James Pine Has A Blog

Jump to Comments

My Flickr account, while public, is not something that gets many views. According to the stats, my photos have been viewed 23,710 times. 10% of those views are for one photo someone decided to link to in a comment on Digg in 2009 (original post is here and someone re-posted a link to my photo as a story here Three Rupicaprid Ninjas Climb A Steep Cliff Wall).

Digg did not allow embedding images directly into the post so I got to see the explosion of views firsthand. Recently I noticed a couple of referrers from tumblr.com in my stats, and they all happend to be for the same image of Yogurt Drinks from Amsterdam.

The Flickr stats showed only a few tens of people looked at the image directly on Flickr, but I was still curious what people were posting about my image. It turns out, there wasn’t much. The image was just one in a stream on various image centric tumblelogs. However, there were 775 “notes”, most of which appeared to be in a particular format: TumblrUser this

For example:
fairyfloss89 reblogged this from fusels
jessicajmaciel liked this
jessicajmaciel reblogged this from urban-princess

The UNIX utilities cut, sort, uniq and grep are my friends, so it was relatively easy to get some stats out of this data:

cut -d\ -f2 notes | sort | uniq -c

 156 liked
   1 love
   1 nomk
   1 posted
 606 reblogged
   1 reeeaally
   2 want

That command says using as a delimiter (-d\ ) print the 2nd field (-f2) of each line, sort them and then count (-c) the number of unique occurances. Looks like there were 156 likes and 606 reblogs, so what where the others?

grep -v “liked\|reblogged” notes

i want a smoothie:(
id love to scoop one of these up
omk nomk nomk :D
I reeeaally want a mango pineapple smoothie right now.
DRINKING ONE RIGHT NOW 
I want them all.
candystaples posted this

That command says print all lines that don’t contain (-v) the words liked or (\|) reblogged. Looks like 6 people decided to post a comment and the source post came from candystaples (here’s a link to the actual post). People can both like and reblog the image, how many of those were there?

cut -d\ -f1 notes | sort | uniq -c | grep “^ \+2” | wc -l

38

That command says using as the delimiter (-d\ ) print the 1st field (-f1) of each line, sort them, count (-c) the number of unique occurrences, print the lines that begin (^) with 1 or more characters followed by a 2, and count those lines (-l).

So at least 606 + 156 – 38 = 724 unique people saw my image and liked or reblogged it. It turns out, when reblogging the comment format is TumblrUserA reblogged this from TumblrUserB. So candystaples was the source, but which other tumblelogs generated the more reblogs?

grep “reblogged this from” notes | cut -d\ -f5 | sort | uniq -c | sort -r | head”

  86 fusels
  75 iamfuckingglamorous
  37 dingyfeathers
  31 fashionistha
  17 madamelulu
  16 s-o-u-l
  14 urban-princess
  11 breakingthehabits
  10 izzyparadise
   9 sellyourseconds

That command says print all the lines that contain “reblogged this from”, using as a delimiter (-d\ ) print the 5th field (-f5) of each line, sort them, count (-c) the number of unique occurances, sort those in reverse/descending (-r) order and print the first 10 (default for the head command).

I hadn’t really thought about what sort of multiplier to attach to my Flickr stats in order to determine how many people had actually viewed one of my photos. Clearly not every photo deserves a 10x or greater multiplier and in cases like this, 10x is likely to small e.g. if 700 people took the time to perform a like or reblog action, is that 1% of the number of people who actually saw the image? And this doesn’t take into account the times photos are just copied and posted elsewhere with no attribution e.g. my experience with a Famous Board Games Photo.