首届世界互联网大会「新媒体新生态」分论坛发言
在首届世界互联网大会「新媒体新生态」分论坛上,张一鸣介绍今日头条两年积累1.8亿安装用户、日活近两千万,讲推荐引擎如何结合内容、场景和社交,说今日头条做的是广义的信息,希望消灭「早知道就好了」这句话,并谈到移动介质给内容形态和创作方式带来的变化。
Source: https://www.cac.gov.cn/2014-11/19/c_1113989679.htm
Remarks at the First World Internet Conference "New Media, New Ecosystem" Sub-Forum
> By Zhang Yiming · remarks at the "New Media, New Ecosystem" sub-forum of the First World Internet Conference · published on the Cyberspace Administration of China website (the organizer) · 2014-11-19 · [Original link](https://www.cac.gov.cn/2014-11/19/c_1113989679.htm) (also [People's Daily Online](http://media.people.com.cn/n/2014/1119/c14677-26054380.html))
> Apparent mishearings in the live transcript are kept as written, with notes.
I'm very glad to have the chance to share with you at this summit in Wuzhen, as an emerging mobile internet startup, some of our practice in media, in the transformation of media in the mobile era.
Let me first give a brief introduction. Actually Chairwoman Ma Li and Mr. Zhu both mentioned Toutiao earlier. Toutiao is indeed a personalized recommendation engine with data-driven at its core, very different from traditional clients; it's the first to recommend content entirely on the basis of technology. [Translator's note: 马利 = Ma Li; 朱总, "Mr. Zhu", is not further identified in the source.]
Let me share some of our data. Over the past two years we've accumulated 180 million installed users, with daily active users close to twenty million. As our models keep iterating, we've found that over the past 6 months of this year, as the model's data scale and range increased further, each user's daily 跃度 time [Note: probably 「活跃时间」 (active time) or 「使用时长」 (usage time).] has kept growing: it has grown 60% over the past 6 months, to close to 40 minutes a day. So we have some different 事件 [Note: probably 「实践」 (practice).] in recommendation engines in the mobile internet era, which I'll share with you. [Translator's note: the source writes 1.8亿 (180 million).]
With recommendation engines, people traditionally think of recommending based on content relevance: for example, you like entertainment; within entertainment you like singers; among singers you like Li Yuchun. That's content relevance.
Actually, besides content there's scenario: for example, whether you're traveling now, say attending the Wuzhen conference, whether you're at your usual location or on a trip; scenario is also an important [basis for] recommendation.
Mr. Zhu just mentioned social. Social is a typical collaborative recommendation: it uses your social relationships in real life, or social relationships online, to serve you in a "me for all, all for me" way. Besides social there are other collaborative methods: collaboration by geographic location, by similar information interests, including by similar occupations, including by similar scenarios.
So collaboration is an important way of exploring information. We often say personalization, personalization: will it more and more be information I 收载 [Note: 「收载」 is probably a transcription error; the intended meaning could not be confirmed.]? Generalization is what's common; social is the commonality of social relationships, and of course there are other commonalities.
Starting this year we've also stepped up pushing major news events, to maintain social consensus. Because we found that besides having their interests met, users also need to stay familiar with public topics, otherwise they'd have no way to talk with friends in various settings.
We have three things to recommend to you. First, we believe we're not news in the narrow sense, but information in the broad sense. In the past, newspapers, broadcasting and radio had either space limits or time limits; because of the bandwidth limits of space and channel, what they mostly conveyed was major social 什锦 [Note: probably 「社会事件」 (social events).]; we place more emphasis on information.
We believe that for someone taking the graduate entrance exam, admissions information is crucial; for a citizen, 公管 information is crucial; for a patient, progress on their disease is crucial. In the mobile internet era, the content media transmit will expand greatly, in both subjects and formats, and this has a lot to do with changes in content production too. [Translator's note: 公管 is as written, perhaps short for public administration (公共管理) information; not confirmed.]
In addition, we actually have a vision. Put simply, we want to eliminate the phrase "if only I'd known". I have a deep feeling about this myself: people often say "if only I'd known about such-and-such". That's the cost the information gap imposes on society. We believe more effective distribution can raise social efficiency and lower social costs.
Third, we believe that in the internet era... before, in the print era, once content was printed on 纸说 [Note: probably 「纸上」 (on paper).] it was out, with no further interaction. In the mobile internet era, this medium's characteristic is that it's two-way: users' behavior can be returned to the server at any time, returned to the content publisher, and the content publisher has the chance to connect directly with users without going through a third party. So we believe the mobile internet will connect people and information, and we're already seeing this trend: many companies have opened Toutiao Hao accounts, and while publishing their professional content, they provide services to users at the same time.
So we not only help users find quality content, but also help quality content find readers. Before, very 强势 [Note: probably 「弱势」 (weak) or 「小众」 (niche), the opposite of the following "had a hard time reaching users".] content had a hard time reaching users. For example, if you studied science history or culture, it was hard to distribute through post offices and airports, because distribution costs were high and users were sparsely distributed. With a recommendation engine, we have the chance to find users' 常规 [Note: probably 「长尾」 (long tail).] information and recommend it to them in all kinds of scenarios. We've now found that some especially 常委 [Note: probably 「长尾」 (long tail).] information interests have already emerged on Toutiao.
In the last two years there have been all kinds of changes, whether social or algorithms; the biggest change is the change of medium, and the change of medium brings simultaneous changes in many respects. The earliest medium was notices, or blackboard newspapers or wall posters; later electromagnetic waves, newspapers and the development of printing; later TV stations. Now in our forms of information transmission, content is no longer just text, nor just pictures, nor just video.
We used to say text, pictures, text-and-pictures, video; now votes and games are more and more forms of content too. In WeChat Moments we often see content forms with all kinds of rich recruitment (招募) methods. Second, we've found it really is two-way: the actions users produce, whether comments or game test results, are also a further contribution to content.
For creators too, shooting and writing content on phones is more convenient, so this medium has brought big changes. This medium brings all kinds of changes from creation to distribution to sharing to monetization, and the media of this era must fit these changes.
Finally, I'd like to share what changes the way content is produced will bring. More and more people are taking part in creation, from traditional UGC to PGC. On our platform we've found that third-party, non-media creating organizations are growing fast and already account for half of our traffic impressions. In future 80% of content may well be created by people, by organizations, as an expression of cognition (认知). Only the rest will be produced by professional institutions: state institutions, like those published by 工商工具 [Note: probably 「工商局」 (industry and commerce bureau).] and the Beijing Public Security Bureau; financial institutions too, like 格善理财 [Note: the institution's name probably has a transcription error; not confirmed.]; and corporate media, like Xiaomi, which publishes a lot of content on Toutiao each time, and their new product information is very popular. [Translator's note: 认知 ("cognition") is as written, possibly a mishearing; 格善理财 is kept in the original characters.]
On this basis, machines free people, and people will take part more in creation, in the transformation of how information is produced. Just as Shuli said a moment ago, I believe there's more creative work in the future waiting for people to accomplish. [Translator's note: 舒立, "Shuli", is given by first name only in the source.]
Copyright: © LookMinds. Our own edited text and translation, not a substitute for the original; rights in the original belong to the author and the platform. Please see the original link.