This article comes from the WeChat public account: Geek Park (ID: geekpark) , author: VV, title figure from: vision China
The algorithm does have values.
If you watched a few more videos about fire on Douyin, you might get a series of fire videos next, as if a fire is happening all over the world.
This phenomenon is often referred to as “filtering bubbles”, and there is a similar saying “information cocoon room”. It refers to the recommendation algorithm of some apps based on the user’s behavior, always recommending content that may be of interest to him. In the long run, users may only focus on content that makes them happy, and will slowly weave themselves in the cocoon room like a silkworm. Here, stay in one-sided cognitive range. Even the ability to look at things in a comprehensive way will be weakened.
From a commercial perspective, it is not wrong to continuously recommend to users who are interested. Relying on this magic weapon, some information flow products have achieved great commercial success, and other products have followed suit, turning this into a “standard configuration.” It seems that only by constantly recommending content that is of interest to users, can they have the opportunity to retain users.
However, one company did the opposite and also achieved commercial success.
This company is SmartNews, a news feed news aggregation app. Both SmartNews and Toutiao were established in 2012, and both started with algorithm recommendations. The difference is that it does not simply please users and recommend content that they may be interested in, but recommends content that they “best fit to watch” and “most should watch” to guide users to break the comfort zone.
Under this “I do n’t want you to think, I want me to think” product concept, SmartNews, which only does news aggregation and has no other business, has won multiple rounds of financing. At the end of 2019, its valuation reached US $ 1.1 billion.
SmartNew ’s product philosophy: do n’t be “fat”, let users listen to my
In addition, SmartNews entered the US as an “outsider” headquartered in Japan when the US news aggregation app was the most competitive and the market was the most crowded. In the past two years, the number of users in the United States increased 5 times in the App. The store’s news category ranks in the top three for long-term downloads.
This proves that not so “personalized” product design can also attract a large number of users.
The unique worldview behind the recommendation mechanism
SmartNews ’product interface is very simple. It is so simple that it does not seem to have any competition barriers, unlike an application that can reach a valuation of 1.1 billion US dollars.
In the United States, SmartNews can log in with an Apple ID or Facebook account, but it is not mandatory and can even be used without logging in. The official explanation is that this can avoid the so-called “filtering bubbles” caused by borrowing the relationship chain of other apps and other reasons, resulting in a narrower news horizon for users.
After entering the SmartNews app, there are only a few tags on the top bar: “Entertainment”, “Lifestyle”, “Politics”, “Sports”, “Science”, etc. Users can choose the accounts they want to follow in different categories. News content is presented in the form of information flow, similar to today’s headlines, as long as you slide down, there will be a continuous stream of content.
SmartNews’ interface design is extremely simple, users can select content according to the tab bar at the top
The difference is that today ’s headline content is more popular, everyone can open an account and post articles on it, and SmartNews ’content comes from publishers and media agencies and only provides information.
SmartNews executives said in an interview with 36Kr: “Our content is more ‘hardcore’, that is, more political and economic. Now we are gradually increasing the types of content, such as entertainment, sports, etc. China News There are a lot of self-media on the App, and there is a lot of UGC content. This makes the content diversity in China very good, but the accuracy is not always guaranteed. The richness of similar news applications in Japan and the United States will be less than similar products in China. ”
For such a relatively narrow range of content sources, SmartNews product vice president Jenny Yang said: “ Our goal is to provide news quickly, rather than waste readers’ time, causing readers to miss some thoughts . This is different from applications such as Flipboard. “
SmartNews content vice president and chief reporter Rich Jaroslovsky also said in an interview with NiemanLab that the core of SmartNews is to present a variety of content, including “you do n’t necessarily know what you are interested in.”
“We deliberately optimized the news recommendation instead of over-personalizing it.” Rich Jaroslovsky said, “Our algorithm is different from other personalized recommendations. This algorithm will be based on what users like and clicked before. The content provides some suggestions, and we hope that users come back because they can find new things from time to time here. “
For SmartNews, how to understand the user ’s comfort zone, how to make the recommended content exceed people ’s interest, make hidden gems surface and truly get rid of their follicles is the key.
Redefine what is “user cares”
In addition to recommending some less “personalized” content to users, SmartNews has made new attempts to define what “user cares about”.
In Japan, users prefer to use SmartNews browses entertainment news, while in the United States, readers prefer political news. “News is international, but the application of news is not. This requires different marketing strategies in each region to pick out the corresponding mainstream cultural news.” Founder Niklas said.
The general algorithmic mechanism will lead to recommend news to users with similar content and similar positions. This may make it more and more difficult for American voters to see articles with different views and angles. Therefore, SmartNews embeds a “political balance algorithm” in the App to provide users with more fair and more diversified news content.
During the 2016 US election, based on past user behavior habits, SmartNews recommended the first 15 news articles that might challenge users ’existing opinions.
Rich Jaroslovsky said: “Participation in politics is a wonderful thing, but we want to make sure that people see different perspectives. Help people break through the stratosphere and see things outside the circle. We hope that people get rid of the filter bubbles of news And regard us as the best place to discover news. “
The algorithm has more than one value, whether it will cause an “information cocoon room” depends on the person who created it
The result is optimistic, so the product concept of “not catering to users” once again won huge downloads for it. In this regard, Rich Jaroslovsky interpreted as “People are eager to find content that is politically neutral, and are motivated to pursue truth.”
In addition to recommending less personalized content, SmartNews also recently launched a special “local news” tab in the navigation directory of the App homepage. However, this is not its first.
NewsBreak is the first news app to introduce this feature. In an interview with PingWest, its founder said, “When you really see the hardships of the people at the bottom: the economic downturn leads to a sharp decline in public welfare and the infrastructure is abandoned, people really feel forgotten … you will start to understand their approach .New placeThe importance of hearing has long been overlooked. However, the more real stories happening around you, the more shocking and shocking people feel. “
Based on specific geographic locations to recommend news that users might care about, NewsBreak daily life has increased more than 20 times in 18 months. SmartNews quickly followed suit, marking local news as another important label outside of its “no personality”.
For example, in the tab of local news, SmartNews will recommend the title of related news according to the user’s location: Prioritize community news, followed by a larger range of jurisdiction news.
New product design philosophy
The product design of “anti-personalization” seems to be on the rise.
Not long ago, YouTube ’s mobile terminal made an adjustment to replace the “Trending” (currently popular) on the bottom navigation bar with “Explore” (Explore) function. As the name changes, there is also a change in recommended content. The official stated that the purpose of the adjustment is to display more diverse and different types of content to users.
Similarly, Google Chrome has opened a plug-in that can reverse recommend content to readers based on users ’reading habits. For example, when a user visits Facebook, the plug-in will import articles from different viewpoints into the user’s information flow.
This is not to deny the value of personalized recommendations. In the article “About the” Information Cocoon Room “, Misunderstandings, Truths and Cracks”, the author’s debater Li Muyang mentioned: “Edition and distribution brings ‘you should know’, search brings ‘you want to know’, algorithm recommendation Bring ‘you may be interested’, follow the ‘people you care about’. “
Different content distribution methods have their value. In contrast, the freshness of personalized recommendations has been discounted. The personalized reading engine represented by Toutiao solved some of the problems, but when almost all media platforms started to be driven by personalized algorithms, in this competitive environment, “personalization” can no longer be the main difference between the various companies. .
After all, when everyone is doing personalized recommendations, you are not “individual” anymore. And the appropriate “not personal” will make you stand alone in the eyes of users.
This article comes from the WeChat public account: Geek Park (ID: geekpark) , author: VV