This is the second part of a three-part series - start with Part I here.
Find Part III here.
Part II - Algorithm Feeding Design
So you couldn’t resist. You’re on TikTok. It’s free, how bad can it be?!
You open it up.
One video plays across the whole screen.
Ok what do I do?
The video keeps playing on repeat.Swipe.
Ah got it - keep swiping.
Ha - that was hilarious.Like. Swipe. Like. Swipe. Comment. Share.
Before you know it you’ve spent an hour watching ten-second videos. If your lost hour feels outrageous, consider that you’ve unconsciously fed the TikTok machine 360 signals.
Damn. How?
TikTok’s algorithm friendly design feeds the machine of this century's biggest human shortcoming - our attention span.
1. Feedback Volume - Who knows you better than your Family?
It’s no secret that TikTok leverages our short attention span.
We get the dopamine hit after watching the short video, which keeps us hooked to watch the next one. The app loves it as you keep consuming more and more content. Contrast that to YouTube. The average Youtube video is 11.7 minutes long. By the time you finish one Youtube video, a TikToker watches 46 videos.
Your sheer volume of content consumption is eye-watering. But what feeds the machine is the “algorithm friendly design”.
Two things stand out in the design:
Micro-interaction - you can only look at one video and you have to do something to continue.
Full feedback palette - you can provide positive, super positive, as well as negative feedback.
The concept is easiest understood when you look at other social apps.
Facebook and Instagram are famous for their infinite scroll. They make us consume even more content than TikTok. The infinite scroll whilst amazing to get us hooked, prevents the algorithm from learning. It shows us 2-3 pieces of content at any given time but only learns when we stop scrolling and interact. And it only knows when you like something. It only has a positive feedback loop, it never knows when you don’t like something. Social apps have decided against dislike or hate buttons for good reasons.
So let’s contrast that to TikTok.
Every single video you watch, you have to do something. You can:
Like - positive 👍
Check out creator - positive 🙂
Check out videos with same music - positive 😊
Share - very positive 🔥
Recreate the video with the same music - fireworks of positive feedback 🥳
Swipe to next video - small negative feedback 😕
Swipe to next video within a split-second - strong negative feedback 😦
Additionally, creators categorise their videos which feed back into the algorithm.
Whatever you do, you give TikTok data on what you like and dislike.
2. Interest Graph - Your True Soulmates
The fundamental difference is that on Facebook you build a network of friends. Whereas on TikTok you’re building a network of people with the same interests.
Why does this matter?
The algorithm learns about who you truly are. It surrounds you with content from people with similar interests.
Research shows after 10 interactions the algorithm knows you better than a colleague. 70 beat a roommate. 150 beat a parent or sibling. Finally, 300 beat a spouse. So, TikTok knows you better than your spouse within two hours. 🤯
If you’re still not convinced, look at one of your friends or a stranger’s FYP (For You Page = Home screen = Newsfeed). I can guarantee that no two FYP pages will be the same and it will surprise you what you see.
We all had that “nobody cares what you ate for lunch” moment on Facebook / Instagram / Twitter.
TikTok’s magic is to find the people who care.
It’s a way to build an interest graph without having to follow anyone. You can skip the long and painstaking intermediate step of assembling a social graph. You get to jump straight to the interest graph.
The magic of the algorithm allows you to go beyond an explicit follower graph. It pierces through networks and in its purity of true feedback transcends cultural boundaries. It allows for thousands of subcultures which are hyper-personalised.
TikTok is the sorting hat for muggles.
This is the second part of a three-part series - start with Part I here.
Find Part III here.