show OCR alternatives (second guess)
Capital I, lowercase L, and the one digit all look the same in some fonts, and OCR may return any combination depending on context. In my application, I often know what to expect. My code corrects for that case, but not if a less-obvious error occurs, like an "A" is mistaken for a "4".
If OCR returned second and third (and …?) possible interpretations, my code could perform the final analysis step and my clients would find the software usable.
This is make or break for my application.
(This alternatives option would let the service-consuming programmer deal with regular expressions.)