Hello, everyone. Today I’d like to talk about the information cocoon room in the recommendation system.
When it comes to information cocoon room, everyone has a sense of crisis. Before, some people in Zhihu asked how to fight against the recommendation system and avoid falling into the information cocoon room. So what exactly is the information cocoon room, and how does it appear? Could our future really be dominated by recommendation systems that keep us out of touch with new things?
Before we answer these questions, let’s take a look at the reasons for the emergence of cocoon rooms.
Matthew Effect and information cocoon room
There is a famous saying in Matthew, “To every one that hath shall be given double, and he shall have more: but from him that hath not shall be taken away even that which he hath.” A similar principle also appears in the Tao Te Ching. The Tao Te Ching says, “The way of heaven damages more than it needs to, but it needs to be replenishing. The way of man is different.
This is complicated, but intuitively easy to understand. The richer a person is, the more resources he has, the more money he can make, and the richer he becomes. The poorer they are, the less means they have to make money, the poorer they are. The same is true in the field of recommendation systems. The more people browse the product, the more popular the product is, the better its sales volume, evaluation and various data are. The better these metrics are, the more likely they are to be picked up by the recommendation system and thus get more exposure, trapping them in a positive cycle. Those that are less popular fall into a negative cycle, becoming less and less popular until they finally disappear.
What is the consequence of this? This will result in the operation of the recommendation system until later, when the recommended products are all highly popular, and the products or contents we see are all similar.
Take a simple example, there was a period of time when B station pushed a lot of little sister, and when I clicked it, it was almost a cover of a beautiful woman. The truth is also very simple, beautiful, attractive, so push to push is this kind of video.
It is not that there is no solution to this problem, and a better solution is personalization, the so-called thousand faces. In plain English, it takes our preferences into account and pushes content that is more relevant to our preferences based on our previous behavior. Now almost mainstream recommendation system is doing so, Taobao, Douyin is no exception. For example, when I open B station, I can tell the recommendation reason of every video.
Personalization works very well, with personalization, the click-through rate of recommended content is greatly increased, nearly twice as high as that of pure content. You think, for the company, only need to hire a number of people and purchase a number of machines, can nearly double the order volume of the whole company, such a good thing, no one will not refuse.
Although personalized effect is good, but not all problems, its biggest problem is the so-called information cocoon room. The information cocoon is similar to the Matthew effect, meaning that at the end of individuation, what we see is what we are interested in. Our access to the Internet has not broadened our horizons; on the contrary, it has hindered our understanding of the larger world. It’s like being wrapped in an invisible cocoon of information, unconsciously becoming a frog in a well.
Strange paradox
There’s nothing wrong with all of this, and it fits perfectly with how we know and how real systems work. But there is one point that is very, very strange, strange as a paradox.
This point is the mission and current situation of recommendation system. Do you know why there is a recommendation system? According to our current understanding, it seems that the recommendation system is created to construct an information cocoon and restrict people’s thoughts through big data. In fact, on the contrary, the mission of the recommendation system is actually to oppose information cocoon house and information occlusion.
It’s easy to figure this out because time is limited and the content on the platform is almost unlimited for each user. For example, there may be hundreds of millions or even billions of commodities on Taobao now. For a single user, it is impossible to browse all these commodities even if he does not do anything all his life. For users, what they can see is always a very small fraction of the product.
Let’s zoom in a little bit, not just for individual users, but for all users. The number of items that users browse on the platform is actually only a small fraction of the total, with some estimates saying it could be less than 10%. One consequence of this situation is that older products monopolize the market due to first-mover advantage, while newer products have little chance. Just like the closed society in ancient times, the children of ordinary people can never become the elite, which is the same reason.
It is to solve this problem that recommendation systems have been built. Hopefully, there can be a system that can efficiently find some high-quality but unpopular products, give such products the opportunity, let the users like it can see it, so as to reduce the traffic difference between the old and new products as much as possible. It can be understood as cutting out part of the flow of the head commodity to other high-quality commodities, so that more commodities get exposure.
We in the we-media business often compare the difficulty of the development of various platforms. Now it is generally accepted that B station and Toutiao are friendly platforms for newbies. The newbie friendliness here is actually reflected in the platform’s traffic support for authors who have just started. The more Matthew effect a platform is, the harder it is to start from scratch. On the contrary, the better the effect of the recommendation system, even if the data is not so excellent content can also find appropriate user recommendation.
When you think about it, it’s not a paradox. The recommendation system’s mission is to fight monopolies, but it will eventually produce the Matthew effect. It’s not that recommendation systems don’t work, it’s that the Matthew effect is too powerful. Recommendation system is not enough to offset its influence, what can do is to screen out high-quality content as much as possible from the mass of content to give more opportunities.
Fight the information cocoon room
For us users, a lot of the time we go online is to get more information, to see the world that we didn’t see before, rather than just see what we like to see. The current situation is that the personalization of each platform is getting better and better. We will be in such a cocoon of information in the future. In this situation, what should we do?
Unfortunately, neither individuals nor industry has much to offer on this issue.
For the industry, there is no better recommendation than personalization. Because quality is relative on most platforms, there is a niche niche for almost any kind of content. Companies and platforms are looking for clicks or results, so they won’t use recommendations that don’t work.
Personally speaking, we obviously like to watch all kinds of cartoons and all kinds of little sisters, so the recommendation system recommended to us is also relevant content. Can’t we force ourselves to watch something we don’t like just to break out of our information cocoon? And even if we do, so to speak, cheat the system and then recommend us something else, isn’t that just another kind of information cocoon? And it’s a cocoon we don’t like.
From my personal point of view, I don’t think this is a big problem. There are two reasons. The first one is that the recommendation system is not just for its liking. We all know that even if we love watching little sister, if we see a full screen of little sister, we will resist. Therefore, in addition to the click-through rate, the diversity and richness of the recommended content is also an important indicator of the current recommendation system. The other point is that for us, the Internet is just a tool for us to get information, not to expand our world view and outlook. We want to achieve this, we need to make efforts to become a broad vision, a wide range of hobbies. A person with a closed mind, even if he is given a lot of meaningful and valuable content, he will probably only click on the area of interest.
There is an overemphasis in some places on the concept of “information cocoons” and the portrayal of these platforms as dark manipulators of people’s minds. These are actually excessive assumptions and guesses by outsiders. Platforms and companies are in pursuit of performance and profit, not boring and unprofitable dark pursuit of manipulating people, so this view is untenable, and you don’t have to worry about it, just laugh it off.
That’s all for today’s article. I sincerely wish you all a fruitful day. If you still like today’s content, please join us in a three-way support.
Original link, ask a concern