Protest Algorithms: Outrageous stuff modern security teams need to know

Introduction

Born in the 1970s I’m the youngest of three brothers. As a child I would protest most of my parent’s decisions. Not because they were right or wrong but because protest got me attention and I wanted attention. I would stomp my feet and throw tantrums to great effect.  

If I was child today, I’d still want attention. But I think the way I’d get attention would change. I’d still scream and shout, but I’d also pester my parents online. Because online is built for attention seekers like little me.

This principle is not lost on adults who want attention. They too protest both on and offline. However, what works online is different to offline.

Modern security teams need to know online, algorithms impact street protests.

What’s an algorithm?

An alogirthm is a set of rules followed in computing calculations. They determine the value of posts on social media and as a result, how much profile a post receives. A post that meets an algorithm’s rules is ‘liked’ and promoted on social media. Therefore, street protestors want algorithms to ‘like’ their posts. 

The holy grail is to have a post go ‘viral’. Algorithms take a post and promote it so much that huge numbers of people outside your connections see it. This holy grail could be labelled a ‘protest algorithm’.

Different algorithms ‘like’ and ‘dislike’ the same post

There is not one but hundreds if not thousands of algorithms that review a post. Each algorithm has different goals and compete against one another.

For example, the ethics department’s algorithm likes safe posts that don’t infringe terms and conditions. Whereas a commercial department’s algorithm likes posts that engage lots of people because that’s good for ad revenue.

Therefore, a controversial post, such as an image of people glued to the M25 may get great engagement (it’s novel) but fall foul of the ethics department’s algorithm (it’s dangerous).

Critics claim these competing algorithms mean no one, not even founders of social media, know what posts will be popular.

Two things we know about algorithms

1) Algorithms are not blind

Algorithms that serve social media are not blind, they are not neutral space like a website blog. Because social media (Twitter, Facebook etc.) are private enterprises their algorithms are motivated by revenue and profit.

Think of social media as a private shopping mall that has successfully moved street protest from the public square to their private space.

Social media, like shopping malls, have lots of upsides and a captive audience. However, just as Westfield controls what protests occur inside their malls, so social media controls posts inside their network.

2) Algorithms always like….

Algorithms always like posts that get engagement.

Because social media is free and ad funded, its algorithms are weighted toward posts that get user’s attention and keep people online. The more people online, the more they view ads and the more social media companies can charge advertisers.

This has the following consequences:

‘The machine-learning models that maximize engagement also favor controversy, misinformation, and extremism: put simply, people just like outrageous stuff.’

Emma Hao, Technology Review, MIT

Protest algorithms like outrageous stuff: What does this mean for security teams?

Protestors want their social media posts to go viral. Therefore, they may do ‘outrageous stuff’ to get engagement. Outrageous stuff means doing something new. Because new gets attention. As a result, the security team should plan as much on what protestors have done, as what they could do.

In practice, this means imagining how a protestor may use your organsation to grab attention.