Why did I buy a lipstick on the shopping platform, and the next day the homepage and push were all lipstick? Why did I search for a phone in the News app and the homepage was all reviews? Why is it that everyone's popular content is different when you open the same app at the same time?
This is no secret, everyone has long been surprised that the algorithm controls the recommended content.
Fortunately, these algorithmic problems that make us a little helpless are gradually being solved.
Searching for "utilization algorithm" in the Regulations, I found 7 main points
In January 2022, the State Internet Information Office promulgated the Provisions on the Administration of Recommendation of Internet Information Service Algorithms (hereinafter referred to as the "Provisions"), which was deliberated and adopted last year, and came into force in March 2022 (it has now been implemented).
The full text of the Provisions is nearly 4,000 words, which defines the algorithm and standardizes the content disseminated by service providers using the algorithm. For us consumers, we only need to search for the four words "using algorithms" in the full text to find the algorithm problems mentioned at the beginning of the article.
In short, the above situations may not exist in the future:
1.
Algorithmic recommendation services must not be used to endanger national security and social interests, disrupt economic and social order, or infringe upon the lawful rights and interests of others;
2.
Algorithms must not be used to disseminate information prohibited by laws and administrative regulations;
3.
Do not use algorithms to register fake accounts, illegal transactions, or operate accounts to like, comment, or forward;
4.
Do not use algorithms to block, over-recommend, manipulate lists, or sort results, control hot searches, or interfere with selections;
5.
Algorithms must not be used to restrict other service providers or to carry out unfair competition;
6.
Algorithms must not be used to induce minors to become addicted to the Internet;
7.
Unreasonable differential treatment in prices shall not be carried out according to consumer preferences and habits.
The seven points mentioned above, on the one hand, restrict the platform (such as e-commerce, takeaway, news platform) to use algorithms to control recommended content (blocking, over-recommendation, sorting, etc.), as well as unfair competition (restricting friends); on the other hand, it ensures that we (that is, service users) will not be addicted to the Network because the algorithm is addicted to the Network, only see the content that the platform wants us to see, or because of high daily consumption.
After the policy is in place, we don't have to worry about the iPhone being more expensive than the Android phone when taking a taxi; we don't have to use the "Gongzong number" and "Princess number" on Weibo to avoid mentioning the WeChat public account and being restricted; nor will we see comments with similar sentences and no nutrition under some accounts.
In addition, the Provisions also require service providers to add the option that "the recommendation service is not aimed at personal characteristics", and if the user chooses to turn off the algorithm recommendation service, the algorithm recommendation should be stopped immediately.
Just like the switch added by iOS before
In addition to the "Provisions" related to software algorithms, the Cyberspace Administration of China also began the "Provisions on the Administration of Internet Pop-up Window Information Push Services (Draft for Comments)" on March 2, which put forward requirements for malicious speculation entertainment gossip, malicious speculation of old news, different push frequencies of different users, and induced clicks, which are expected to be implemented within the year.
The final implementation date of the draft opinion is 2022
From the Ministry of Industry and Information Technology's attention to app open-screen advertisements last year, to the Ministry of Network Information and Information Technology's rectification of algorithms, pop-up windows and other content this year, the problems exposed by the barbaric growth of the Internet for many years are gradually becoming standardized and orderly.
We who use apps are being shaped by apps
Recommendation algorithms always let me see what I like, and there's nothing wrong with that.
Are you sure that what the algorithm recommends to you is really what you like?
The algorithm is very simple, what you like to watch, give you more pushes, but also guesses the content you may like, pushed to you at the same time; the algorithm is very evil, it will occasionally push a few pieces of content that you may never touch, but most people of the same age or region like, to help you open the door to a new world - if you are interested, it expands your hobbies, or changes the content you like.
Shouldn't I like Texas Hold'em or is it aimed at the little sister in the video?
Since the algorithm recommends content based on your personal usage habits, what will the app recommend if there is no usage habit?
We used three mobile phones from different brands and different network environments to download Douyin, and refreshed 20 short videos each without logging in to the account, the types are summarized as follows:
It can be seen that among the 60 videos, nearly 1/3 of the videos with the themes of beautiful women dancing, photo editing, and daily life, and news and food beauty account for another 1/3. These short videos are basically the good-looking content that the Douyin algorithm thinks most people like.
Do I like it? Not easy to say.
Just like when you meet a beautiful woman on the road, you will look at it twice, and when you smell the fragrance on the way to work, you will turn your head to see what kind of food it is. When I actively search for content, I don't pursue this, but because I look at it more than two times, the app thinks I'm interested in relevant content, and will recommend more related content next.
Give you what you like to see, give you (it thinks) looks good. Does the algorithm recommend content that we really like, or does it recommend content that it thinks we like?
In addition to content fetchers, algorithms have also changed content providers.
Every creator wants their videos to be played by a million views, articles with 100,000 plus, likes and coins flying together. If you want to achieve this achievement, it is inevitable that traffic is supreme, so how to put traffic first? All you need is a traffic password.
After searching for keywords on station B, you can see the traffic password of the cover of the video cover of station B at a glance: the picture is matched with a large-character caption.
As long as a type of content is on fire, you will see a large number of similar content in a short period of time, taking the video of the digital circle as an example, some time ago, "MIUI 12.5 BUG" has become the traffic password of many creators.
Keywords such as "shock", "must read", "retweet" and so on have become ridiculed by everyone about some outrageous news, and they have also been the traffic passwords of some platforms.
The head creators of the platform may not be anxious about the title and cover (even these passwords are driven by them), but for a large number of emerging creators who pour into the platform every day, if they want to quickly increase the number of fans and monetize, it is the most correct way to follow the platform's traffic password, obey the platform's recommendation mechanism, and think about the user weight of the system.
It is said that technology is innocent, but when technology emerges, the privacy, competition, policy and other issues it may lead to are difficult to predict. After we experienced the convenience of sharing bicycles, we saw the collapse brought about by vicious competition; after experiencing the disorderly expansion of community grocery shopping, there was the introduction of the new regulations on community group buying "nine noes".
Usually after enjoying the convenience brought by technology, it will not find its problems, and the same is true for algorithms. We use the algorithm to recommend the content we like, and because the algorithm changes its preferences; because the algorithm searches for the local food that suits us, and because the algorithm can only see the recommendations at high prices.
We're shaping apps, and apps are shaping us.
Please give us the right to nod and shake our heads
Years ago, an elementary school textbook described an information superhighway connected to the Internet: "On the Internet, you can consult information and easily understand the world around you."
But now, the Internet, which was supposed to connect us to the larger world, has created separate information cocoons.
As a big 6 member who watches Station B every day, my homepage recommendation is like this.
Digital reviews, app recommendations, gameplay videos, anime commentary, and recommendations for me to dance in a little dress. Click on the dance video, and the first comment in the comment area is the familiar "brothers have come to learn to dress again" joke. Everyone loves youthful girly dance, but we want to use the recommendation algorithm to get more than just dressing skills.
Is the recommendation algorithm useless? Of course not. In this information explosion network environment, you can't access all the commodity information and news hotspots. When searching for food, who should be the first to rank between "meat sandwich buns" and "snail powder"? When looking for directions, should I prioritize the subway or drive? When I open the News app, should I recommend current politics or entertainment? The "vote for your favorite" algorithm can indeed improve efficiency.
But on the brink of use and abuse, we also need tough policies and let sinless technology do what it has to do.
The implementation of the "Regulations on the Recommendation and Administration of Internet Information Service Algorithms" is a good start, but just like the Ministry of Industry and Information Technology began to rectify the app's opening screen advertisements, the new "gameplay" of shaking to open advertisements was another magic. What kind of practical effects can be achieved by the "Provisions" also depends on the understanding and practical operation of each manufacturer of the contents of the "Provisions".
I only hope that between "costume dressing" and "the larger world", we can have the right to nod and shake our heads.
Click "Watching"
It is the greatest motivation for us