Friday, August 20, 2010

Quantification of the Attention Span of Star Trek Internet Users vs Surfer Internet Users

Since we're coming up to the end of summer, and I haven't posted for a while (I admit I've been busy over at http://hn-review.blogspot.com ) I thought it might be time for a little fun. Here's a little "study" I conducted way back in 2007. It's based on the work of Rispin (2005) [1] who analyzes the propensity of surfers (as in the wave-rider kind) to write dude with multiple u's when posting online. I looked at those who wrote Khan with multiple a's and analyzed the differences between the two groups:

Introduction: Fun while waiting for a delivery - saw a news article on the "defection" of a Canadian MP named khan entitled "khaaaannn!!" [1] and it reminded me of a colleague's email in April about the work of Rispin (2005) [2]. Rispin contended that for a word with a vowel that is often extended by forum posters, such as "dude" as spelled with multiple u's, the number of repeated characters typed can be fit to a decaying exponential function ( N(n) = N0*exp(-tau*x), where N is the number of users who still have the u key pressed at time x, N0 is the number of users at time 0 and n is the number of u's ) which is related to the typical attention span of the group of users typing 'dude' with multiple u's.

I decided to expand upon this work and compare the exponential decay constant for people who typed 'dude' (more than one u) [1] and those who typed 'khan' (more than one a) and to determine what differences, if any, exist between the two groups. A total of almost 27 million dude writers and 68 million khan writers (according to Google [3]) participated in my study under natural conditions - no participant knew that their propensity for writing excessively long words would be used to determine their attention span.

Results: The frequency of Kh(a^n)n where n is the number of a's is plotted in figure 1:


Figure 1


Figure 2 shows the number of users who still had the a key held down after a fixed period of time compared to the number of users who used only 10 a's, plotted on a log scale to allow for easy calculation of the decay constant:


Figure 2

The time is based on my computer's character speed which is about 29.6 chars/s. Examining at points 20 to 40 (20 to 40 'u's) of dude [1] we get a decay constant of 4s-1 and points 10 to 50 (10 to 50 'a's) of khan, 3s-1. So half the folks typing 'dude' gave up after holding down the u key for 0.17s and half the 'khan' folks let go of the 'a' key after 0.23s (both after the initial wait for the computer to realize you want to type more then one character).

Conclusions: Possible explanations for the discrepancy include: (1) that the people who type dude have a shorter attention span then those who type khan or (2) that people who type 'khan' typically choose to type it on machines which deliver more characters per second.

References:
[1] Singh, K.N. (1982) The Voice of Khan. Internet Site. http://www.khaaan.com/
[2] Rispin, C. (2005) Duuuuuuuuude: like, an analysis, right. http://www.tropic.org.uk/~crispin/dude/
[3] Google Search Engine (2007) www.google.com

No comments:

Post a Comment