Banner
    Racked Up A Million: Should I Retire?
    By Johannes Koelman | January 16th 2012 08:03 PM | 23 comments | Print | E-mail | Track Comments
    About Johannes

    I am a Dutchman, currently living in India. Following a PhD in theoretical physics (spin-polarized quantum systems*) I entered a Global Fortune

    ...

    View Johannes's Profile

    No, not a million of those... I am talking about page hits. And no, there is no reason for retiring: the next million is awaiting!

    I didn't give much thought to page hits when I started blogging. And I probably still don't give it enough attention, but I do realize it is page hits that make the internet tick. And yes, it is stimulating for me to witness a piece that I have written attracting many thousands or even tens of thousands of hits. Yet somehow it feels unreal. Who are all these folks clicking on a link to this blog? A million hits for a nerdy and rather inaccessible physics blog is way more than I ever could have expected. 

    Or is it? Let's put this figure of one million in perspective. If this blog continues attracting page hits at its current rate, it would take a respectable ten centuries to get to the number of hits of a video of two toddlers uploaded onto YouTube. And that is a video that got created in a minute and happened to be uploaded for the mere purpose of allowing it to be watched by the toddlers' godfather. And that figure of ten centuries probably creates a too rosy comparison for the blog you are reading: it seems safe to assume that a significant fraction of the 400 million hits of the toddlers video resulted in a full 56 seconds view with all the nuances ("That baby knows perfectly well what it's doing!") getting conveyed. Can the same be said of the million hits of my blog? I doubt it. 

    Well, in fact, I don't doubt it: I am pretty sure that the number of hits to the posts on this physics blog tells me very little about how often the various articles have been read and digested. I would be surprised if the number of true readers isn't at least an order of magnitude smaller than the number of bare hits.

    Let's have a look at the available statistics.

    I started this blog May 2009, 32 months ago. A total of 65 articles were posted, a modest two per month.* These attracted a million hits, and 2,350 comments. It seems reasonable to assume that the number of reads is in-between the number of hits and the number of comments. Lacking a better method, and to settle on a number, I am going to apply the 'Hammock Rule'. This rule can be applied to blogs every reader can comment on: the number of people actually reading a blog post is the geometric mean of the number of hits and the number of comments posted by the readers:

       #Reads  =  Sqrt( #Hits  X  #Comments )

    This rule is based on the assumption that reading the full blog post given one has stumbled upon it, provides as much a hurdle as reacting to a blog post given that one has read it. This assumption translates into the expectation that if one in every N people who happen to hit upon a blog post do make the effort to read it, also one in every N people who happen to  read it do post a reaction. The rest is simple math. 

    With #Hits = 1,000,000 and #Comments = 2,350, it follows that my blog posts have been read in total roughly 48,500 times. A number that translates into an order of magnitude estimate of close to 750 faithful readers that pay careful attention to each Hammock Physicist blog post. That doesn't seem an unreasonable figure, given the fact that my least popular posts have delivered a number slightly north of 1,000. With close to 750 recurring readers, and taking into account duplicate hits, one would indeed expect a thousand or more hits also for the least popular posts. So let's settle on a number of 750 loyal followers who pay attention to each of the posts on this physics blog.

    As many as 750 loyal readers, that is a number that I am more happy with than the million hits. Thank you, all seven-hundred-fifty of you!

    Now back to the hard data. On average each hammock post attracted slightly more than 15,000 hits. The variation around this average is huge, with the most popular articles attracting over 100,000 hits. This strong variation must contain some important insights on what info the pop-science-hungry are seeking. The twelve most popular Hammock Physicist articles each attracted over 25,000 hits:

     1) What's Wrong With E=mc2?
     2) It From Bit - The Case Of Gravity
     3) Survival Of The Stupidest
     4) Holographic Hot Horizons
     5) It From Bit - Entropic Gravity For Pedestrians
     6) Shut Down The LHC?
     7) The Art Of Acting Rational
     8) Limits To Science: God, Godel, Gravity
     9) Big Bang, Big Bewilderment
    10) Metric Versus Imperial
    11) It From Bit - Dark Energy
    12) Gravity Of Free Will

    Can you spot a pattern? I can't. I have no idea why precisely these articles come out on top. What do energy-mass equivalence, entropic gravity, game theory, the LHC, and units&measures have in common? If you see a commonality between these twelve posts, let me know!

    One last piece of interesting statistics. It is well known that rankings such as the population ranks of cities, corporations size rankings, income rankings, citation rankings, meteorite size rankings, earthquake amplitude rankings, etcetera, all tend to follow Zipf's law. This rank-size rule states that the top ranking item will approximately be twice as large as the second largest, three times as large as the third largest, etc. Does the same regularity hold for Hammock Physicist page hit rankings? 

    Not entirely surprising, the ranking of my blog posts by hit frequency indeed does display a Zipfian behavior. This can best be seen by plotting the number of hits for each of the blog posts versus their page hit rank in a log-log plot. If Zipf's law holds the result will be a line with slope -1. Such a behavior is indeed observed for the "Hammock Physicist" page hit statistics (see figure, the purple straight line indicates Zipfian behavior with slope -1). 

    Why is this Zipfian behavior so ubiquitous? Why does it describe phenomena as diverse as your blog surfing behavior as well as city sizes and meteorite impacts? 

    That question deserves its own blog post...



    Notes
    * Yes Hank, I know: attention span on the Internet is short, and to build a significant group of followers, I need to make that into posting at least twice a week, or even better: twice a day. If only I could find the time...

    Comments

    Thor Russell
    Oh dear, your first comment isn't from a regular follower ;-)
    Thor Russell
    Hank
    I am not sure how you are using 'hits' versus 'readers'.  Are readers only people who read to the end? Because otherwise the number you see per article is the number of people who clicked on to read it - readers.   How many never finish it is unclear, stats can only tell us how long they stayed, so if they left the tab open for 2 days and never read it, we don't know.

    I agree about never being able to predict an audience - I write just about every day but some pieces only get 300 readers, though I average around 300,000 readers per month, so the metric you use won't work for everyone - I assume I have more core readers than I could get if I divided readers by comments - plus, Google News is a big variable in reads, as are social news sites like Reddit.  I would have a hard time making an equation that accounted for that variation.
    Johannes Koelman
    For me a 'hit' represents a scenario like: stumble on link, click on it, scroll through article, shrug shoulders, click browser back button. A reader is more like: click on link, read first paragraph, become intrigued, read on...

    According to the above, I estimate some 20 'hitters' for every 'reader'. I might be slightly too pessimistic on this, but I don't think my pessimism is excessive. I have not included in the analysis the number of hits that spend less than 30 or 60 seconds on the page. This represents a sizeable fration of the total number of hits, and as such vindicates a finding of vastly fewer readers vs hitters. I certainly would not label anyone spending less than a minute on a Hammock Physicist article as a reader.

    The top articles indeed all got traffic from Reddit and/or StiumbleUpon. But that begs the question "why were these articles targetted by users of these sites"? It does seem that anything perceived as mysterious (dark energy, gravity, etc.) and anything controversial (entropic gravity) has a higher chance of attracting more attention. Apart from that: I really have no clue.

    And what are the top articles in terms of the more relevant 'number of readers' is a question shrouded in even bigger mysteries. ..
    ”A number that translates into an order of magnitude estimate of close to 750 faithful readers that pay careful attention to each Hammock Physicist blog post.”

    *raises hand* Not sure who the other 749 are but if we each tell two friends, and they tell two friends, and they… world domination! Betcha, that’s how those two Brits got all those hits (I just learned from you about them in this post and already showed my wife; it works!.)

    Thank you, all seven-hundred-fifty of you!

    No, thank you! And I’ll get to that ‘tell two friends’ thing ASAP.

    Regards.

    +1... 748 to go...
    Congrats, on reaching a million!

    Sorry about the 'blockquotes' not showing... not sure why. They have never worked for me on this site?

    I always read you and Vongehr. You're both very good in your own distinctive ways, and when you two get to butting heads, it makes for an excellent point/counterpoint exchange and is well worth the price of admission.

    Kudos on the one million hits... +1 and 747 to go...

    Cool. Three of my favorites made the top 12.
    Keep them coming please.

    Hi, it's one of the faithful cadre. About the list of top articles: can you just consider that they are your finest pieces for your audience? Because if I compiled a list of my favorites it would very much look like that (with Physical Reality: Less Is More, which I think is one of your finest).

    Looking forward to your article on Zipf's, it's a "law" which I find fascinating. It will be fun to read your thoughts about it.

    Johannes Koelman
    I agree. I consider Physical Reality: Less Is More one of my best. Maybe the best so far, as it is a prime example of what I want to achieve with this blog: making modern physics accessible to interested readers. It achieves this in the widest sense, as it covers all of modern physics. But the point is (as you noted): it doesn't feature anywhere near the top-12.

    More examples of articles that make modern physics concepts accessible are:  The Principle Of Maximal AgingMarking The Grand Arena Of Physical RealitySquare Root Of The UniverseThe Gravity Of Free WillEinstein Got It Wrong. They all would feature high on my personal list of best blogs. None of them attracted many hits. (Which is fine, it's just that I don't understand what made the other articles so popular.)
    I also remember fondly all the articles you cite: they are very good. It is that kind of hardcore science made easy which I (being an ex-physicist myself) like most. But I said "finest for your audience", which is wider than the faithful and probably has less of a Physics background.

    In the top 12 you make science in general accessible, and the masses like that. Also: they are shorter, and the medium favors short pieces.

    Johannes Koelman
    All -- thank you all for your kind words.

    And keep reacting and commenting to the various blog posts. I won't have the time to react to each comment, and I try to avoid reacting to comments I agree with and can't add much to. But rest assured I do appreciate each and every of the thoughtful comments you all place here!
    blue-green
    Question for Hank: suppose I open a new article by Johannes, start into it and am interrupted. Perhaps I close the browser .... reopen later ... start reading again .... click on a hyperlink within the article .... get interrupted .... come back later ... read some more from the same computer and ISP connection ... click on another hyperlink .... come back later. Get to the end. Make a comment. Edit the comment three times to correct typos and improve clarity. Get 10 email notifications that something has changed in the blog ... click on a few of the email links .... wander around looking for changes ... feel inspired to comment more .... meanwhile ... by now I have refreshed my browser (as I often do for other reasons) so I have to sign in again ... then I have to find the article to which I wanted to review or comment ... recheck my email ... find a link to the article ... click on it .... scroll down .... make a new comment ... edit that comment three times .... 
    In all of that activity, how many hits did I generate?  
    If my behavior is even a little bit typical, then J's 700 club is more like four score ....

    Too often great opening essays are left to whither on the vine .... by their authors .... as if everything here is a one-night stand .... with no further correspondence necessary .... everything tied up in a neat bundle which the author can refer back to in a hyperlink. Problem solved. Moving on.

    The more Johannes and Sascha write, the more hyperlinks they make to back articles and the more Google and other robots make links to the links. There is no honest way to count hits. An author has to just pull in his or horns horns and write as well as s/he can without contemplating hits or loading an article with proper names and popular phrases .... to generate hits.
    Hank
    Fault it if you want, but if these guys are going to do the work to educate people and I can get them 1,000,000 readers a month or 100 readers a month, I am going to choose 1 million (though we had 1.4 million uniques last month, I just picked a million as a round number) because they're doing the work and it's good stuff and deserves to be read.

    The most accurate metric is created by the people paying to be here - advertisers.  They don't have your cynical stance on pageviews or titles.  Obviously if it were as easy or as mercenary as you seem to be claiming, everyone would do it.  
    Statistically-challenged bloggers and pathological serial clickers... 




    blue-green

    Some of those hits could we be from people entertained by pithy comments that don't just tow the line.

    As a paid content provider to the WWW since 1996 …. I have been through quite a few phases of the Hit Count Parade. My ISP stopped logging hits to my business's website many years ago because the monthly log files that were being generated were bigger than the 700 plus page site many times over. I know what my customers want. I create new material on a regular basis with no tracking of hits whatsoever (and no outside advertisers). Don't misread me, Hank, I'm not complaining about your ads. I'm just trying to add a wider perspective here on the issues, web crawlers and all that. Seriously, I was asking how many hits my type of activity generates. I think that with YouTube, if the same user repeatedly clicks on a video (even over several weeks), the tracking software makes sure that this pathology is counted as just one view. If that user, goes to a completely different computer interface, he creates another view. If the creator of the video edits the description of the video, the views meter at YouTube does not mark it as a new view. I am speaking from experience here. Just trying to be helpful and a wee bit confrontational to generate hits. Thanks for answering my question, Hank. Cheers.

    (my edits Derek are often to add space between words that are unintentionally jammed together when I cut 'n paste from Open Office ... so just where does the pathology rest?)

    http://www.mountainlake.com/mlp/mlptube.htm 

    Did I mention you?
     
    blue-green

    Instead of blithely adding up hits as if they were all equal-valued real numbers, let's take an approach inspired by Feynman. Each hit is a number, an amplitude, in the complex valued plane. A clear-to-the-theme of the article comment gets the value one. A comment from a high-ranking scientist gets a much higher real value. An entertaining comment that is completely tangential is off on in the imaginary direction. A completely tangential remark that is not interesting is pegged in the negative imaginary direction. A completely annoying horsefly comment (destructive interference) is off in the negative direction of the reals with perhaps some plus or minus imaginary component. Hits from webbots are given a tiny positive number … and a swing in the negative imaginary direction if they come from robots combing for porn.

    You get the picture. To get the grand number for the total hits, the impact, you vector sum all of the little amplitudes and only then do you take the norm of the resultant. That's my 2+2i cents here.

    (OK, I came here to post this …. I noticed I was not signed in …. I backed out to signed in … I re-enter and post this  and hopefully see no need for an edit. What is the amplitude of my visible one hit on Feynman's scale versus Hank's? ... (don't forget to weigh in that I used a famous name that will be picked up by the heavy-hitting bots ....))

    Bonny Bonobo alias Brat
    A comment from a high-ranking scientist gets a much higher real value. 
    Oh dear, sounds very elitist and who's going to decide that rank? The Erdos number maybe, that should cause a few more punch ups here....
    My latest forum article 'Australian Researchers Discover Potential Blue Green Algae Cause & Treatment of Motor Neuron Disease (MND)&(ALS)' Parkinsons's and Alzheimer's can be found at http://www.science20.com/forums/medicine
    Hi Johannes,

    I'm one of the 750, and I do notice a pattern in the 12 most popular articles. They're the ones that keep popping up whenever I try to find other, less well read posts from before I started following your blog ! very annoying indeed.
    Websites should have an 'attention normalizer' so that articles (or people) that already get too much attention are tuned down a bit and other, perhaps equally interesting (you just won't know, would you ?) are promoted to read.

    Anyway, thanks for all the fish !

    Hank
    Artificial Intelligence just isn't that good - but people en masse are quite smart. A self-tuning way to sort articles by reader, especially people who will be 95% anonymous, isn't possible.  And penalizing a writer's most-popular articles to artificially subsidize less popular ones is mandating that the opinion of the public in the past means little in what the public of the future might like.  That's in defiance of everything we know about crowd sourcing and its benefits.
    nothwithstanding the truth of what you're saying, if you haven't been one of the 750 *all* the time, it's nice to be able to go through older posts without a priori qualification. but then, that's an archive, which is there :)

    Hank
    Yep, there's no ideal solution but we have (a) related articles, which works quite well but isn't author-specific and is topic instead and (b) search just one author by topic under their avatar box and on their profile and then (c) reverse chronological on their profile.  

    Many authors with a lot of content also create their own page of links because then they manually highlight stuff they are proud of that may not have had a lot of readers, or break them into sections, and they refer to that in the bottom of articles or in their signature files.