The weird of web "metrics"

Saturday, 06 December, Year 6 d.Tr. | Author: Mircea Popescu

Motto : <kakobrekla> Isn't Alexa rank like rand()*2/2?

On one hand, there's


So like... really ?! I'm THAT big in India ?

It got me thinking that who knows, that Shanti Dynamite chick is so into me she's reading Trilema compulsively and has an Alexa toolbar installed or something.

Then, I checked the Awstats for last month (which, while not necessarily more informative than any other random measure of nonsense, are nevertheless somewhat closer derived from actually verifiable data) :


That the US and its smaller, poorer cousin the UK are shown as 20%/10% in one source and 500k / 30k in another source I could perhaps see as a result of you know... errors. But picking up the 34th, moving it to first position and giving it a 20`000% stat boost in the process ? It's just a teensy bit rich, no ?

Sotto : See motto.

Category: Meta psihoza
Comments feed : RSS 2.0. Leave your own comment below, or send a trackback.

7 Responses

  1. They fixed it now!

  2. Mircea Popescu`s avatar
    Mircea Popescu 
    Sunday, 7 December 2014

    Lol look at that! How things can swing in a day ? (I thought that's a longer average, maybe a month or three).

  3. So what's the problem? For a 30 day average to move 10% in a single day you need merely a 300% outlier that day, no big deal.

  4. Mircea Popescu`s avatar
    Mircea Popescu 
    Sunday, 7 December 2014

    Upwards only, but who's gonna notice such things.

  5. Also depends on how divergent the dataset is.

  6. Mircea Popescu`s avatar
    Mircea Popescu 
    Monday, 8 December 2014

    This is true, but really, the #1 spot ? Shouldn't be that divergent.

  1. [...] this give you pause ? Where's my niggers at, yo! ———It's still unclear exactly what part rand() plays in there. [↩]Reading Demographic Graphs 1. Index. This compares audience composition of the [...]

Add your cents! »
    If this is your first comment, it will wait to be approved. This usually takes a few hours. Subsequent comments are not delayed.