When the algorithm dominates all content

You may have heard how hot live-streaming sales are in China, where the most famous camgirl @Viyaaa had sales of more than $382 million a day at the online shopping festival on November 11, 2019.

Yes, I wrote it correctly. The daily sales of this girl were the net worth of Kim Kardashian in 2018.

Selling goods through live streaming is considered part of the business philosophy of Private Traffic in China, and the concept of Private Traffic is a buzzword in China in 2019.

Since e-commerce operators first use this phrase, most of the explanations for it on the Internet are full of confusing business terms.

But in fact, Private Traffic in a widespread sense refers to the attention or impression you can get in your Facebook, Twitter, and Wechat Moment personal accounts. A series of theories based on this explanation of Private Traffic is actually telling people how to expand the fans of their personal social networking sites fully, and how to market, distribute content, and sell goods or services to the fans.

This is a strange phenomenon: why the Chinese replaced the fan economy, with the phrase Private Traffic in 2019 and made it a hot word again.

Is it because the fan economy is just beginning to be popular in China?

However, the truth is not that.

The logic behind the popular concept of Private Traffic is a counterattack by Chinese Internet users (including some companies and self-hires) against algorithm domination, a Renaissance in a business model.

All information is no longer sorted by time

Back in 2016, a new form of content organization, Feed (信息流, xìn xī liú, It means a list of all kinds of content), suddenly became popular on the Internet in China.

In the Chinese field, Feed refers to neither RSS nor NewsFeed released by Facebook in 2004. It goes a step further, mainly relating to the mixed news aggregator with multiple contents driven by the recommendation algorithm.

In most cases, a Feed-like App, from China does not follow the chronological order of content, nor does it need to follow certain users or topics, but pushes content according to users’ preferences-just like Tiktok.

Douyin, the Chinese version of Tiktok, has been a great success in China, where it has achieved more than 400 million daily active users according to reports at the beginning of 2020, making it the second-highest number of active users in China after Wechat.

Anyone who has used Tiktok or Douyin can understand how recommendation algorithms make users addicted to it. 

But Douyin is not a typical Feed App, because it only recommends video. 

Another product of TikTok’s parent company Bytedance, Today’s Headlines, is typical of Feed App. It mixes long articles, videos, pictures, and micro-blog, in the same interface to meet users’ needs for all types of content, and all the content is presented according to users’ interests. 

In fact, Today’s Headlines is also the reason for Bytedance’s initial success in China, long before the launch of Tiktok. 

Today’s Headlines has an average daily use time of more than 60 minutes among Chinese users and have more than 100 million daily active users.

( Click here to learn how Bytedance uses the recommendation algorithm to create Tiktok and the remaining 20 successful App. )

The strong user stickiness shown in Today’s Headlines makes all Chinese Internet companies integrate Feed into their original product. Here are some simple examples:

Weibo, China’s Twitter, began experimenting with its “smart sorting” function in 2015. Contrary to Twitter’s cautious attempt, Weibo removed “sorting by time” altogether later that year, making it impossible for users to view messages sent by their following accounts in chronological order at all. Despite some opposition, it has been followed by a sharp rise in key data such as Weibo’s user activity, user size, and how long users spend in a single day.

In 2016, Chinese search engine giant Baidu launched the Feed feature on its home page, a new feature that makes it no longer look like a Chinese version of Google. Users can get unlimited information on the home page of their PC site and App without any search, and the information is recommended to be sorted entirely according to the user’s historical search.

It is worth noting that Google launched a similar feature the following year.

In 2017, the UC browser under Alibaba launched the Dayu platform. This platform allows content creators to post content on it. These content will be pushed to the default page of the UC browser by the recommended algorithm.

Now, Feed based on recommendation algorithm has penetrated almost all Chinese Internet products, even if those products themselves do not provide any content at all. For example, in the chat software, in browsers, in a photo app, in a bank App or fitness App.

Feed occupies a blank space in these apps because it significantly increases the number of time users spend in App-you may want to read a few health-related articles during your fitness break. Moreover, after adding Feed to an App, enterprises can collect more user data to improve products or increase revenue through the occurrence of user browsing, likes, retweets, and comments.

Some Chinese solution companies have even made Feed an embedded SDK, similar to Google AdSense, to facilitate small startups that do not have the ability to build Feed to add this feature.

Of all the mainstream Chinese App you can use by 2020 (with more than 100m daily active users), only WeChat Moment still insists on faithfully displaying users’ concerns in chronological order, rather than randomly sorting them according to their “interests.”

It can even be said that this is one of the most significant products form differences between the Chinese Internet and overseas Internet at present.

But this does not mean that this phenomenon exists only in China’s Internet. As anyone with a little knowledge of China’s Internet industry knows, the development of business models and product forms of Chinese Internet companies surpassed the United States a few years ago to take the lead in the world (such as mobile payments).

Google, Facebook, and Twitter have been slowly increasing the weight of “interest” as a ranking element in Feed in the past few years. Maybe sooner or later, users around the world will face a Feed that is no longer sorted by time.

Why does the recommendation algorithm cover everything?

The reason why Feed based on recommendation algorithm is so prevalent in China must have some distinct advantages. To be fair, it is not only good for enterprises but also good for users.

Almost all companies have observed an increase in nearly all their key business data when they integration somewhere in the product a Feed. This means that users prefer to use their products and use them for longer.

There is a reason for this, because since the popularity of the mobile Internet, Internet users are no longer the early middle class or elite who can afford computers. With the influx of low-educated or low-income people, the “people looking for information” way of search engines is no longer in line with the trend.

With the popularity of the Internet, this situation is applicable all over the world, but it is particularly prominent in China. According to CNNIC, the number of Internet users in China has risen from 20 million to 904 million in the past 20 years. However, the proportion of Internet users with a bachelor’s degree or above fell from 62.5% to 9.6%, while those with a high school degree or below rose from 8.7% to 58.3%.

A person with low digital literacy may not know what to search for in a search engine or who to follow on a social networking site. As a result, a product that proactively pushes exciting content to them can significantly improve their user experience.

On the other hand, for enterprises, Feed based on algorithm recommendation not only increases the user activity of the product, but also helps to insert advertisements.

Generally speaking, advertisements in Feed are more hidden, they appear directly in the middle of the content and have a similar appearance to unsponsored content. Because the algorithm collects a large number of user preferences, these ads also appear in content that users are more interested in.

More crucially, when the recommendation algorithm replaces the old “follow model”, companies will have more access to display whatever they want to display on the user’s screen rather than what the user chooses.

This is especially true on Weibo, where it’s hard to imagine a Twitter site saying goodbye to “follow and unfollow”-content posted by people you don’t following will also appear on your page, while content posted by people you following may not be displayed because it is judged by the system as “you’re not interested.”

If people get the right to decide what they want to see from human editors in the era of Web 2.0, then this right is transferred to a virtual algorithm editor in the algorithm era.

Another feature of Feed based on recommendation algorithms in China is that it is unlimited. As overseas users are familiar with Tiktok, you can never finish watching videos moving Tiktok.

In China’s App, almost all Feed have such infinite feature.

As mentioned earlier, Google has also added a Feed function similar to Baidu to its App, which shows users some content that may be of interest to users without the need for users to pay attention in advance.

However, Google’s Feed is updated regularly, updating only a limited number of about 10 or 20 pieces of content at a time, and the list will not appear new content until a few hours after. Baidu and other similar Chinese App unlike this, their Feed provides 8-20 pieces of new content every time users drop down and refresh, users can refresh forever and the list will never stop growing.

This means that once a user enters and starts browsing Feed, in an App, his usage time will increase significantly. There is also a hidden and underlying benefit for enterprises: a user’s daily screen time is limited, and if he stays in my App for one more minute, he will stay less in a competitor’s App.

Therefore, no matter whether users can bring direct business benefits to the enterprise or not, they want users to spend more time on their own Apps.

Fight back by taking back fans.

Obviously, recommendation algorithms can cause problems.

One of the results of using algorithms to cover all Feed is to form an Information Cocoon. This concept was put forward by Cass R. Sunstein, a professor at Harvard Law School who wrote Infotopia.

In simplified terms, it means that people will only “follow” what they are interested in on their users, which will narrow their horizons in the long run.

When Professor Sunstein came up with this concept, people can at least choose what they want to “follow.” When the recommendation algorithm covers everything, most users can’t even actively choose the content they are interested in.

Due to the abuse of recommendation algorithms to some extent, the concept of information cocoons is far more prevalent in China than in the English-speaking world.

In the public domain, the search results of “Information Cocoons” on the Baidu news channel were 24800, compared with 266 in Google News. In the academic field, China’s academic search engines show that 378 academic papers related to information cocoons were published in 2019 alone, compared with 72 in the English-speaking world.

However, the information cocoons is only an intermediate result, and as it changes from an academic concept to reality, it will have a complex and far-reaching impact in many aspects. And this impact involves all aspects related to information-politics, economy, entertainment, lifestyle.

In business, Feed, which is based on recommendation algorithms rather than social relationships, makes it easier for Feed operators to insert ads. 

But sometimes it’s not just ads that get messed up.

[sociallocker id=”5614″]

In the past, every publisher of information had the right to post any information to his fans-if you are a fan, no matter what I post, you will see that. If my content is really worthless, then you will stop following me.

But now, everything is different, and some information that is actually of interest to users can be hidden, but its publishers may not be able to find any evidence.

Because the enterprise can simply tell the content publisher: the reason why your content did not get the expected number of views is that our algorithm thinks that users are not interested in it. Although you have some fans, your content didn’t have time to appear on their screen because they browsed the more interesting content we recommended to it in a limited time.

This makes more and more content creators have to pay advertising fees to App instead of getting advertising share from App while creating content.

On Weibo, the ad service is called fan’s headlines. Its essential function is to “make sure my fans can see my content on their Feed.” And its advanced version is to “show the content to people who don’t follow me.”

There is such an advertising business in Today’s headlines, Douyin, Bilibili, and other Chinese websites. And this advertising business has been proved to be an efficient source of revenue.

Similar to advertising programs exist on Facebook and Twitter.

On Facebook, this feature is called Boost Post, and you can pay Facebook a sum of money to make your content appear on more people’s phones.

But Facebook and Twitter are not entirely dominated by recommendation algorithms on Feed, and if a user is your fan, he can see all of your content in most cases.

So there is no business ethics problem with Facebook’s Boost Post function. Because the companies don’t block your fans from receiving your content and then charge you a fee to remove roadblocks.

This is not to say that this business model is completely unreasonable, but there should be a limit. Just like Google can set the first two items of search results as advertising content, but it should not set advertising content for the first five pages of search results.

It is a pity that most users do not care about the negative effects of the information cocoon room. Because for them, the recommendation algorithm is so sweet that it even helps them filter out a lot of ads they don’t want to see-especially those from content producers.

But for content producers, it’s a nightmare-because you don’t actually have any fans anymore.

Imagine this scene, and you can understand:

You are a Youtuber, and you have accumulated 3 million followers through years of producing quality video content, which allows all your videos to have more than 500,000 views. A generous sponsor decided to put an ad on your channel, and you did your best to make a video that, even though it was an advertisement, still had the same quality as regular content.

You clicked on publish, and the video was played only 500 times in 24 hours. Because Youtube’s algorithm sees it as an “ad”, it assumes that users will not be interested in advertising unless you are willing to give some of your money that was given to you by your sponsor to Youtube.

At this point, we can finally return to the problems mentioned at the beginning of the article.

Private Traffic has been reinvented is that “fans” are no longer critical in current products, what is essential is “private”. Most of the “fans” on the Chinese App are no longer of practical significance, and only accessible “private fans” are the ones you should pay attention to.

WeChat Moment has attracted attention as the only Feed in China that has hundreds of millions of daily active users (in fact, more than 750 million daily active users) and maintains a chronological sequence.

Some content creators with sales attributes create content on various platforms, such as product reviews, food reviews, travel strategies, and then leave the author’s Wechat personal account (not Wechat public account) at the end or comment area of those content. When you add these Wechat accounts as friends, they will continue to send you subsequent content-including what they want to sell.

Another way to obtain Private Traffic is live broadcast, which is a form of strong exposure of the personal image, which makes an emotional connection between users and content producers, which to some extent avoids the weakening of social relationships caused by App.

Some camgirls and camboys with personal brands get more revenue from their brand image negotiations with the platform: I am willing to share the sales revenue with App, but if App takes too much profit, then I will turn to other platforms, because fans focus on my face rather than my account ID. I can switch to the platform at any time.

Camgirls and camboys, which do not have this brand value, will find ways to direct their fans to WeChat or QQ groups during live streaming: if you want to chat with me when I am not live streaming, you’d better join this group. Of course, I will not only chat in the group but also post some ads in the group without restriction.

Some oral ads that are randomly interspersed in a live streaming are also allowed, because in fact, the claws of the recommendation algorithm have not yet gone deep into the interior of a live broadcast. Some video game camboys and travel camgirls, ads are part of their live content, which cannot be filtered by the recommended algorithm.

As mentioned earlier, the popularity of recommendation algorithms in Internet applications seems to be an inevitable trend. Maybe these problems will play out later on sites such as Facebook and Twitter.

[/sociallocker]