We are brushing WeChat, Douyin, and Taobao every day, and we can always brush up on our favorite articles, videos and products. But have you ever wondered what is behind all these "guess you like"?
Simply put, it's personalized recommendation algorithms, as well as behavioral data analysis and collection — they do "know you better than you," but at the cost of ceding privacy. This also makes more and more people pay attention to the privacy of personalized recommendation algorithms, and the sound of adding corresponding setting switches is also endless.
In January this year, the Provisions on the Administration of Algorithm Recommendation of Internet Information Services were promulgated, clarifying that algorithm recommendation service providers should inform users of their provision of algorithm recommendation services in a conspicuous manner; provide users with options that do not target their personal characteristics, or provide users with convenient options to turn off algorithm recommendation services.
The regulations came into effect on March 1, and around the "3.15" of this year, apps such as Weibo, WeChat, Taobao, Bilibili, Douyin, Baidu, Dianping, and Xiaohongshu have successively launched "personalized recommendation" switches, but most of them are somewhat hidden.
Next, APPSO will take you to see where they are "hidden".
How does personalized recommendations work?
Taking some common domestic Internet apps as examples, we will highlight the position of the "Personalized Recommendation" switch and show the main changes to the corresponding pages of the app.
Taobao
Taobao's personalized recommendation switch is already relatively easy to find. The specific closure method is:
Enter the Taobao app, click "My Taobao" in the Tab bar, click the "Settings" button in the upper right corner, enter "Privacy", click "Recommendation Management", and turn off the "Personalized Recommendation" switch
On the Recommendation Management page, Taobao will detail what personal data their personalization algorithms use to track individuals, but does not provide a link to the Taobao Privacy Statement.
Before closing, Taobao homepage recommendation will usually recommend related products such as digital product accessories, model play handmade and other related products based on the categories you have purchased, and after closing, the homepage recommendation will become a "big supermarket", and the recommended products are basically daily necessities.
▲ On the left is the recommendation that turns on personalization, and on the right is the recommendation that turns off personalization
Bilibili
The personalized recommendation switch of Bilibili (hereinafter referred to as "Station B") is hidden more deeply.
Enter Station B, click the "My" drop-down in the tab bar, click the "Settings" drop-down, click the "Privacy Permission Settings" drop-down, click "Personalized Content Recommendation Management" to close "Personalized Content Recommendation"
In addition to hiding deeply, Station B did not elaborate on what kind of personal data their personalized recommendation function would use.
For most B station "omnivorous" viewers, before and after closing the "personalized content recommendation", the recommended content on the home page of the mobile terminal of the B station will actually hit your preferences, but reduce the video of some segments that are often watched.
▲ The left is the homepage of Station B before closing "Personalized Recommendation", and the right is after closing
Weibo has launched a variety of clients, and the personalized recommendation switch for each client is also different, but the basic location is the same.
The normal version of the shutdown method is:
Click "I" in the tab bar, click "Settings" in the upper right corner, click "Privacy Settings > Personal Information and Permissions" to close "Personalized Content Recommendation"
The closure method of Weibo International is similar, but it is relatively not so deeply hidden.
Although there is no separate page switch like Taobao and B Station, Weibo's personalized recommendation algorithm is relatively thoroughly closed - before closing, my "recommended" list will basically recommend the content of bloggers in various game fields and technology fields, and after closing, it has become a smorgasbord of entertainment circles. Well, from this point of view, Weibo should be the least suitable to turn off the personalized recommendation function...
▲ The left is the recommendation to turn on personalization, and the right is to turn off the personalized recommendation
Another reminder, turning off personalized recommendations doesn't affect the timeline sorting of your homepage, it's still messy, sorting the content sent by the system recommendations to the people you follow.
Therefore, students who have the habit of brushing Weibo in the timeline sequence are still recommended to continue to use "Weibo International Edition" or other third-party Weibo clients.
Now WeChat is already a collection of multiple services, and each service is operated by a different team of WeChat, so WeChat's personalized switch can be said to be the most complex of several apps.
Here, APPSO uses the common subscription number and video number services in WeChat to introduce how to turn off the personalized switch.
The first is the subscription number.
Using an Android phone, click on the menu in the upper right corner of "Subscription Account Message", click the More button, and select "Settings" in the pop-up menu to find the switch of "Subscription Account Message Personalized Recommendation".
Next is the video number.
Click the "Villain" logo in the upper right corner of the video number, click "Privacy > Personalized Recommendation" in "Browsing Settings", and turn off the two personalized recommendation switches of "Live" and "Video".
Meituan
Meituan's personalized recommendation switch is easier to find:
Open the Meituan app, click "My" in the bottom tab bar, click "Settings" in the upper right corner to find "Privacy Management" and "Personalized Content Recommendation", and turn off the switch
In addition to providing switches, Meituan also has a separate page that shows in detail what personal information their personalized recommendation algorithm will use, which is the most detailed category of several apps mentioned in this article.
After closing the personalized recommendation, the recommendation on the home page of the Meituan group will obviously have a decrease in relevance: before closing, I will recommend the recommendation of a milk tea shop that I wanted to place an order before but gave up, and after closing, it is basically a shopping punch card guide. It is worth noting that after turning off personalized recommendations, Meituan will also mark the distance between the merchant and you in some recommendations, and it seems that the off switch does not completely turn off the use of location information.
I know
Zhihu's personalized recommendation settings are not only hidden deeper, but also play a small word game. The closing steps are as follows:
Open the Zhihu app, click "My" at the bottom, click the "Gear" icon in the upper right corner, find the "Privacy Center" click "Permission Authorization Settings" to find the "Turn off personalized content recommendation" switch and turn it on
It should be emphasized that the switch is turned on to indicate that Zhihu's personalized content recommendations are turned off. Zhihu also did not indicate next to the switch what personal data would be used to participate in the recommendation decision.
Zhihu's personalized recommendation switch belongs to a more thorough faction: the very subdivided topic area content recommended before closing will disappear without a trace after closing, and will become the content related to hot events.
Providing a "switch" is only the first step in a journey of a thousand miles
Does personalized recommendation have to be a "flood beast"? Not quite.
In the era of rapid expansion of information dissemination speed and quantity, the quality of information seems to be difficult to fully control, and under the premise of rational use, it is also a means to help filter information by allowing algorithms to help make some decisions.
Our applause for personalized recommendation doesn't necessarily mean how much we hate personalized recommendations; on the contrary, there are bound to be many people in the world who need to use personalized recommendations to help them process their information on an initial basis. The more important meaning behind this switch is that Internet companies and society as a whole are willing to start respecting and listening to the opinions of users in addition to their interests; rather than blindly forcing the recommendation algorithm into the user's mouth, and using this as an excuse to excessively collect personal privacy.
From this point of view, everyone will be happy to see more and more manufacturers respect the needs of users, comply with regulatory requirements, and provide personalized algorithm switches. Regardless of whether these switches are effective or not, at least manufacturers are beginning to be willing to provide users with such switches, which is an improvement in itself.
As things stand, a simple switch in the app is obviously not enough: the personalized recommendation algorithm is still a black box for the average user, and the user does not know what kind of personal data participates in the algorithm's decision-making throughout the process. In addition, after the personalization algorithm is closed, the implementation of the recommendation algorithm in each app is also different.
We might as well refer to some well-done examples, such as the privacy page in google account or Apple ID settings, to provide users with more detailed settings. Here, you can see how much personal data such as your purchase orders and browsing history have been collected by Google or Apple, and you can also correct, delete or export it, or even directly prohibit them from collecting your personal data.
In this comparison, those apps just provide a simple and rough switch, which seems to be a lot more "primitive".
For ourselves, trying to reduce "psychological dependence" on these apps by turning off personalized recommendations is not the best option.
We choose to "swipe the phone" largely not because of personalized recommendations, but because of the imbalance in the demand for information intake in the brain. Just like in the era when there was no Weibo and WeChat, we would pick up the shampoo bottle and look at it when we went to the toilet.
In the same way, the "information cocoon" problem of "personalized recommendation" will not be completely solved because the recommendation algorithm is turned off - this problem is essentially caused by our brain's preference for rapid processing of information.
There are no "personalized recommendations" in shampoo bottles, and our brains (most likely) will not actively reduce the intake of useless information because the recommended content is not satisfactory. In order to really get rid of "mobile phone addiction" and "social addiction", the more effective way is obviously to try to actively adjust their information intake sources and open up new choices for themselves outside of mobile phones and apps; for a point of view and news, it is far more important to understand and dig up these things and sources than to be mobilized by these content and unnecessary emotions.
▲ A vain attempt to solve the problem of privacy and information cocooning with only one switch will only make the problem more complicated. Image courtesy of Unsplash@yungnoma
In short, this "late" personalized recommendation switch has finally landed, which is a good thing after all. But for manufacturers and users, its landing is only the first step, in order to make manufacturers truly respect users, so that users are responsible for their own intake of content, obviously there is still a long way to go.
Hopefully that day doesn't come too late.