TikTok is launching a new feature that allows you to start your TikTok feed from scratch if you don't like the videos that are being recommended to you, while it's also developing new systems to avoid repetition and limit exposure to harmful elements.

First, on the new feed refresh option – in the last few months, TikTok has been experimenting with a new process that essentially allows users to basically start their account over again, by resetting what the algorithm thinks they might be interested in.

Now, TikTok is rolling that option out to all users.
As TikTok explains:

"On TikTok, For You feeds help people discover a diversity of content, creators, communities, and products.". But we also understand that there are times when people's recommendations don't feel relevant anymore, or provide enough topical variety. So, we're rolling out a way to refresh For You feed recommendations if they no longer feel like they're for you. When enabled, this feature allows someone to view content on their For You feed as if they just signed up for TikTok. That would begin to bring up even more content based on new interactions into our recommendation system.

Good option, of course; the other social apps likely cannot do this for users and control more of their in-app experience - or have not done this at any point anyway.

And with it, Meta's longtime competitive advantage in the social space is its all-powerful social graph: everybody you know is on Facebook, so if you want to be up to date and see everything your friends and family find interesting, you have to be there too.

Over time, Facebook and other applications have perfected this, where through your likes and follows it determines what the algorithm showcases next. But TikTok's algorithm is different: it learns much faster through your actual viewing habits, since your behaviors in the app will constantly update and refine its algorithmic system, which then helps it to showcase the best-performing videos from anywhere in the app, versus being limited by your connections and explicit activity.
This has changed the paradigm for social apps. Whereas once, social platforms were reliant on manual actions to show you more relevant content, TikTok's automated algorithms are now so advanced that it can give you a better user experience, even if it knows nothing about you, other than what you do in the app. Now, every other app is playing catch up, and working to improve on their own automated recommendations, but so far, none of them have come close to the compelling experience provided by TikTok's highly attuned system.

Which also enables it to be able to offer updates like that, because it knows that its system is so good at showing you content based on your activity that it can even let you start all over again, and be confident that it'll align with your interests quickly.

Which is a big advantage, and it'll be interesting to see what the user experience is like in refreshing your feed and shedding your past interests in the app.

On another front, TikTok also provided some new insight into how it's improving its recommendation systems to avoid content repetition and, in particular, to ensure that users are not shown too much content that could be harmful or triggering.

“An inherent challenge of any recommendation system is ensuring the breadth of content surfaced to a viewer isn’t too narrow or too repetitive. We’re intently focused on this challenge, and work to design a system that intersperses a variety of topics. For example, viewers will typically not be served two videos in a row created by the same creator or that use the same sound, and we try to avoid showing people something they've seen before."

This doesn't always work, but again, TikTok's advantage in this respect is that it's not bound by your social graph, in regard to what it can show you, so it has a huge, never-ending stream of content to choose from when populating user feeds.

"In addition, we work to carefully apply limits to some content that doesn't violate our policies, but may impact the viewing experience if viewed repeatedly, particularly when it comes to content with themes of sadness, extreme exercise or dieting, or that's sexually suggestive."

According to TikTok, based on expert advice, allowing users to see how others cope with difficult emotions can be helpful, especially for teen users, but there does need to be limits on over-exposure, which is where it's focused.

"Our systems do this by looking for repetition among themes like sadness or extreme diets, within a set of videos that are eligible for recommendation.". If multiple videos with these themes are identified, they will be substituted with videos about other topics to reduce the frequency of these recommendations and create a more diverse discovery experience. This work is ongoing, and over the last year alone, we’ve implemented over 15 updates to improve these systems, along with expanding to support more languages.”

The core working principle is users should be able to share and discuss sensitive topics but not feel dragged down rabbit holes that repeatedly expose them to such; thus, TikTok is working with a range of experts in determining optimal exposure limits which will also be built into its algorithmic process.

Both are important areas of development that can have a huge benefit to the user experience. And while it may seem like enabling users to reset their feeds could impact ad relevance, in that you potentially lose some of the context of what people may be interested in, it could also, conversely, improve such, by refreshing people's interests over time, which will enable brands to get more relevant promotions in front of the right people.

It'll be interesting to see how each changes the user experience, and whether it leads to improvement – potentially to be emulated by other platforms.