> [...] and this enabled others to OCR it, and recompile it offshore.

Did it? Or did it just give them plausible deniability?

I remember playing with OCR as a kid and all the software I could get my hands on gave horrendous results, even if the input was as perfect as one could hope for.

And even today I sometimes run tesseract on perfect screenshots and it still makes weird mistakes.

Would be interesting to know if the book had any extra OCR-enabling features. I'm sure the recipients would get access to proper tools and software but OCRing source-code still seems like a nightmare back then.