The Political Economy of Social Media Influence Operations in the Philippines (and Elsewhere)

And the final speaker in this AoIR 2023 session is Fatima Gaw, whose interest is in the political economy of social media manipulation. Thus far we only have a very partial knowledge of this political economy; there is work focussing on bots, trolls, and fake accounts, using big but limited social media data, or occasionally doing ethnographic work. There is also much reliance on secondary sources. Further interdisciplinary methods combining these and other approaches are needed to determine the scope and scale of this political economy.

A starting point here may be the covert campaigning by political influencers. This involves the demand side (clients and their political machinery), the products (influence operations and campaigns), and the supply side (the influence operations industry and workers). At the centre of this are the influencers, who perform ‘gray’ political influencing and are more visible and higher-ranking than bots and trolls.

Fatima’s focus here is on the Philippines, and specifically the Filipino election of 2022 during with Ferdinand Marcos Jr., son of the former dictator, won power. Countries like the Philippines are also a testing ground for influence operations that are then rolled out in the WEIRD nations of the Global North. The project engaged in ethnographic and computational work to analyse the operations of influencers, and analysed these in terms of network, behaviour, and content dimensions; it focussed on Facebook, Twitter, YouTube, and TikTok.

This mobilised some 16 categories of analysis across the three dimensions, and was especially interested in influencers that behaved in unusual ways; it worked with a dataset of some 44,000 influencers and scored them across each of the dimensions and categories, and clustered them based on similarities, to identify the outliers. This identified some 1,400 accounts that were suspected of covert campaigning, with more than 500 each of these present on YouTube and TikTok, in particular. (Facebook had the most influencers overall, but only 200 of these outliers.)

’Network’ means different things here, incidentally: on Facebook, sharing; on Twitter, retweets; on TikTok, hashtags; on YouTube, recommendations from high-prestige channels. This also manifested in very different network metrics patterns across these platforms. Behavioural findings were also diverse across the platforms; while there was more similarity in content patterns across the four platforms.

Overall, then, this produced some eight types of covert political influencers: amateur commentators and curators, hyper-partisans, stans, trending influencers, alt-news and entertainment media, mainstream popular influencers, and polarisers. This, finally, also enables a further analysis of the economic models that might underpin such activities: some may operate under a pay-per-post model, for instance, while others may be on the payroll of specific political campaigns.

All of this is also covered in an Internews report on the Political Economy of Covert Influence Operations in the 2022 Philippine Elections, incidentally.