## Colophon %% title:: All The World's A Stage type:: [[output]] url:: file:: %% tags:: [[performative politics]] [[professional incentives]] status:: [[brewing]] created:: 2021-10-19 ## Content # Most of the world's a stage... most of the time ### Prevalence and Percentiles On November 9th 2021, Facebook released a number of transparency [reports](https://about.fb.com/news/2021/11/community-standards-enforcement-report-q3-2021/). What they highlighted was the reduction in 'prevalance' metrics across categories like - - Hate speech - On Facebook, down to 0.03% in Q3 2021 from 0.05% in Q2 2021 (that means for every 10,000 views either 3 or 5 were views of content classified as hate speech). - On Instagram, 0.02% for Q3 2021 and this was the first time they reported this metric. - Violence and incitement - 0.04-0.05% on Facebook - 0.02% on Instagram - Bullying and harassment - 0.14%-0.15% on Facebook - 0.05-0.06% on Instagram I get why Facebook is highlighting average prevalence across its very large platform(s). But as Samidh Chakrabarti pointed out in a [thread on Twitter](https://twitter.com/samidh/status/1450105907531059216), one needs to look at percentiles to get a better sense of the distribution and effects.[^1] ![[Prevalance Performance 1.png]] ### Performance, Politics and Escalation Another area where Facebook likes to cite average prevalence type of numbers, is for the amount of political content. In posts dated [November 2020](https://about.fb.com/news/2020/11/what-do-people-actually-see-on-facebook-in-the-us/) and [February 2021](https://about.fb.com/news/2021/02/reducing-political-content-in-news-feed/), it cited a 6% number for the amount of content political content **in the US** *(In other parts of world? We just have to guess and hope that we'll know someday. Other platforms? Same answer for now - we don't know)*. > political content makes up about 6% of what you see on Facebook We're not sure how this content is classified, even if my gut tells me that this figure probably understates the actual amount of political content. But, just like the average timeline prevalence of content classified as hate speech, incitement, bullying, etc. may not be particularly useful to understand the effects of its distribution [^2] - I think a similar logic applies for political content, across any platform. As a metric, it doesn't tell us much. And, whether this number 6% or 10% or 20%, it may not matter much.[^3] Why? Because politics has always been performative. The difference in average prevalence of political content probably doesn't lead to vastly different outcomes - within a certain order of magnitude. Even distribution though percentiles may not tell us much given that it is such a broad category. It likely matters more (this is speculative) that those who consider themselves newsworthy feel the need to perform more (for longer hours, across more places) than ever before, and what effects these resultant dynamics have.[^4] [^5] **Most of the world is a stage for most of the time (if not all).** #### So what? Yes, politics has always been performative. And that performative nature has probably always tended towards being escalatory. But the pace and scale of these escalatory cycles were limited - the public sphere could only carry so much information (i.e. limited supply) which meant: 1. The pool of candidates one had to stand out against was constrained by size and geography. 2. Opportunities to build a public profile were fewer and required greater investment of time/effort/resources. 3. Information consumption was restricted to when it was available across a finite and limited set of venues. The number of ideas/viewpoints, and the speed at which they moved/gained momentum is thus affected. 4. While some form of near-instant communication has been around since the telegraph, its capacity to carry information was capped, or entry barriers were high (TV, Radio), therefore, multi-point communication was limited and developments were slower. *Related [[Velocity of society]]* That's clearly changed. Professional incentives now require standing out from an ever-expanding pool of competition, not only in terms of people, but also viewpoints and ideas (irrespective of whether they are ideologically aligned or opposed) in always-reducing timeframes, across many possible venues (physical and digital ones) pushing political speech to greater extremes. These dynamics play out whether it is a local-level politician trying to gain fame / build clout / break through the clutter, or an established regional / national figure. Of course, the scale of effects will differ. This may explain why videos containing bigotry and hate crimes are willingly circulated, or why birthday wishes are a Twitter affair, or why it seems like political leaders are on the hunt for that 2 minute video they can upload to YouTube and share on Twitter and Facebook, or why the Finance Minster quote tweets and tags a Board Chairman instead of a direct call (which may also have happened), or why politicians dunk on each other on Twitter, or why so many letters - no matter how inane are posted across a range of platforms, or why relatively mundane events are used for different sorts of political grandstanding. Shifting professional incentives in related fields such as news media (not in scope for this post) further amplify these performative tendencies. All this ultimately affects the kinds of conversations we have about governance and regulation - and indeed, what we get. And though it didn't need to be this way, the escalation itself has escalated ... into a race to the bottom [^6]. [^1]: For another post, but I wonder if displaying these metrics across groups / profiles for content they post would be useful. Note that I haven't fully thought this through. And of course, before this happens we need a much better understanding of how these classifications are made. I don't think we have that today. [^2]: I'm saying distribution not demand because it seems misplaced to claim that all content on timelines is a result of demand - even if user choices do shape what appears. [^3]: Of course, degrees matter here. 6, 10 or 20% may not matter much, but 50, 60 upwards will. [^4]: This is, by no means, restricted to politics, I am just drilling down on that aspect since evolution of political discourse is likely to have faster effects, since it may affect governance, regulation. The choice of faster v/s greater is deliberate since we may not see other effects for a number of reasons. [^5]: It raises the question of whether a better understanding of the supply side of political content across categories like hate speech, incitement will be useful. Same caveats as [^1] apply. Also worth noting that this could limited utility as it is unlikely to be cross-platform and it is possible for actors to shift to a model where they are not directly performative, but engage in certain rhetoric by proxy. [^6]: While I am always trying to track the intersection of the performative aspects of politics and social media, and the role of incentives, I should add that the trigger for this particular post was this [tweet by Ariel Bogle](https://mobile.twitter.com/arielbogle/status/1460364302368448514) ![[Ariel Bogle Tweet Escalatory Politics.png]]