Puzzlingly Appropriate

September 3, 2015 at 10:27 PM

In 2013, I did my ‘2500 Clue Challenge‘, where I tried to solve cryptic crossword clues for charity.

Interestingly, Indian scientist Sundaresan Naranan has contacted me to say he has used the data from the challenge to confirm something he’s found from his own solving: The success (or failure) rate of someone solving crosswords follows a “Negative Binomial Distribution” (this is subtly different from the binomial distribution, maths fans!).

Using data from his own solving of over 5000 crosswords over the past 16 years (WOW), he found that with a particular set of parameters the negative binomial distribution was a good fit to the number of crossword clues he failed to solve per crossword. But he was only one man using cryptics from the The Hindu, an Indian newspaper. When he discovered my data, he was able to see if it also fitted my solving statistics based upon a different solver and a different pool of setters from a different newspaper. Pleasingly, he was able to find that with slightly different parameters, the same distribution applied to me too.

I don’t follow all the maths, but you can see his write up in his pdf here: vindhiya.com/snaranan/VNS-xw2015w1.pdf and see more of his work here: vindhiya.com/snaranan/

I think it’s awesome that someone used the data from the Challenge in an interesting way. Thanks for sharing it with me Sundaresan!

Meanwhile, I’m still trying to put my thinking cap on for future potential challenges. I haven’t quite found any that have been as unusual or interesting as the 2500 Clue Challenge yet… so if you have any ideas, please do send me a tweet!