Given that scientists can use x-rays and other techniques to fathom things like the antikythera device from a mashed up hunk of ocean debris, couldn't similar techniques be used to deconstruct Yoshizawa's unpublished models?
It seems as though step 1 would be to x-ray each model, and identify the actual crease pattern and layered topology of the model. That should be fairly easy. Step 2 would be to figure out a linear set of folding instructions that would produce the model in question. This is essentially impossible because we can never know what Yoshizawa really did. But over a period of years, it's likely that members of the origami community could develop increasingly refined guesses that seem to better and better match Yoshizawa's design and instruction style. Would it be great to have 50,000 Yoshizawa models diagrammed, instead of just a few hundred? And since there is literally no other way to eventually diagram these models, why not do it? If cost is a factor, crowdfund it. I'd donate! Be well, Zack -- Zack Brown