Since people have repeatedly asked about my progress with transcribing the handwritten pages of the manuscript, I’ve decided to post updates on my progress on this here.
As of tonight, I have transcribed 8 pages of 108 total.
Page 8 / 108
█░░░░░░░░░░░░░░░░░░░░░░░░
Yes! Excellent. Look forward to reading it.
Page 10 / 108
██░░░░░░░░░░░░░░░░░░░░░░░
Page 12 / 108
██░░░░░░░░░░░░░░░░░░░░░░░
Page 22 / 108
█████░░░░░░░░░░░░░░░░░░░░
Page 25 / 108
█████░░░░░░░░░░░░░░░░░░░░
So much for trying cutting edge technology AI tools… This supposedly specialises in historic documents. How is my handwriting harder to read than some scribbling by a medieval monk??
A medieval monk has probably had hand-copied thousands of pages before he wrote the page that’s being analyzed. Which leads to the monk’s handwriting having two features: 1) every letter looks (almost) the same as the other instances of that letter on the page (except for ligatures, perhaps, but those can be identified separately), and 2) the handwriting in general is at least above-borderline legible for those literate (otherwise, the monk would have been given some other task—like weeding turnips—after his first two hundred pages).
Your handwriting lacks the first feature, and is quite close to lacking the second one. 
You just had to seriously analyse that, hadn’t you? Fuck you, man! 
Page 31 / 108
███████░░░░░░░░░░░░░░░░░░
Page 35 / 108
████████░░░░░░░░░░░░░░░░░
Page 37 / 108
████████░░░░░░░░░░░░░░░░░
Page 40 / 108
█████████░░░░░░░░░░░░░░░░
I’ve started training a machine learning model to transcribe my handwriting. It is far from perfect yet, but it is certainly speeding up the process.
Page 50 / 108
███████████░░░░░░░░░░░░░░
Page 57 / 108
█████████████░░░░░░░░░░░░
67 of 108 pages
62.04%
71 of 108 pages
65.74%
78 of 108 pages
72.22%
81 of 108 pages
75%
