Reimagining the Internet as a mosaic of regional cultures

Angela Xiao Wu, Chinese University of Hong Kong and Harsh Taneja, University of Missouri-Columbia

Most online maps of the Internet are architectural plans, engineering blueprints, anatomical drawings or statistical graphics. For example, the Internet has been represented as millions of devices connected to each other by 300 “[c]ables lying on the seafloor” with its center in a huge hotel in Manhattan.

The Internet can also be viewed as a network of hyperlinks between world languages used to produce online content or represented through Wikipedia as a map of human knowledge.

Yet we learn from historians of cartography that maps reflect the preexisting interests, desires and preconceptions of the society from which they emerge. The same goes with how the vast virtual territory is mapped.

Consistent with the rhetoric emphasizing technical connectivity led by US-based transnational corporations, the prevalent maps of the Internet privilege technical features – such as hyperlinks, content of Web pages, Internet infrastructures and service providers. In these maps, the Web tends to center on the West with the rest of the world at its “peripheries.” These, together with other representations of the global digital divide, highlight the dominance of the West.

Such views limit the public’s ability to envision the Internet as a globally inhabited cyberspace.

We mapped usage of the Internet, as distinct from its technical features. Viewed this way, the Internet is much less West-centric, and rapidly diversifying as the world’s populations engage with it in their own ways.

Mapping global Internet usage

Actual traffic patterns on the Internet differ from its technical architecture. Reimagining the Internet according to global usage, our research reveals a fairly decentralized Web with significant participation from the global South. Our mapping makes visible, on an unprecedented scale, aspects of Internet use that remain “largely invisible” when “viewed from the perspective of network centers.”

We analyzed traffic to the world’s most popular 1,000 websites – which consistently account for 99 percent of global Web traffic – during the month of September in 2009, 2011 and 2013, respectively. These data come from comScore, a world leader in Web audience measurement.

For each of the possible pairs of those 1,000 websites – more than half a million pairs in total – we looked at the traffic shared by its two constituent websites. For example, for the pair comprising The New York Times and Google USA, we looked at how many people visited both sites.

We viewed website pairs as connected if they had traffic overlaps greater than would be expected by random chance, as with the Times-Google pair above, or the pair comprising the Times of India and Google India. Examined in this way, pairs of websites serving users from different cultural backgrounds – such as the Times of India and Yahoo Japan – tend not to be connected.

The Internet as Global Usage: 2009 (left), 2011, 2013 (right).
The dots are websites and the lines represent the existence of significant traffic overlap between them. These show that global Web usage clusters itself into many communities of websites based on shared traffic. What the member websites of these clusters have in common with each other allows us to identify them as expressions of online regional cultures (see legend).

Analyzing online regional cultures

Mapping sites based on how much traffic they share with each other revealed interconnected clusters or communities of shared Web use. These corresponded well with major geo-linguistic regions, and we called them “online regional cultures.“ In addition, there are a few online cultures that span geographic regions; they tend to include either user-generated or adult content.

To conduct our analysis, we borrow the anthropological concept of ethnology, a scholarly tradition that characterizes relationships between cultures based on common traits in beliefs, emotions or practices. To examine these regional cultures comparatively and historically, we calculated how distinctly a regional cultural community stands out on the Web, and the strength of its online activities.

In general, we find that geographical regions where people speak languages not widely spoken elsewhere (such as Japan and Korea) are the most distinct online cultures; regions with geographically dispersed languages (such as Spanish or Russian) or those of multilingual geographies (such as India) less so.

Our study suggests that the Web, when mapped based on its usage, does not have its core in the West, but is a mosaic of online regional cultures that associate with physical places.

In such maps, the Internet is becoming more decentralized, or to be more precise, de-Westernizing, as more users from disparate cultures are taking over its topography by bringing in their own cultural identities. Between 2009 and 2013 the Web witnessed a gradual process of “de-Americanization”; the cluster corresponding to the U.S. has separated from the “global” websites such as Twitter and Instagram – primarily user-generated websites, which are neither centered in North America nor on the English language.

In this process, the American sites have taken their own “corner” of the Web, just like other online regional cultures. Simultaneously, non-Western online cultures have strengthened, especially those linked to Brazil, Russia, China and India. Unsurprisingly, in these places, local Internet industries are thriving and domestic content is flourishing.

Compared to the prevalent technological Internet maps, our user-centric maps from 2009 to 2013 challenge, rather than reinforce, the existing concept of an Internet anchored by Western knowledge, norms and activities. They encourage the (Anglophone, especially) general public to confront the narrow online world with which it is familiar. Further, the trend captured by our maps may encourage Westerners to refresh their own preconceptions by exploring the vastly heterogeneous cyberspace.

These user-centric maps also inform policymakers about how better to empower the global South. Technical connectivity alone is not enough. For online regional cultures around the globe to strengthen, users must be able to build and shape the content they find appealing. For this to happen, local governments need to introduce civic, economic and social opportunities with new technologies. Left to a market dominated by West-based transnational corporations, the global South may not achieve healthy domestic Internet landscapes and online cultures.

The Conversation

Angela Xiao Wu, Assistant Professor of Journalism and Communication, Chinese University of Hong Kong and Harsh Taneja, Assistant Professor of Media and Communication, University of Missouri-Columbia

This article was originally published on The Conversation. Read the original article.


Do event outcomes affect how we rate their coverage? CNN IBN and US Elections 2012

Media feeds on elections. Counting and results declaration is when news becomes a “mass audience” genre, and the US Election results night more globally so is such a mass audience event. As I write, Obama has just been declared victorious. Apart from watching the coverage on BBC, I have been following the conversation in India on the topic on Twitter.

Rajdeep Sardesai, (Editor of CNN IBN , an important national cable news network in India) just tweeted that he was thankful to people for liking their coverage , on the basis of unsolicited instant feedback he received on twitter. Sounds very good!

BUT I couldn’t help ‘speculating’ if this enthusiastic support of the coverage had something to do with the outcome ( Obama’s Victory). Let me explain. Although a small fraction of Indians are really interested in US Elections,  many of them talk about it on twitter and also watch English language news. On average, they support Obama (they are young, globally mobile, have US connections mostly in North East and West Coast). So perhaps a Romney victory would not make them jump with joy. Then they (Obama supporters) may have liked the coverage a little less. Or at least not expressed unsolicited admiration.

I am not taking any credit away from Rajdeep or IBN. I am merely fishing for people’s thoughts on a hypothesis:An outcome of an event favorable to us makes us think more favorably of news media’s coverage of that event. So by corollary – On average, Americans who watched the 2012 elections night unfold on CNN (more liberal)  may rate CNN’s coverage more favorably than FOX News’ viewers (more conservative) would rate its coverage . Thoughts, comments?

Facebook ads impact is just more ‘measurable’ than TV ads, not necessarily ‘effective’

They have been talking for a while about the death of the 30 second commercial. And social media were not the first technologies that inspired such claims.Yet, the size of Facebook with its tightly connected ‘anatomy’ has made these claims more vociferous. (The average degree of separation between any two people on Facebook is just 3.8, means that just in four steps a message can actually travel from anyone to anyone else, on average).

A recent post exemplifying this refrain, claims that Facebook promotion ( paid ads + all other kinds of messages) deliver 70 times more impressions ( eyeballs that see the ad) for the same cost than an ad on prime time television in the UK ( on coronation street) and 180 times more impressions than the same campaigns in the US . They argue that a well run Facebook campaign can generate 4 to 5 billion impressions versus the TV campaign that reaches a 100 million for the same duration. The claim is problematic due to the way impressions are counted for the two media.

The Facebook number is reached by counting the people touched directly (who friend or like the brand pages), their friends (who see it in their feeds, comments or messages) and friends of friends of those directly touched. In other words upto 3 levels removed from the campaign. For TV they simply count the number of eyeballs that saw the commercial ( the way marketers have computed ad effectiveness for TV ads for decades). There seems an implied an assumption that prime time television advertising only reaches those who see it directly. However just as Facebook page activity, television advertising too inspires conversations and recommendations through word of mouth that can make it reach a significant multiple of that 100 million, especially over the course of the year. In fact, the worth of one such conversation or recommendation (including on Facebook) ‘may’ be worth a few hundred Facebook impressions, a miniscule number of which may actually count themselves. However, obtaining any similar accurate measurement for the extended impact of TV advertising is next to impossible. However, his does not mean that there is no extended impact.

On first degree impressions, TV campaigns clearly achieve much more than the biggest Facebook pages( even Coca-Cola has just 53 million likes). Next, it may be worth pondering why these extended 4 to 5 billion impressions may not really count for a lot. Consider that for each person who likes a brand page, that activity appears on some laundry list of notifications of 16% of their 190 friends (Currently 16% of FB users are able to see such posts, and the average number of friends on FB is 190). A miniscule number of these impressions are likely to be welcomed or even consumed. Those receiving them may be just users who haven’t optimized their news feeds enough to avoid such notifications. Finally the post claims that Facebook itself did not have any data on the third level of sharing (to friends of friends of directly touched). This will be presumably much lower than the 16% on level 1. Think of chance that a friend of your friend is interested in the activity by their friend ( who is not a mutual friend) with a campaign or brand page. Since these data were not available, they used some triangulation to estimate some multipliers, and do not mention either the multipliers (no of people reached further by each person) or the attenuation (what % of people’s friends will receive such a message).

Of course other points worth considering are 1) why people like brand pages on Facebook (to get discounts!) and 2) how did they learn about these brands (surely not exclusively through FB campaigns or noticing friends who liked FB campaigns, but perhaps through prime time advertising and conversations with friends and family). These concerns make the value of these impressions questionable, even if their sheer volume is estimated properly. Of course the same can be argued for TV advertising that people do not want to see, but end up watching.

In sum, the key point here is that just because Facebook due to the traces users leave, allows one to measure how a campaign circulates beyond the first degree impressions and TV does not, one cannot conclude that Facebook campaigns are that much more cost effective than prime time TV.

Quick Updates on the NASA Diwali Image Phenomenon : Socially Constructed Thruths and the Persistence of Popularity

October 26th ( 2:40 Pm Chicago Time Less than 48 hrs on) ;66, 248 Shares  ( three times over 22,000 before)

October 25th ( 14 hrs on)

Remember in my original post I had speculated that as India would wake up , we would see an exponential rate of increase in the alternate truth. Yes the 1000 odd shares on that thread have grown to 23,000 ( in 12 -13 hrs) in that thread ( potential audience of 7-8 million just from this source).

Interestingly, there were many counter-voices in some of the threads that circulated who argued with similar logic as to why the image was not one of Diwali.Some of these even pointed out to the original website of the image confirming that it wasn’t a Diwali image.  Yet the Euphoria continues to persist.

To me it reflects that on Social Media, once certain facts get socially constructed, it is very hard to let alternate facts percolate. In other words, their popularity will continue to persist, even though they have been proven wrong! There seems little room for course correction.

Social Media and The Problem with ‘Socially Constructed Truths’ : Why all Believed that the NASA Picture Was Clicked on Diwali

India and her diaspora worldwide was excited about Divali, (the festival of lights), and for good reason. A day before Divali,  October 25 2011 3:30 pm CT, I noticed that a map had become a ‘Facebook hit’. In less than 5 hrs,  the link had first appeared it had already been shared by 1000 odd people ( giving it a potential audience of at least 200,000). Remember I am reporting figures for only 1 thread that I could trace.

An idea took form that the picture was an illumination of the country on Diwali night. Of course most people related the distinct, yellow, green and red dots to the colors of fireworks.  When it popped on my wall ( as the 950th share) I immediately questioned that Colombo(Sri Lanka), Lahore (Pakistan) and Delhi could not be celebrating divali with the same ‘illumination’. (Some even speculated these were the remnants of Hindu legacies in these cities!) Also why would people in certain areas burst green colored fireworks when the rest of the country was bursting yellow.

Leave that, it was easy to explain the map if one understood some   development parameters and grade school geography of the region ( Electrical Connectivity and Population Density, Location of the cities, Mountains).

So I decided to question how could the illumination be different from any other night (when the Power grid hadn’t failed)? I commented with my reservations on the wall of the friend who had shared this with me. Seeing some merit in my reservations, he digged the original source of the picture and indeed found it had nothing to do with Divali. He found the original source on FB itself, when someone else like me ( a friend of a friend of his) had commented on their common friend’s share of this post with similar concerns and found this link.

But our counter-currents are far and few between. That they are the complete truth does not matter . As I write, the Indian diaspora on Facebook believes that the map is indeed a NASA picture clicked on a Divali evening. I am sure in a few hours when Indians wake up this ‘socially constructed truth’ will spread further and more widely. ( See my update tracking the numbers)

To me this is the grim reality of online social media!

Happy Divali Regardless

Lets Abuse TAM Data: Indian TV Industry

An example that illustrates how one should be careful when reading television audience ratings data.

I read an ad in Mint about UTV-Bloomberg claiming that they have grown to no. 2 in the English Business News Space.

Lets Abuse TAM Data

Surprised I decided to look closely:
Well in the Male CS 25-44 (all SECs) Segment All India market they had slightly greater channel share (relative) than NDTV Profit and ET NOW according to TAM. For the uninitiated channel share is the product of no of people who watched at least one minute ( reach) and the average time spent by them.

I have three issues with this kind of abuse of TAM numbers :
1. The target for English Business News Channels is much smaller than the one used in the ad. Perhaps CS 25-44 M SEC A and B in 10 lakh plus towns would be more plausible.

2. The use of relative shares is again a misleading measure. In reality the absolute of share of viewing of these channels is minuscule to total television viewing so how many combined minutes of television viewing does this really mean?

3. The third is with the use of TAM for such data.English business news is a very niche segment.
a.) This genre is consumed by respondents who are hard to recruit for high involvement research like TAM.
b). Besides consumption is outside the home ( at public places, offices etc).
c.) The sample sizes for these channels is so low that these fluctuations may be due to sampling errors.

But the saddest part is that our media planners will continue to spend advertising money based on these irrelevant numbers and think they are making scientifically correct decisions.

I hope students of media/ advertising /marketing can learn this before going to the industry.

Hail the web and the blogosphere: Finally my blog has had ‘real impact’

I have been updating this blog somewhat regularly now. I have often wondered whether I write this weblog for a particular audience . Perhaps the diverse nature of posts (yes I go all over) suggests that I don’t. So why do I then bother to use tags, categories to make my blog more search friendly. However I have been proven wrong and I am thrilled!
Regular readers (sigh!) will remember a few weeks ago I wrote about my rather funny but adorable roommate from Russia, Mikhail Safronov on this blog. I had forgotten about this post altogether. A few hours ago Mikhail came running to my room, excited and blurted out loud ” Harsh, A real Bhains ki Aankh has happened. My parents saw your blog and now are making fun of me and my mom is a little worried that Barack Obama should not see this blog !”

We were amused that how did they get to this post. I used the word-press dashboard to see what search terms had been used recently to get to this blog. And found that someone on 2nd Feb 2010 had used the phrase “northwestern university safronov” to get here. So now Mikhail knows that his parents (or their colleagues) try to stay more informed about their son than his phone calls can inform them. Maybe the economist will elevate his views about a PhD in Media, Technology and Society!

So a piece of content produced in the USA which perhaps appealed to only one unit in Russia – Mikhail’s household – reached there in less than two weeks after it was produced. The recipients had to be just more active in seeking what they wanted to read/view. I hope Mr Safronov gets to read this one too!