That first example in the article was interesting because it made me realise that I have a strong aversion to pushing the complexity into the data rather than into the code. Modifying the list seems to me that it could be a cause of other bugs (what if the list is re-used and something depends on it?).