On the surface it sound silly but it can be a very complex problem where automatic tools would take too much time and work compared to having a person doing it

On the surface it sound silly but it can be a very complex problem where automatic tools would take too much time and work compared to having a person doing it manually for the data which is relevant at any given day. Maybe the output from the database is old an application using mixed character encodings (with corruption and errors), spanning relationships design long forgotten, with the primary function to get any useful data from the application being to print it out. To take a real example, an organization has a internal website where workers post reports. What the programmers did not expect is that what workers actually do is to write the report first in a word processor and then drag and drop the text into the form. This produce a bunch of garbage characters which the database accept but when it need to be printed back out things crashes. While the fix they did was to add a filter to the input, I can easily imagine others fixing the output so that it doesn't crash and which then any future applications also need to copy in order to handle the garbage data in the database, especially if you need to move data from one database to an other. A better solution is naturally to print to file and have a person manually go through it in order to detect corruption. No need to waste paper.