The Internet race gap.

Among the strange properties either revealed or inventedⁱ by the Quantcast report for Qntra published alongside its monthly report for November, the incredible African American index stands out :

The naive interpretation upon first seeing that thing would be that you know, thirty black people saw the site for each 108 white guys or 168 azns. As Quantcast itself explainsⁱⁱ, this is not actually correct. The values represent "indexes", whatever that actually, mathematically means (Quantcast does not bother to explain, because whatever, we live in Disney stupid world, there's no need to expose the actual underpinnings of any mechanism, just make some vague claims, wave hands a lot and everyone's happy) and a value of 100 is representative of some sort of an Internet Average!

To understand each other : if 100 pink people and 30 purple people live somewhere on an island, upon which island there exists also an icecream shop, which then hires Quantcast to do racial profiling for itⁱⁱⁱ, should ten pink people and three purple people consume icecream in some interval the indexes of both would come out as 100. They would not come out as 10 and 3 respectively, as the naive approach above suggested, they're not absolute values. They would also not come out as 10 and 10, ie, 10% of the population in each case. They would come out as pure 100s, ie, "out of the 10 pink people consuming icecream, 10 consumed icecream ; whereas out of the 3 purple people consuming icecream, 3 consumed icecream". "All", aka 100, is both the average, the maximum and the minimum for both groups in this simplified example of ours.

If we go deeper to consider icecream flavours, and if it so happens that out of the ten pink people, six had chocolate and four vanilla, whereas out of the three purple people, one had chocolate and two vanilla, it is entirely unclear how the Quantcast reports would look, but I am quite warrantlessly assuming that they would come out as 120 / 66 for chocolate and 80 / 133 for vanilla. Because the "Internet average" is 5 pink and 1.5 purple people per flavour, and then proportionally 6 is 120% of 5 and 2 is 133% of 1.5. This is at least what seems to be implied by the "Very High Indexes" thing.

So equipped with this highly speculative understanding of the statistical processing involved, we can now proceed to look at the Internet. For instance : maybe Qntra being "independent" in its political leanings discourages black people, more likely to congregate on democratic-leaning bastions of thought, either because of Ms. Walker or for any other reason. But then, Gawker looks like this :

Check out all the big boned, great-sense-of-humor, respectful friendzone-fodder grouped up on gawker, while the 20% of women that actually matter aren't even there! Who knew, seriously. But, and this is the truly strange part, black people at only 73% ? On the largest site^iv on the Internets catering to what white asocial asexual aspie boys think black people want ?

The mystery continues : linkedin 70 ; ijreview.com 28(!) ; topix 77 ; whitepages 82 ; thoughtcatalog.com whatever that is 65 ; nbcnews also 65 ; examiner 78 ; drugs.com gets close at 99 but still no cigar ; sbnation 98 ; diply.com 71 ; comcast 60 ; legacy.com 63 ; stackexchange.com 74 (meanwhile azn ? 239 yo! record breaking) ; goodreads.com 88 ; hubpages 99 ; sportly.tv 63 ; photobucket 91 ; guff.com 50 ; eonline.com 77 ; vox.com 90 ; nbcsports 92.

Amusingly enough bleacherreport.com (which apparently is a big thing I never heard of) shows a slight overaverage : 108. Cheap lulz aside, there are also a few more substantial overperformers : nydailynews 137 ; urbandictionary 126 ; tmz 120 (and pretty much the only place where blacks and blacks only are over average) ; quizlet.com 127 (very popular with all minorities, this thing) ; babycenter.com 113 ; inquisitr.com 106 ; anchorfree.us 109 ; uproxx.com 103.

Count it : 21 under, 9 over. On average, the 21 under are at 70.5 (ie, 29.5 under) while the 9 over are at 116.5 (ie, about half that over). On top of which : these are those sites among the 100 most viewed on the Internet that chose to publish their demographics. A larger chunk chose not to publish their demographics, and given the race hysteria that's been going on in the US for a while now I suspect this decision is not independent of the fact that their demographics show underaverage representation of black people.

So to sum up : the average number of black people on the Internet is, on average, about 15% lower than the average numer of black people on the Internet as far as we can see - and it's probably even worse as far as we can't see.

Doesn't this give you pause ? Where's my niggers at, yo!

1. It's still unclear exactly what part rand() plays in there. [↩]
2. Reading Demographic Graphs
   1. Index. This compares audience composition of the site or mobile app to each platform population. The higher the index number, the more concentrated the property is in a particular demographic. As an example, if a property indexes 100 for age 18-24, that means a given visitor to it is as likely to be 18-24 as any internet user chosen at random. An index of 200 means the visitor is twice as likely to be 18-24, 50 means half as likely, and so on.

   2. Segments are represented with icons. Segments include gender, age, household income, and education.

   3. Very High Indexes (over 200) are denoted with a plus symbol.

   4. Internet Average is represented by the dotted vertical line. [↩]

3. Hey, it works for da police, right ? Why wouldn't it work for the Internet fictitious icecream parlours on fictitious islands. [↩]
4. It's really the whole conglomerate, 25 different "properties", 10 allegedly different sites (gizmodo, lifehacker, gawker itself, deadspin, io9, jezebel, kotaku, jalopnik and of course everyone's favourite spamsource, kinja.com). Oh, what, you thought gawker & jezebel could survive on their own, if they didn't have gizmodo and lifehacker to push the numbers ? Think again.

   In fact, the whole charade looks so exactly alike the cheapo Romanian spamrings I was taking apart and mocking all the way to their grave a few years ago it's not even funny. [↩]