What is the algorithm?

What is the algorithm?

What you see on your social media feeds could be about to change, and it’s all to do with the algorithm - a tool used by social media companies to show us what they think we want to see. How it impacts our society has increasingly come under scrutiny, and now the government has revealed draft legislation that could give us the option to opt out of it. So in this Squiz Shortcut, we’ll dive into:

• What the algorithm is

• How it got this way

• And how it might change

🙋🏻‍♀️ This newsletter was written by Andrew Williams and Larissa Huntington

Prefer this in your ears?

Listen to our podcast 🎧

Listen time: 12 minutes

Squiz the Shortcut

Let’s rewind a bit… How did social media work in the beginning?
You used to be shown photos or updates only from the people you followed, and they came in chronological order - but it’s been a long time since that was the default model.

Why did it change?
The main reason is that as social media grew in popularity and people started following more and more accounts (and accumulating more and more friends), it became unwieldy - you could easily miss things you cared about if they were buried under a bunch of other posts. So the big platforms - Facebook, Instagram and the like - began searching for a solution. That’s where the algorithm comes in…

So what actually is the algorithm?
An algorithm by definition has nothing to do with computers, or tech - it’s simply a list of step-by-step instructions for doing something. Technically, a recipe is an algorithm. The word comes from al-Khwarizmi, a Persian mathematician, whose name gets Latinised as Algoritmi - so it’s a maths thing, really, before anything else. 

Where do tech and social media platforms come in?
Social media companies brought in their own instructions - their own algorithm - designed to boost engagement with their platforms. 

What do you mean by ‘boost engagement’?
It means getting more people to actively interact with social media content rather than just scrolling past it. So, instead of just looking at a post, users take a measurable action like sharing, commenting or liking it.

What change did the algorithm make?
Originally, it worked by showing you things based on accounts you interacted with a lot, or the types of posts you liked - for example, if you tended to like photos more than text posts, you would be shown those more often.

Which app was the first to change?
Facebook was the first to move away from chronological updates in late 2009. Its system for doing this was called EdgeRank and it had 3 main factors - affinity (your closeness to the person posting), weight (what kind of post it was - photos and comments scored higher) and time decay (how recent it was).

Did EdgeRank stick around?
No… Within a couple of years, Facebook had dropped the system. It was a bit clunky and needed engineers to set those weights by hand, and it changed to a model largely designed by machines… But that made it harder for Facebook to explain why it was showing you what it was showing you - because on some level, it didn’t know. But basically, it was tuning things so you would enjoy the platform more and stay on longer. 

Did the other platforms follow?
Eventually… Twitter and Instagram held out for a while, but they both made an algorithmic timeline the default a decade ago. And despite a wide blowback to the decision from many users - there was even an #RIPTwitter hashtag on Twitter itself - the platform said it worked in terms of boosting engagement. 

Couldn’t people just opt out?
They could, but hardly anyone did. At the time, Twitter said the number of people opting out was in the low single-digits, percentage-wise. So, while lots of people complained, very few people actually did it.

What about TikTok?
Things changed again when it came along - its whole design was about showing you things you didn’t know you wanted to see. It had 2 main pages, and its very popular For You page was entirely built on the premise that the algorithm would show you things you liked. After that, algorithm-driven feeds have only become more popular. 

So the algorithm shows me things I’ll enjoy?
Not necessarily… Mostly, they’re geared towards keeping you engaged - and that can mean many things, including making you feel angry, scared, or curious. There’s a difference between liking something and not being able to stop watching it. The main goal for most algorithms is the second one.

How does it work out what’ll keep me there?
Without even realising it, you’re training it constantly. If you watch a video through to the end, you’ll likely get more similar to it. If you interact with a certain type of video, you’ll probably get served more of those. On the flip side, if you skip past certain types of content, you’ll probably see less of it. 

Is it only looking at what I choose to watch?
No… The algorithm takes all of the information you feed it - intentionally and unintentionally - and then predicts a probability that you’ll engage with each video that comes along. That includes what you search for, what you hover over, and also what other users like you (so people who engage with the same content as you) enjoy. So your activity is also shaping what other people see too. Exactly how each company does it is another matter - they keep that tightly held in-house.

Got it… So where does this become a potential problem?
It becomes an issue when the constant need to engage an audience risks radicalising them. Let’s say you get served a video that shows you something shocking or violent. You didn’t ask to be shown this, but you might not be able to look away. If you watch lots of the video, the system learns from that and then shows you more like it.

Is there a push for change?
Yep… Consent advocate and chief executive of Teach Us Consent Chanel Contos has spoken about this pattern as a reason to give people the choice to opt into the algorithm - meaning you’d have to ask to be served content you didn’t explicitly choose to follow or ask to see. 

What does she reckon is happening?
She says that the algorithm - not entirely intentionally, but as a byproduct of chasing engagement - is feeding misogynistic, violent and hateful content to young men. She argues this is contributing to higher rates of sexual violence amongst young people.

Got any examples?
Yes… Content served up by influencers in the manosphere space like Andrew Tate, alphas, looksmaxxers, and certain fitfluencers are all targeting young boys and men - often to boost their own income from advertising dollars. And asking social media companies to just change the algorithm to avoid feeding that type of content isn’t an option - the algorithms are so big now that most social media companies can’t even explain how they work.

So what is the federal government proposing?
Draft laws put together by the Albanese Government would prompt users to opt out of having the algorithm decide what they see. It’s not going quite as far as Contos is calling for, but it’s a step in that direction.

Is there any pushback?
It’s early days and still a moving story, but it’s worth noting that there is some concern from the business community, particularly from small businesses that rely on the algorithm to reach customers. Communications Minister Anika Wells was asked about that, and she said businesses would still be able to reach customers because opting out would be optional, and being offered the choice doesn’t mean people will take it, as Twitter found in 2016. 

Where do the other political parties sit?
The Coalition has raised some concerns too - arguing the laws could be tantamount to censorship, and that the algorithm is just showing people what they want to see in the first place. And the Greens are calling for big tech companies to restore the original settings on social media apps as a default in Oz so that people have to opt into the algorithm, instead of it being in reverse.

So what would actually get switched off?
That’s the bit nobody can answer just yet. If you use Instagram, you’ll know that there are several feeds now - reels, stories, posts - and they all use algorithms to determine what you see. Exactly what we’d end up with is unclear.  

So what’s the bottom line?
Wherever this legislation lands, it’ll be balancing all of those competing interests with the goal of reducing the negative impact the algorithm might be having on us as a society, particularly when it comes to young people.

Onto our Recommendations

Reading: This article in The Conversation is a good explainer about the whole opt-out discussion.

Reading: And this piece from the ABC is an interesting read where the reporter tries to outsmart the algorithm - without success…

Recent Shortcuts