Loved this:

"...it was completely unnecessary to use a stemming algorithm because if you typed in a misspelled search, given the size of our index, there was probably someone who tagged their file using the same misspelling."