Pursuant to my earlier investigation into Unintentional Pilish occurances, a correspondence with Mike Keith, author of “Not A Wake: A dream embodying (pi)’s digits fully for 10000 decimals“.
Thanks for the interest in this problem, and for your results. I’ve also done similar searches in years past with various subsets of PG, but I think you have used a few more GB than I have, and you have found a few examples that I did not find. My favorite one (which I had already found) is the Captain Cook, since it starts at the beginning of a sentence.
I know that I haven’t seen the one with the 15-letter word, which is nice.
However…I recently found a 9-word example, which I’ve attached as an image file (this book is not available in PG, of course, since it is recent and copyrighted). This is the current world record. Somewhat shockingly, I found this one by simply playing with Google for a few minutes using partial-Pilish phrases that I thought would tend to be common. Nonetheless, it seems clear that searching massive amounts of text mechanically is the way to go for further records.
So, we have the following records (as far as I know):
Starting from the beginning of a sentence: 8 (Capt. Cook)
Not necessarily at the beginning of a sentence: 9
I would, of course, be very interested to hear of new records in either of those categories.
Best Pilish Regards,
Great! Thanks! Would you mind if I were to post what
you wrote on my blog?
I’m planning to make my little script for searching for these strings
more efficient (via multithreading etc) and maybe I can rig up a
webcrawler so it can be searching continuously. I’ll let you know if
anything interesting or any new records emerge!
Also I was trying to come up with a way of determining the probability
of finding a Pilish string of length n in a document length m. The
most obvious way is with a Markov Chain, based on probabilities
calculated from the dataset, so I’m planning to give that a try also.
Sure, no problem – feel free to post what I wrote.
Your 15-letter-word example led me to do another Google search with “straightforward” as the root of a manual search. I eventually found another new record! Look at the 4th post on this page:
David has given you a very straightforward procedure to follow. Since you have not done any CV programming you should…
with the part I have colored in red giving 3.141592653 (10 digits). Interestingly, this was written just a few months ago, in Dec 2009.
Keep me apprised of any new results. It’s great to have another person interested in the search. Would love to see some probability results based on Markov chains.