common to see references to surviving texts than to lost texts. I'm thinking of late antique papyri documents like the Cairo Genizah that insofar as so many of them are letters or tax documents or otherwise mundane materials that they make reference to other, lost, mundane materials.