A month or two ago someone related their experience with OCRing code and getting it to work; even when it compiled correctly there were still subtle errors that he/she had to spend hours finding. With the recent threads on OCRing the DES cracker I was thinking of ways to make it more foolproof without going so far that the FEDs would get excited. If the author/publisher was to include a hash of the source code the person doing the scanning would know when it was good without having to compile it and then run it to discover there were still errors in the source. A hash per page of code would be even easier as it would localize the errors to a more manageable area, or even more sophisticated methods could be used to localize any errors. How far you could go in providing feedback on the correctness of the OCR process without getting the FEDs all excited is the question. Just a thought miner