“WeChat” or “Today”, which one has a better chance of becoming the “headline”?

[[126484]]

At the Sogou Search Mobile 3.0 version release conference, Sogou CEO Wang Xiaochuan officially released a new feature called “WeChat Headlines”, which, like “Toutiao”, focuses on personalized intelligent recommendations. Zhang Chaoyang specifically liked the Weibo post “Sohu sues Toutiao for copyright infringement and demands 11 million yuan”.

The purpose of WeChat Toutiao is self-evident: to kill Toutiao

After three years of data accumulation, Toutiao has accumulated enough user data, making it the undisputed leader among independent news aggregation products. Yidian Zixun, which was born a little later, is also growing rapidly, but it is difficult to compete with Toutiao in terms of data volume and user base.

However, the newly released WeChat Toutiao can be said to be standing on the shoulders of giants. Its mother is Sogou, its milk is Tencent, and its grandfather is Sohu. As a product born with a golden key, it not only has the massive data collected by Sogou for 10 years, but also the content accumulated by the WeChat public accounts of the top companies. It can be said that the two are basically on the same starting line in terms of data volume. As a personalized recommendation platform, Toutiao finally ushered in a heavyweight opponent.

Below, the author will analyze the advantages and disadvantages between the two from five aspects.

1. Personalized recommendations, precise details are king

Personalized recommendations are the main features of both, but Sogou still has certain shortcomings in technical preparation compared with Toutiao.

Perhaps due to strategic considerations, Sogou's "WeChat Headlines" currently only supports WeChat account login, which goes against the current user experience principle of loose registration through multiple channels. However, the strange thing is that the public account information is not synchronized after logging in to WeChat. WeChat has not opened more data interfaces separately, which is quite puzzling. If WeChat can open data to Sogou, analyze the types of articles in users' previous Moments, and make personalized recommendations based on this, then it is only a matter of time to catch up with Toutiao. However, Sogou and WeChat have not done this yet. I don't believe that Sogou can't think of this. I can only say that Sogou is currently in the experimental stage and is not well prepared.

On the other hand, Toutiao supports multiple social platforms such as Weibo, WeChat, QQ, Renren, Kaixin, etc., and the account information collected by using third-party APIs is relatively more comprehensive. According to a person close to Toutiao, the author has obtained eight recommended methods for Toutiao:

√ Recommendation of similar articles based on topic similarity: Recommendations are made by obtaining articles similar to those that the user has read.

√ News based on the same city: For users with the same geographic information, popular articles from matching cities will be recommended.

√ Recommendation based on article keywords: For each article, extract keywords as a feature to describe the article content. Then match and recommend articles with the keywords in the user's action history.

√ Universal recommendations based on popular articles on the site: Find popular articles based on the reading habits of users on the site and recommend them to all users who have not read the article.

√ Recommendations based on reading habits of social friends: Based on the user’s off-site friends, get the articles forwarded, commented or published by the off-site friends for recommendation.

√ Recommendations based on users’ long-term interest keywords: Recommendations are made by comparing users’ short-term and long-term reading interest topics and keywords.

√ List recommendations based on similar user reading habits: Calculate the similarity of user actions within a certain period of time and make cross-recommendations on reading content.

√ Content recommendation based on site distribution sources: Calculate 20 news sources that users like based on the source distribution of articles they read and recommend them to users.

However, don’t be too eager to worship when you see so many recommendation methods. You should note that Toutiao’s personalized recommendations are not entirely based on big data. It is not difficult to come up with recommendation rules based on popular information, social relationships, geographic information, etc., and it is not a problem for Sogou to write and apply the rules in a short period of time. If Sogou wants, the gap between WeChat Toutiao and Toutiao in content recommendation will be greatly reduced in a short period of time. You should know that it is difficult for most users to perceive the difference between a product with an 80-point score and a product with a 70-point score.

2. In the competition of refined content, what are the chances of machines winning against human brains?

√ Toutiao’s main robot

Toutiao focuses on personalized recommendations. Its founder, Zhang Yiming, is also a technical person. He attaches much less importance to content than one might imagine. "The IQ of a machine may not be higher than that of a knowledgeable person. This is mainly because people have limited memory and limited data processing, while machines are tireless and can input massive amounts of data at any time." Zhang Yiming has expressed similar views on many occasions. However, in actual practice, Toutiao has shown a significantly different direction in the refined operation of content.

In the latest version of Toutiao, many details about content operation were added, such as imitating the keyword subscription of "Yi Dian Zixun" and the media account subscription of "Sohu News Client", which shows that it is still impossible to achieve accurate personalized recommendations of Toutiao simply by relying on machines and technology. The author privately interviewed an operator of Toutiao media platform and learned that the ratio of recommended content to read content of Toutiao account is about 20:1, which is still a long way from accuracy.

√ WeChat headlines screen manual number selection

WeChat Toutiao has added obvious interest tags, such as jokes, private dishes, and likes. These tags allow users to quickly find content that interests them. In addition, the content on WeChat Toutiao does not currently move all the content of millions of WeChat public accounts to this platform. Most of the content is content posted by high-quality accounts after screening. A careful observation will reveal that the number of readings of WeChat Toutiao content is relatively high, which means that these contents are high-quality content that is widely welcomed on the WeChat public platform and have undergone a second filter. In terms of quality selection, Sogou still has a great advantage at present, but in the future, after opening up to include various public accounts, whether the content can still maintain the standard will be a big problem.

It should also be noted that the personalized recommendations featured by Toutiao make it possible for these contents to be read. Unlike WeChat Toutiao, the more popular the content on the Toutiao platform is, the more people will be recommended to it. Whether the content is recommended depends entirely on the group of users who initially recommended it. The average user quality of the entire platform is far behind that of the elite group, so the positioning of the content is completely determined by the novice users. It is difficult for high-quality content to be effectively recommended, and the entire platform is full of clickbait titles. At present, these contents are difficult to deal with through technical means.

On the other hand, the more important issue is the operational perspective of social news. Today’s headlines are presented in chronological order, and it is difficult to effectively trace historical content. However, Sogou’s WeChat headlines are not arranged in chronological order, which can well organize and integrate historical content, especially to form more readable topics. Therefore, there was a huge gap in content quality between the two sides from the beginning.

3. Risks of copyright competition, and doubts about the safe harbor principle

√ Toutiao still faces risks

In June last year, Toutiao experienced a copyright dispute, and behind this dispute, Sohu was involved. Since then, the two sides have been fighting openly and covertly to a certain extent. There are still great doubts about whether Toutiao's content transcoding behavior at that time was infringing. However, from the perspective of content copyright, Toutiao still faces great risks.

Not long ago, many WeChat groups launched a campaign to criticize Toutiao for stealing content from Baidu.com. Many authors’ articles were grabbed onto the Toutiao platform without permission. The content of Baidu Baike authors was reprinted by many media outlets such as Qudongzhijia.com and Kanchai.com, and then because these media outlets had copyright cooperation agreements with Toutiao, these contents could be spread unscrupulously on Toutiao.com. According to Toutiao’s publicity after the copyright storm, all the content currently included and released by Toutiao belongs to the content partners. On the surface, there is indeed no problem, but a closer look shows that this is still obviously an infringement. In fact, Toutiao also understands this risk, so it has tried its best to launch Toutiao accounts, hoping that content producers can settle in the platform and resolve this risk.

√ No worries about copyright for WeChat headlines

Compared with WeChat Toutiao, which is backed by two major portals, Sohu and Tencent, as well as the WeChat public platform, the copyright risk of its content has actually been completely resolved. In addition, as a parasite on Sogou Mobile Search, WeChat Toutiao does not transcode the content, but directly diverts it. WeChat Toutiao's advantage in copyright may become a killer weapon to counterbalance Toutiao in the future.

4. Platform openness, whoever can divert traffic will become the king?

Which platform is more open, Toutiao or WeChat Toutiao? In my opinion, WeChat Toutiao still has more advantages at present.

As mentioned earlier, WeChat Toutiao actively directs traffic to WeChat public accounts to help content producers expand their influence. However, Toutiao platform transcodes all self-media content, and only a few media websites cooperate in traffic diversion (the author counted about 300 mainstream media, and only less than 10 of them cooperate in traffic diversion). From the perspective of openness, Toutiao is still lagging behind and has missed a great opportunity for development.

In fact, for mobile-based media platforms, it is not important whether the traffic belongs to them. You just need users to stay on the platform and not jump out. Mobile Internet apps are basically equivalent to browsers in the PC era, and all user behaviors take place within the platform. The role of traffic diversion is not only to help content producers grow, but also to attract more traffic from outside with the help of content producers, forming a win-win situation.

At present, Toutiao has fallen into a bottleneck of traffic growth due to its inability to fully open up. This can be seen from its launch of a series of independent apps, such as Jinri Toucai, Jinri Specials, and Neihan Community. It has been frantically developing a series of products and hopes to increase overall DAU growth through other apps. But for Toutiao, perhaps full openness is the real way to win.

5. Active or passive, which one will represent the search engine of the future?

In an interview, Zhang Yiming believed that Toutiao is not news in a narrow sense, but information in a broad sense, and is a search based on a recommendation engine. As WeChat Toutiao is parasitic on Sogou, it is obviously also closely related to search.

Obviously, under current technical conditions, it is impossible to completely replace traditional passive search with active recommendations. However, as far as the author is concerned, since using Toutiao, the number of times I actively type in keywords to search has indeed been reduced. However, this reduction in the number of searches is not because Toutiao can pre-perceive the news I want to accurately view, but because Toutiao actively arranges a large number of recommended news streams to be piled up on my mobile phone screen. When I keep pulling down the information flow, the major industry news I follow is "brushed out". After clicking in, Toutiao knows my habits over time, and the content recommendations are relatively more accurate. But the question is, does Toutiao involve an infringement of user privacy in this way? The author renovated some time ago and selected a few home improvement articles to click in. Now, the information flow is full of articles sharing decoration experience.

Sogou has a combination of both. WeChat Toutiao currently does not make intelligent recommendations based on user clicks on articles, but it is directly integrated with Sogou, a search engine, so that you can choose WeChat Toutiao when you read the news, and when users have specific keywords to search for articles, Sogou search comes in handy.

6. Social relationship chain, the next battlefield for content recommendation

In the update history of Toutiao, Toutiao version 2.8, which was launched on August 22, 2013, is actually an important watershed. From this version, Toutiao strengthened the social relationship chain of friends and began to de-microblog. The friend dynamics of version 3.5.1 updated on July 28, 2014 was further upgraded, which completely sounded the clarion call for Toutiao to enter the mobile social networking. However, as a pure information platform, too much trivial information interferes with users' communication outside the article comment area. Toutiao's socialization attempt is still in its initial stage, but it still provides a good development direction for content recommendation.

The content of WeChat headlines all comes from WeChat public accounts, and 80% of the reading of WeChat public accounts comes from sharing in Moments, which largely reflects the reading habits of most netizens in Moments. This recommendation strategy obviously completely ignores the human factors behind WeChat public accounts. Therefore, this is also the key reason why I am not optimistic about WeChat headlines to a certain extent. If it is just an information reader, WeChat is actually enough, and we don’t need to open a new entrance outside of WeChat.

Compared with Toutiao, WeChat Toutiao is more inclined to manual screening, and many high-quality content cannot be disseminated and distributed. At present, Toutiao's machine recommendation has a greater advantage. However, as Toutiao needs to accumulate data, it also captures a lot of junk content. The rapid growth of Toutiao accounts has led to an explosive growth in the amount of content on Toutiao, but the quality of the content has seriously declined.

Machines will never be able to read people's minds. New media platforms can only effectively solve the problem of information dissemination, and the massive amount of information determines the obsolescence of the traditional editor recommendation system. Regardless of WeChat Toutiao or Toutiao, the next focus should be on people. With people as the hub, the content is reorganized and processed, from "everyone is a self-media" to "everyone is an editor." For example, which articles Wang Xiaochuan has read and collected, which WeChat public accounts Zhou Hongyi has followed, and which articles Zhang Chaoyang has liked, this kind of recommendation based on people will be the next battlefield.