Ewen Cheslack-Postava me@ewencp.org @ewencp

April 14 2012
Tags: url shorteners | twitter | followup

I posted yesterday about URL shorteners wrapping already shortened URLs. I didn’t have any tweet data to work with, so I could only look at the current state of URL shortening using Twitter’s public stream. Jeff was kind enough to provide some sample tweets from 2009 so I can look at how URL shortening has changed in the past couple of years.

Although the 2009 data had many more tweets, I extracted about the same number of tweets (100,000) and links from those tweets (~20,000) as I collected for the previous post. Here’s a histogram of the number of redirects taken per link found in the tweets from 2009:

Histogram of number of redirects for links from 2009. The vast
 majority have zero or one

and here’s the histogram for 2012, repeated from the last post:

Histogram of number of redirects for links from 2012. The vast
 majority have one or two wrappers around the real

The Y-axes differ a little, but the trend is clear: where links were previously almost all wrapped at most once (average .66 redirects), they are now wrapped twice or more nearly half the time (average 1.48 redirects).

I also checked the top shorteners (involved in any redirects) and re-shorteners (redirects to another shortened link) as in the previous post. The top shorteners aren’t surprising – bit.ly dominated in 2009 with about 78% of the links, and t.co dominates in 2012 with 72% of the links. But here’s the data showing the raw count of shortened links that redirected to another shortened link, for the top 10 reshortening services in 2009 and 2012, sorted by the 2012 numbers:

Site 2009 2012
t.co 0 7929
bit.ly 19 31
j.mp 1 5
ow.ly (owl.li) 3 5
fb.me 0 4
is.gd 2 4
mrkt.ms 0 3
apne.ws 0 2
w3e.us 0 2
lnk.ms 1 2
dzone.it 2 0
yit.me 2 0
migre.me 2 0
retwt.me 2 0
cnn.cn 1 0

The number of tweets in the two data sets aren’t exactly equal, but it doesn’t really matter because the numbers are so skewed. There might be a slight increase in reshortening among the services represented in both years, but t.co links dwarf all the other services. Twitter seems to be single-handedly increasing the degree of indirection to the destination sites linked on Twitter.