A friend told me an anecdote from the time he was working in a Chinese restaurant. A customer, called Robert, wanted to get a tattoo of his name in Chinese and so asked one of the Chinese staff to write down his nick name - Robbo - as a pictogram.
There's no Western "r" sound in Chinese, and the staff member subtly altered the pronunciation of Robbo to something closer to Lobbo - which means "carrot" in Chinese. The pictogram was duly drawn, and Robbo happily went off to get himself tattooed large as being "Carrot".
This resonated with my data quality genes in two ways. The first is what I call the Chinese Whisper effect. You know that game - one person whispers a word or phrase into the next's ear, and so on down the line, and what comes out at the end is often completely at odds with how it started. Data quality is like that - at every interface between information and data and between data systems, the quality of data goes a little more awry.
The second has to do with ignorance. Most organisations think their data is great simply because they don't understand it, and that's especially the case with names and addresses. If you can't see the problem, or, indeed, see that there is a problem, you can't correct it.
Robbo lives in ignorance about his tattoo and is probably still mighty proud about it. He may get a few sniggers if he ever goes to China, but that's about it. Unfortunately, data quality issues arising from processes which work like this can be much more dangerous.