Faces of Open Source

Changing topic slightly, I'm just mucking about with OCR. Why can't OCR process a PLAIN TEXT FILE CORRECTLY?

My document looks like it came out of nroff, with nicely aligned sections and indents. I don't know how it was OCR'd (probably with one of Google's Artificial Stupidity tools), but the OCR software has decided that the document is formatted in columns (no it isn't), so all the text fragments are all over the place and it's a fing nightmare trying to put it all back together again!

If it's a simple document, ffs OCR it as a simple document! Left-to-right, top-to-bottom, DON'T assume fancy formatting that isn't there!

Cheers,
Wol