01-07-2025, 02:05 AM
I find this really striking when I'm looking at the text. It's even weirder than the copying with occasional permutations or errors. It's like a Cambrian Explosion of mutations. My weak guess is that this is a behavior pattern of one of the scribes. I do hope we learn something. At one point, I saw something (I must find this spot again) that looked as though the size of the copy-buffer shrank below the size of a typical word, down to the size of a prefix or suffix or syllable even smaller. This made me think of the process as fractally self-similar, it could go all the way down into the ligatures or combined glyphs or even what we describe as single glyphs, but that can't be. I just can't imagine people behaving like broken robots. My first-draft code that looked at this did a "for each space-delimited word on the page, look for that string in the page-as-one-string." That led to finding words that can be prefixes, or suffixes. Which is old news, but it was interesting to see it in this context. I don't seem to believe that this process is 100% of what causes the Maybe I should regenerate the results and stare at them some more. I didn't really know where to go from there because tracking a word through permutations and then showiing it as a sub-word prefix or suffix would be harder and it's hard to see what we'd learn. There'd have to be a lot of threshold-setting.