Social Intelligence Stories That Caught Our Attention: Volume Twenty-Three

In this week’s round up we have, a philosophical look at the way photography has been re-shaped, how frequently emoji’s are used, Kantar and Frrole’s new partnership, bad data and bad decisions, fooling deep learning AIs, Instagram data scraping issues and being kicked off Facebook and Instagram, needing to do better around detecting propaganda, Reddit censorship to attract advertisers.

Social Photo: What Does it Mean to be Your ‘Authentic’ Self When Everyone is Watching?

Interesting article that has a philosophical look at the ways photography has been re-shaped, augmented and re-imagined in online spaces. It questions when images become more accessible and omnipresent, how does it change the face of photography, or underscore previously-held truths?

There’s also a question on whether the internet hasn’t actually changed us that much; rather than online spaces operate as auxiliary amplifiers of long-held beliefs about self and society.

Read the full article here.

How Frequently Are Emoji’s Used?

Unicode Consortium has run research to identify how frequently emoji’s are used. Here’s what they found?

Read the full article here.

Kantar and Frrole Partner to Launch TGI Social+

In India, Kantar and Frrole have partnered to launch a service called TGI Social+, which combines consumer profiles from Kantar’s consumer survey TGI with social media analytics to provide a ‘holistic’ understanding of online consumers. A big congratulations to Frrole on this partnership.

Read the full article here.

Bad Data Series: Blockbuster Decided Not To Netflix and Chill

In the bad data series, NetBase explores what happens when brands rely on bad data. They warn that the wrong social analytics solution, offering inaccurate insight, could result in brands misunderstanding potential market opportunities – and missing out on category-defining moments in time.

Read the full article here.

Why Deep-Learning AIs are so Easy to Fool

Interesting read across at Nature on why deep-learning AIs are so easy to fool. Great background and examples on where AI has been fooled into misreading objects. From implications to self-driving cars to auto-detecting cancer and other visual objects.

Read the full article here.

Facebook Reviews All Official Marketing Partners Over Instagram Data Scraping Issue

In August, Business Insider reported that Facebook was reviewing hundreds of its official Facebook Marketing Partners after reports that companies were downloading more data from Instagram than the social network allows. These companies included Hyp3r, Cambridge Analytica and Brand24.  More recently, it was reported by NewsBeezer that Brand24 and its founder Michal Sadowski had their accounts on Facebook and Instagram blocked.

Read the Business Insider article here.

Read the NewsBeezer article here.

Twitter Needs to Start Exposing the UK's Murky Online Propaganda

Wired report that both Twitter and Facebook are getting better at disclosing inauthentic accounts and state-backed disinformation campaigns, but they’re largely silent on those emanating from the UK and the USA.

Read the full article here.

Reddit is About to Further Censor its Communities to Attract Advertisers

I love Reddit as a data source, but this article from Reclaim the Net suggests that Reddit are censoring conversations to bring the platform up to the arbitrary standards imposed by political and media power-centers.  There’s an interlinked relationship between censorship and revenue growth – the article questions how far this policy will go and how much it will hurt communities to increase profit.

Read the full article here.

‍