Anonymous ID: 37b131 Dec. 11, 2019, 2:27 p.m. No.7483478   🗄️.is 🔗kun   >>3504 >>3523

>>7483422

>>7483337

 

gov documents are printed then re-scanned and OCRed, presumably to avoid information about redactions etc leaking out.

 

OCR doesn't do a good job at differentiating between "r-n" and "m" (and some other combinations too).

 

so the kerning is mostly irrelevant, what's relevant is the frequency of errors and whether the errors change when you re-print and re-OCR a document. we've seen that C-o-m-e-y is never recognized, even though other m's usually are, and that's the part that's damning.

Anonymous ID: 37b131 Dec. 11, 2019, 2:38 p.m. No.7483539   🗄️.is 🔗kun   >>3577

>>7483519

sure. just saying this kerning stuff is wrong and (imo at least) not notable

 

>>7483523

it's more than that. in the IG report at least, both versions, the ocr is completely blocked from recognizing comey. this would never happen if it was just a dictionary.