Monday, March 31, 2008
Compressed Sensing: Who are you ? and the poll result.
This blog was initially there so I would be able to put those nagging ideas in a written form somewhere. It has evolved to be a little more than that in many ways but the most profound change recently has been a focus on Compressed Sensing or Compressive Sensing. With that in mind and after a year in that mode, here are some of the statistics of the blog that might account for how a nascent subject is faring on the interwebs.
First things first, there have been about 170 posts on the subject in the past year. More recent entries provide the abstract of the new papers because I believe that in this day and age, unless there is a location that does that, the papers won't get read. It's a shame to say this but my own experience suggests that unless you have been hooked on by an abstract and maybe some graphs, it is very difficult to get people to read your stuff. No matter how good the paper is. The reason ? Everybody is busy and at the same time drowning under a pile of things to do. As one of the reader mentioned, each entry provides a good read when commuting. The blog serves the purpose of getting people's attention on every new paper but it also provides for Google a way to index it. While Mark Davenport and other folks at Rice do an excellent job at gathering new papers on the subject when it is submitted to them, sometimes a more pro-active means is necessary to find new papers on the subject. Mark also does a great job at adding other subjects areas that have, in retrospect, great relevance to the subject.
Eventually, I am still bewildered at the different communities that contribute to the subject. The side effect of this mix is that the community sees itself as loosely tied both in terms of subject areas treated and geographically (as witnessed from the heat map above.) The figure below provides actual hits and page views to the blog. It shows a steady increase of the readership mostly through search requests from Google. This March was the first time the blog saw more than 4,000 page views in a month. This is not a Slashdot effect as I have seen before but rather a growing trend over several months.
The blog is also read through RSS readers that are traditionally not counted toward page views or site visits. Feedburner says that about 40 readers read the blog daily, while Google Reader says that I have 62 readers using that application. Those two numbers should not be added but rather should be understood as saying that about 50 people read this blog daily through their RSS readers.
Feedburner also provides page statistics, and the 160 average daily visit squares well with the 4,000 visit/monthly given above. That 160 number is also well correlated with Google analytics numbers shown below.
But the statistics I like the most is this one: In the past week, about 20 percent of the people visiting the blog directly, ended up spending more than 5 minutes on the blog. Five minutes is a loooooong time......
Now with regards to the poll. Answer 2 "Compressed Sensing: A Paradigm shift in sensing, don't be the last one to say DUH" is the one with the most hits. It looks like going for the Simpsons is always a winner.
One more thing, the ads ran for 18 days and the words "compressed sensing" and "compressive sensing" seemed to have been searched on Google about 1,600 times or about 88 times a day (number of impressions) during that time period.
And remember, this statistic 50/160/88 squarely goes against George Constanza's routine: It's not me, It's you :-)