laitimes

Byte "de-algorithm" experiment

Written | Fish three falcons

Edit | Jingyu

Another "invitation-based" app, a code is hard to find.

As early as January this year, there was news that Byte was targeting "interest social networking" and would launch an app called "Zhi District". Recently, with the concentrated coverage of "Zhi District" by many media, this app that focuses on interest reading and combines social and content has triggered a wide range of discussions.

In the past decade, a large amount of information generated after the explosion of data has overwhelmed everyone, so "algorithm recommendation" has become a weapon to help people filter information and recommend content, and apps such as today's headlines that focus on "algorithm recommendation" have undoubtedly achieved great success with the efficiency and accuracy of their algorithms.

However, when the drawbacks of relying entirely on "algorithm recommendation" gradually emerged, and the Cyberspace Administration of China strengthened its regulations, an experiment on "removing algorithms" was unfolding.

In addition to providing a "close algorithm recommendation" button in today's headlines, Byte's "recognition area" App is also unusual, not only does not take its consistently recommended "algorithm recommendation" as the core, but tries to return to the "human recommendation" with interest and social as the core.

What is the Zhi District?

The feeling after many users experiences is that in the "recognition area", they see the shadow of many other products.

Byte "de-algorithm" experiment

First of all, the concept of "recognition area" is understood as a "group" of Douban, and users with the same interests can freely choose to join one or more "knowledge areas", and after joining, they can publish content that is consistent with the theme of the "knowledge area". Each "district" will have a "district chief", similar to the "group leader" of the Douban group. Each user can create their own recognition area, just follow the app's guidance, one step in the next name, introduction, set the head map to complete the creation.

Byte "de-algorithm" experiment

In addition, after the "knowledge area" is successfully created, the district chief can also choose to equip his knowledge area with a bot (robot that automatically grabs content), when setting the bot's filtering rules, the system will first investigate the topics that the user is interested in, and extend more keywords from this topic for users to choose, so as to determine the bot's subscriber source. Among them, most of Bot's subscription feeds come from today's headlines, but users can also expand the scope of content subscriptions by adding RSS subscriptions.

Byte "de-algorithm" experiment

This setting is reminiscent of the early "instant" and long-faded RSS (Simple Information Aggregation) readers.

In terms of content browsing, "Zhi District" provides two different content push methods, one is the content update and recommendation inside the "Zhi District" column, and the other is the "Recommendation" function that can see other posts in different Zhi District.

Byte "de-algorithm" experiment

Users can interact and communicate with the post below the post when browsing, and the "light up" button next to the "Reply" button acts as a function of like and favorite, and the user can also record the feelings at that time while "lighting up".

Byte "de-algorithm" experiment

In terms of content release, "Zhi District" also provides two forms of release, one is a text mode that requires user-generated, and the other is a forwarding mode that can be shared by quoting external links.

A "de-algorithmic" experiment

On September 6, 2006, Facebook launched the News Feed feature. At that time, people did not buy this feature, and even protested for a period of time because of privacy concerns. But when this recommendation model significantly improved the efficiency of people's access to information, more and more people became dependent on it, and to this day, algorithmic recommendation functions similar to News Feed are so common.

In China, "Today's Headlines" is the first product to embrace "algorithm recommendation". In August 2012, today's headlines were officially launched, with 10 million users in less than 90 days, and soon competed with traditional news portals such as Tencent, NetEase, and Sohu.

But there are two sides to everything, with the occurrence of algorithmic manipulation events such as the 2016 US election and the Brexit referendum, the negative recommendation of algorithms is gradually presented in front of people, and the public cannot help but begin to reflect.

In March this year, the "Provisions on the Administration of Internet Information Service Algorithm Recommendation" was officially implemented, and under the requirements of supervision, Today's Toutiao, WeChat, Xiaohongshu, Zhihu and other Apps have launched the "Close Algorithm Recommendation" option, but many users have questioned this.

Once a user tried to check the "turn off personalized recommendation" button on a certain platform, and then the content presented on the platform was almost disorderly, and in desperation, they had to reopen the personalized recommendation. If in order to achieve the independent choice of algorithm recommendations, it is necessary to lose the quality of reading or use experience, which is obviously what most people do not want to see.

Byte "de-algorithm" experiment

Most apps use "algorithmic recommendation" | Unsplash

How can we avoid the drawbacks of "algorithm recommendation" while still maintaining the user experience?

Although in 2013, when Google Reader, as the representative product of RSS, announced the closure, Zhang Yiming once wrote in an article: Compared with relying on manual information only from the portal, the subscription model has actually improved, but it is obvious that it has not been able to meet the information needs of most people. What kind of reader is smarter and more suitable for the public and will become an alternative to Google Reader? My team and I answer yes to algorithm-based personalized recommendations.

But this time, Byte bet the chance of a breakout on RSS.

RSS is a tool that aggregates all the content that users subscribe to to read on the same platform.

The traditional RSS use threshold is high, users must first clearly know what kind of information they need and find the corresponding feed, and secondly, they need to have strong self-control, can regularly sort out and control the number of RSS feeds.

Byte "de-algorithm" experiment

Many RSS reading apps "inherit" Google Reader's user | Internet

Most of the content in the "Knowledge Zone" is crawled by bot in the feed, so the Knowledge Zone and RSS are essentially user-owned subscription content, but the difference from the above traditional RSS is that the Knowledge Zone also adds interest and interaction.

Since each member of the "Zhi District" has common interests and hobbies, the aggregation of content in the unit of "Zhi District" is equivalent to the App automatically helping users to filter and organize the feed based on interest classification, which not only reduces the high threshold of traditional RSS, but also provides a space for users with the same preferences to communicate and interact with each other.

Kyth, the founder of Small Universe, once expressed his thoughts on the failure of Google Reader in "Twenty Years of RSS", he believes that Google Reader failed because "people with similar interests and hobbies should have conversations, at least you can get resonance when you receive the same information, but Google Reader did not do this well."

Nowadays, it seems that the emergence of "recognition zone" just makes up for the flaws of Google Reader.

In addition, although the Bot still has the participation of algorithms in the process of crawling content, the importance of algorithms has been weakened, and interest and interpersonal sharing have occupied an important part.

The "Recommendation" function listed at the bottom of the "Recognition Area" function bar will not only push the content of the knowledge area added by the user himself, but also push the relevant content of other knowledge areas to help users further broaden the source of information in addition to their own interests.

The recognition area that weakens the sense of existence of the algorithm breaks the single-column information flow distribution mode of traditional byte-based products such as today's headlines and vibrato, and the multi-dimensional content push of the "recognition area" further breaks the user's "information cocoon".

In the early days of the Emergence of the Internet, simple classification directories and search engines can meet the search needs of users; and with the development of network technology and the growth of information content, only the use of algorithm recommendation can help users get rid of the "bitter sea" of information and improve efficiency; nowadays, in addition to efficiency, users are more concerned about content quality, and new needs will inevitably lead to new products.

From Douban, Instant, Tieba to RSS, rather than saying that "recognition area" is their "hodgepodge", it is better to say that this time Byte intends to gather the strengths of various apps and strive to find a content growth point in the "de-algorithm" era.

As Zhang Yiming once said: "What model and method a product chooses is secondary, and the most important thing is to what extent it meets the information needs of many people."

Read on