Dave Munger over at Cognitive Daily has posted the results of his poll asking readers to pick a number between 1 and 20. The number 17 was picked almost 18 percent of the time, compared to the expected 5 percent from his sample of 347 responses. Comments on the results included the following attempts at explaining why humans seem to prefer 17:

Austin wrote:

As someone who chose the number 17 perhaps I can offer my personal insight - I was looking to select the most unlikely number (sorry!). In this sense i wonder if this is actually a measure of perceived least common number rather than random selection - there is a body of research about the ability to actually generate random numbers being very difficult for human subjects (methinks Sallice, T).Michael said:

I think the answer is bisection. Our minds first go to (20+0)/2=10 but then we think, oh that's too obvious. So we next choose (20+10)/2=15. Nope, still too obvious. Then we go (20+15)/2=17.5, round it to 17 and the loop ends, because we are getting too close to 20 to be comfortable. Nonsense, probably, but that is what comes to mind...

Craig Pennington wondered,

...if the fairly uniform distribution of n*7 modulo 10 (7, 4, 1, 8, 5, 2, 9, 6, 3, 0) has anything to do with the popularity of 7 and 17. I'd bet (x*10)+7 is preferred for most "random" ranges (more so for primes -- and just under 1 in 4 primes less than 100 is 7 modulo 10.)i wonder what would be result of an equivalent experiment carried out on some of our primate cousins? Experiments do suggest that rhesus monkeys, at least, do posses the capacity for "spontaneous number representation."