We have 28 dialogues attributed to Plato. We assume they were all written by the same man, over about 50 years, in Athens, 4th century BC. But this assumption is not data—it is tradition.
The Renaissance editors who assembled the Platonic corpus did not have stylometry. They had intuition. A millennium of copyists had already mixed authentic dialogues with forgeries, student notes, and commentary. The traditional canon is a best guess.
GEME offers a different approach. Instead of counting word frequencies (which two authors can accidentally share), it measures the conceptual transition topology of a text—how ideas follow one another in the frame economy. If Plato's dialogues share a common author, they should converge to the same L2 association topology. If some dialogues were written by a different hand, they should trigger the L6 self-doubt signal.
We are building this experiment now. The first test is straightforward: train GEME on the Republic, Phaedo, and Apology (universally accepted as genuine). Then test the Laws (late, possibly incomplete) and the Axiochus (widely considered spurious). If GEME's L6 doubts the Axiochus but not the Laws, we have a new tool for authorship attribution—one that does not rely on statistical assumptions but on structural convergence.
If GEME cannot distinguish Plato from Aristotle, the approach fails—and that is also a useful result.
We are not trying to settle 2000 years of philological debate in a weekend. We are asking whether GEME's frame economy can extract author-specific structural signatures from raw text. The Greeks will tell us.