Did the “R” mess up on both the father’s name and the son’s?
That is a very good point. Not it is not. It is all in the same resolution. That tends to indicate the name was not pasted together. If it were, you would expect both names to have the same sort of flaw.
Upon examining it though, it brings another point to mind. The Checkbox with an "X" in it is in the same resolution as most of the other letters, yet if OCR is going to fail, it will most assuredly be on that "X" that is crookedly on the checkbox. (Just under the Child's name "Barack")
I also notice that the letters making up the fathers names are not exact copies of the letters making up the son's name. What that means is that the process treated those letters as images, not as Ascii characters. One of the compression methods is to substitute a specific token image for all representations of the same letter (Takes one byte instead of several hundred.) That is obviously not occurring here. The letters are being treated exactly as if they were images.
It still makes no sense why the pixel resolution and bit depth change from one letter to another. Even if you assume it was some sort of error, it makes no sense. The "X" not quite aligned up on the checkbox should have been an OCR error, and it wasn't handled the same way as was that "R".