This newsletter is published at techpolicy.substack.com
An excerpt from Edition 17 is reproduced below.
Hate is Trending, Trends are Not
Hey Prateek – you spoke about hate just last week!
Well, sorry, I did tell you it was unlimited, didn’t I? Unlike 1984’s 2-minute Hate 2020’s is the 24-hours a day, 7-days a week variety (I cannot claim originality here, this was inspired by Maelle Gavet’s piece in FastCompany).
Over the weekend, there were violent protests in Malmo, Sweden. Anwesha Mitra wrote an explainer in The Free Press Journal. I won’t go into details of the ‘on-the-ground’ politics because I am far from being an any sort of expert in Swedish or Norwegian politics. What I did see though was that as these events were taking place – terms like SwedenRiots, WeAreWithSweden and NorwayRiots were trending on Twitter in India. In the early phases of this, as I was trying to find authoritative news sources to figure out what was happening, even searches on google/bing pointed to mainly Indian publications (yes, I use Bing sometimes).
Of course, on the face of it, there is nothing wrong with this. Twitter is a global platform, the world is global village (isn’t that the phrase?), and India is a free country (It is, no?). There is nothing to say that somebody can’t share their opinion(s) on something that happens in different countries. What they choose to say, has some relevance though.
From when I was looking into Twitter activity around the time BLM/George Floyd protests started in Edition 6:
like USARiots and USAonFire. In fact, on both these hashtags, of the accounts that actually publish location information on their profiles, the top ones were from India. Even the accounts with most number of tweets using the hashtag appears to be be from India. Now, this is very different from saying that most or even a substantial portion of the activity was from India. It was still higher than I expected.
See? Nothing wrong, nothing even out of the ordinary. Also, from earlier in the same paragraph:
I spent some time looking at the hashtag USA riots, because I noticed a similarity in anti-protest narratives from HK, Anti-CAA, J&K and now George Floyd, “rioters”, “looters” etc. I was also surprised to see a lot of RW seemingly Indian (or Indian-origin) handles active on hashtags like USARiots and USAonFire ….
That pattern persisted in the Sweden, Norway case too. I wrote about it for TheWire, but here’s the gist.
- Using a combination of Hoaxy, Twitter Search and APIs I tried to find what kind of messages gained traction.
- Hoaxy clusters indicated that the accounts which a got a lot of traction typically associate themselves with anti-minority content.
- It appeared that a substantial amount of the activity was driven by accounts that listed themselves as belonging to locations in India.
- Hashtag clouds indicated many references to Bangalore violence/riots, Delhi Riots, Kashmir, Poland, Islam, Jihad and ‘Just Saying’
- The top 1% of tweets accounted for 55-72% of the activity.
- The content of the top 1% was largely anti-minority in India.
- There was 1 account that purportedly belonged to a caucasian Swedish citizen (which got a lot of engagement on one of the hashtags) while one of its first tweets was in Hindi/Urdu in Roman script.
I’ll include the Hoaxy clusters and tag clouds here, but all other content is in the article.
It seems like it only last week that I was talking about Dangerous Speech (that’s because it was) so I won’t belabour the point. I’ll just include the definition: *‘any form of expression (e.g. speech, text, or images) that can increase the risk that its audience will condone or commit violence against members of another group.’*
“Whether something comes under hate speech has to be defined by a consistent policy and has to have neutrality of ideology,” the official said.
“This debate around hate speech is skewed because what constitutes hate speech or otherwise will be determined by India’s rules and regulations and constitutional frameworks, not by community standards of a particular social media platform. It also needs to apply uniformly,” Malviya had said.
Now, that brings me to another point – hashtags and trends. Hashtags/Trending/Most Viewed type of sections enable people identify what is popular and also to coalesce around topics. Except, these have always been possible to game. All the way back in 2010, a writer at DailyBeast demonstrated that it took ~1300 mails to break into New York Times’ Most Viewed section. In 2017, Pranav Dixit wrote about the use of Twitter Trends for political propaganda in India. And Kiran Garimela tweeted recently about some upcoming research.
Go here for the complete edition.