Experiments with interpretation of the labels. The label otol.

[-JKP-]

...
But expecting anything about the VMs to follow in a predictable and logical fashion is asking far too much, especially given the possibility that the VMs was designed to hide information and confuse the curious.

It is actually quite predictable. Not in the sense of being able to predict the next token (although Torsten has a few important things to say about that), but in terms of predicting the uniqueness or otherwise for a token.

After inputting a few folios into my transcript, I could predict, with about 70% accuracy, whether the next token was a new one or one that had already been used. This was not based on memory (I can't memorize that kind of thing), it was based on the construction of the token.

But this, by itself, doesn't mean it's meaningful, only that there are common patterns.