I didn't read it. I've been coughing up phlegm for like 2 hours.
update: I just skimmed it. My conclusion: Hoey said he thought that a better word list and a smarter program could get to ten times his 540-word palindrome, using only noun phrases with indefinite articles. I'm pretty sure that will never happen. The problem is a dirth of "a"s. According to Hoey's rules, every phrase must start with the letter "a". That means that either the rest of the word must be an exact reverse of another word (and we know there are 1100 of these) or the phrase must have another "a" in it somewhere, and it must be matched by two or more other phrases. Phrases such as "a man", "a plan" and "a canal" work well because they contain multiple "a"s. Now consider a phrase such as "a biologist". If that appears in the palindrome, then somewhere else the letters "tsigoloib" must appear. But note that those letters must all appear in one word/phrase, because there is no "a", and we only get word boundries at "a"s. And of course, there is no single word that contain those letters. In general, take a word (such as "an asparagus" or "a biologist"), split it into components around the "a"s (yielding ["n", "sp", "r", "gus"] and ["biologist"]). Collect the set of all such segments, from all the phrases in the dictionary. Now go back through the dictionary, and for each word, see if the reverse of each of its components is in this set. So "an asparagus" is good, because its reversed components all appear in the set: "n" appears in many places (including "an asparagus" itself), "ps" appears as a component in "a psalm", "r" appears in many places (such as "a karat"), and "sug" appears in "a sugar". On the other hand, "a biologist" is no good, because the component "tsigoloib" does not appear.
so in conclusion, I gotta go make my lunch for tomorrow.
)
