Though being dry doesn't diminish the value of the text, though. Very exciting. I'd also be worried about ambiguity; humans can (sometimes) detect when they ma

Though being dry doesn't diminish the value of the text, though. Very exciting. I'd also be worried about ambiguity; humans can (sometimes) detect when they may be parsed the wrong way in context. I wonder if there will be a way to flag results that don't properly convey the data. How would that be integrated into the generator? (There's probably an answer in the literature.) Lots of fun questions to explore.