2013 钛媒体 Ti 创新者大会演讲

Speech at the 2013 TMTpost Ti Innovators Conference

> By Zhang Yiming · live speech at the TMTpost Ti Innovators Conference · edited and published by TMTpost (the organizer) · 2013-12-19 (speech given 2013-12-17) · [Original link](https://www.tmtpost.com/84589.html)
> Subheadings are the organizer's. Apparent mishearings in the live transcript are kept as written, with notes.

Today I saw that the theme of TMTpost's conference is "new media, new technology, new models", and I think we fit it pretty well, so I'm here to cheer the event on, not to crash it. Just now during the tea break, quite a few friends came to ask me, since as a startup we've been fairly low-key before and haven't talked publicly about our views on this industry or our philosophy on media. Today TMTpost invited me here to share, and I'll take this chance to share some of our different ideas with you.

Many people ask me: did you study journalism before? Did you work in media? Did you come from some newspaper? Actually not at all. Before this I was doing internet startups, working on product and technology. Some people ask, how did you get a news client to the top of the rankings, with over sixty million users? What I want to share is that the way we look at media is quite different from many of our peers, and today I'll mainly share where we're different.

How do I actually understand media?

The word media is often mentioned, and I don't know how you understand it. I see two understandings:

One is an organization that creates content: an organization that creates content is called a media outlet. In the English dictionary definition of media, it also has another meaning, the medium that distributes information. My understanding is that the medium that distributes information is what's called media; the word itself doesn't include creating content. So I think the very core of media is the distribution of information, and distribution is changing a great deal.

This wave of mobile and this wave of big data will bring very big changes to the distribution of information. For example, changes in devices: phones, Glass, letting you get the latest content anytime, anywhere; creating content becomes lower [in cost]; now more and more small media organizations and new media are popping up. As for information distribution, the likes of WeChat, Toutiao and Weibo have become platform-level channels for distributing information. [Translator's note: the source reads 创造内容变得更低, "creating content becomes lower", with no noun; "in cost" in brackets is the translator's reading.]

Let's look at a summary I made myself. I looked back: from the earliest notices on walls or bulletin boards, to newspapers brought by the development of printing technology, broadcasting brought by loudspeakers, the campus broadcasts from our school days, radio stations brought by wireless technology, then later television. From radio to television there was a big change in the information conveyed; from television to the internet, one of the biggest changes of the internet is that it's two-way and interactive; then to the mobile internet, where the main changes are in time, scenario and location. The internet era can be further divided: first portals, then search, then Weibo and recommendation engines, different ways of sending and receiving information. We can see that in every generation of change, technology played a very important role.

I added and subtracted (做了增减) along several dimensions. First, the radius of transmission: a campus broadcast, for example, is within a range of a thousand meters, while radio can cover the whole globe, a very important, landmark medium. Next is speed: one push on the mobile internet can reach users across the country within 10 minutes. Bandwidth means the capacity to transmit; after the internet it's the only one [that carries] content in every format, text and images, video, pictures. Next is the mode of interaction, one-way or two-way, and whether it can be one-way and two-way at once. Scenario and distribution mechanism: I put the distribution mechanism last, and it's very important. Listening to the radio or watching TV is simple one-to-many, but with the internet there's more and more personalization, and personalization is implemented in different ways. The earliest was subscription-based: subscribe to a source; on Weibo you can follow a person; later you could follow some topics. Personalization has been a very big dividing line in recent years. [Translator's note: 做了增减, literally "made additions and subtractions", is as written, perhaps for 做了比较 "made a comparison"; the sentence on bandwidth has no verb in the source, and "[that carries]" is the translator's.]

What I just did was summarize the relationship between changes in media and changes in technology. From another angle, I divide the links involved in media into these parts, and each of these links evolved differently 20 years ago, 10 years ago, 5 years ago and this year, with different products emerging. Take traditional media first. We all know media needs creation, needs topic selection; after topic selection the editor has to review; after review it has to be published, then filtered, which goes on the front-page headline; after filtering it has to be distributed, in what form of text and images, then distributed; after distribution there's interaction, with user comments and readers' letters. In the print era interaction was very slow; feedback on this issue's content only came in the next issue. In the internet era it was very real-time, and on the mobile internet you can give feedback and interact at any time, and then create again.

After a piece of content is distributed, readers have many opinions after reading, many thoughts, votes, shares, and some readers can create very good content again. Their comments and re-created content can be shared on further, to Weibo and WeChat, and finally there's the monetization of media itself. We can see that technology plays a big role in every area. In the earliest traditional media, news agencies, newspapers, reporters and editors did the gathering and editing of content; later, with Wikipedia, Tieba, Zhidao and Weibo, more and more people took part in technology creation (技术创造). This is a change brought by technology: creating content gets easier and easier. Many peers sign up for WeChat public accounts, or sign up for Weibo; you don't need to find a printing plant; you finish registering in 5 minutes and can start creating, and after creating you go on to distribute it. [Translator's note: 技术创造, "technology creation", is as written; the context suggests content creation.]

Let me give an example. Math enthusiasts used to be unable to get their work distributed; now it's relatively easy. On Weibo and WeChat you can have some accounts and filter for interested people; you talk up front, and people behind follow you. Of course this isn't very easy either; growing followers is a hard thing. Toutiao made a big innovation here: as long as the content is well written, as long as it has information in it, from the moment it's created it starts being distributed to interested people. As long as good content is created, even on its first day it can be distributed to interested users, and according to the article's quality it's distributed to different groups of people.

Monetization models

There's also a lot of innovation in monetization. It used to rely on media subscriptions and advertising; now in the internet era, forms like AdSense and Baidu Union are more and more common, and some internet products are trying direct publishing, like Douban Read; Qidian also created a monetization method, where authors attract loyal readers and are paid directly per creation. I think throughout this process, technology is driving every link, driving innovation.

Friends who know more about the internet, about foreign internet, will find that even in the internet era there are specialized companies for interaction, creation and sharing: in the US there's Diqus, in China there's the Bshare sharing plug-in, which also has very many users, specializing in sharing articles. Many web pages have a little button (按纽); click it and a row of icons appears for where to share to. These are internet technology companies making improvements specifically for sharing. 索伊美 one link is constantly innovating, with new things constantly coming out. In creation too, from blogs to Weibo, new ways keep being created. It may be a change of brand; for example Weibo is a change of format, and the ways of creating content keep evolving, and these evolutions have a lot to do with internet products and technology. [Translator's note: "Diqus" is as written, probably for Disqus; 按纽 is a typo for 按钮 "button"; 索伊美一个环节 is kept in the original characters and is apparently a mishearing of 所以每一个环节, "so every link".]

Some important changes have already happened. As bandwidth has increased, in these three to five years we've already seen a major change: text is turning more and more into multimedia. Before, few websites opened with a row of big pictures; now this design style is more and more common, and the precondition is the change in bandwidth technology. If you remember, 6-7 years ago both China and the US had a series of products that surfaced popular content, like 大企网, where netizens clicked up the hot content; now it's also more and more personalized, and topic-based content has already happened, such as an internet technology section or a media section; Tianya, for example, is changing toward topic-based content. [Translator's note: 大企网 is kept in the original characters; it may be a mishearing of Digg (掘客网 or similar); not confirmed.]

The biggest one in progress is 末国民教育互联网 turning toward PC internet. Now on phones we can do communication, do social networking, do some games and entertainment, and there's one most basic function, I think at least in the top three, which is getting information. At first there were mobile newspapers and phone push notifications; the likes of Toutiao, 资深 clients, Weibo and WeChat are all partly meeting the need to get information. [Translator's note: 末国民教育互联网 is kept in the original characters. 资深客户端, "senior clients", is probably a mishearing of 资讯客户端, "news clients".]

> [Note: 「末国民教育互联网向PC互联网转化」 is identical on both TMTpost pages (www and prewww), and no other version was found. Probably a transcription error at the event; from context it may mean "PC internet turning toward mobile internet"; the original is kept.]

The way information is organized also keeps changing. Portals are classification and navigation; then came search, which is a very precise distribution of traffic but needs active user behavior: you have to go and look actively. Then came following: around '08, the words subscribe and follow were very common: follow this person, follow this thing, follow this topic; information distribution based on follow relationships. In the last two years, with the development of technology, came distribution by recommendation technology, where you don't need to follow, because following also has many problems: is the granularity of what you follow big or small? If I follow technology, I'm not interested in pharmaceuticals, oil and natural gas, but if I don't follow it I'll miss a lot of content. If I follow a portal's tech channel, I can get a dozen or so pieces a day. The granularity and conditions of following can't be combined very richly, and the next step is recommendation. [Translator's note: the source writes 08年 for 2008.]

In this change, technology is different from before. In the review of media just now (刚刚回顾媒体的对话中), there were changes brought by printing and changes brought by radio; you can just watch video directly on phones, and as far as we know, watching video directly on phones is less than 25%. It's mainly not a change in communications technology; in short, it's a change in algorithms, or a change in big data. Big data can deepen machines' understanding of content: what this piece is about doesn't need an editor to judge; the machine can quickly know who it talks about, which category it belongs to, which topics it involves, which companies it involves. Understanding users is also very easy. For example, I'm someone interested in startups, interested in technology, interested in algorithms, interested in Beijing, perhaps interested in travel on weekends, but 工作不日感觉兴趣; machines already have the conditions to understand people very accurately. [Translator's note: 工作不日感觉兴趣 is kept in the original characters; it is garbled, perhaps for 工作日不感兴趣 "not interested [in travel] on workdays". 对话 ("dialogue") in the review phrase is as written.]

On these two foundations, understanding people on one hand and understanding information on the other, it becomes possible to let the machine connect (让机器这把) the two. [Translator's note: 这把 is as written, an apparent transcription slip.]

I often get this question: can machines really understand that accurately? Can pure algorithms really beat people? Algorithms can't beat people, but algorithms built on top of people can beat people. Because so-called algorithms stand on the basis of a large amount of user behavior: knowing what a media person in Beijing has recently liked to read, they'll guess what another media person in Beijing likes to do at this time. Standing on people's basis, you can write the ten most popular pieces, or the ten best pieces for each channel, but you can't show (显出) more niche channels; without algorithms there's no way to filter information for people at scale. This is the part where, as I understand it, technology plays the most critical role in this transformation.

I made a simple list: in this change there are different types of media, and I grouped them by mode of creation, content format, state of development and so on. Toutiao has brought a big change in distributing content, and I also summarized the several ways of distributing content, which I mentioned earlier: from 1999 to 2013, every three or four years a new way becomes popular and to some degree replaces the need for one type of product.

Amid these changes, how will media organizations change? Creating content has become easier: in 5 minutes you can set up an account and create content. On our platform content can be seen; within 5 minutes the content can be seen, without building followers, without advertising, without accumulating word of mouth; as long as the content is well created, tens of thousands of people see it on the first day. The only remaining difference now is that monetizing content isn't so easy. Writing a piece with a hundred thousand readers, that's actually not bad, but in the current industry environment you may still not get five thousand yuan of income from it. Some well-marketed public accounts can reach that, but it's still not that easy; it takes effort to run. If monetizing content also became very easy, if you only had to concentrate on creating content and its monetization could also happen automatically, with high conversion efficiency, it would bring very big changes to media organizations.

Some links are being replaced by tools and services. With the development of IT you don't need typesetting and printing, and when you write there'll be an editor program correcting you, so proofreading and editing in these links are replaced by tools. Distribution is replaced: you don't need to find distribution channels; it's replaced by platforms and software. On this basis, media organizations are becoming smaller, content creation is becoming platform-based, and there'll be a few big platforms 攻 broadly creating content. [Translator's note: 攻 is kept in the original character; it is probably a slip for 供 "supplying", i.e. big platforms supporting broad content creation.]

Our product

Finally, let me introduce Toutiao. We officially launched in August 2012, and in less than a year and a half we already have 70 million users, 10 million daily active users and 30 million monthly active users, with users making over 1 billion requests to our servers every day. We recommend different content to different people according to the information (根据信息不同的人), understand different people's interests, and give people different content under different conditions; the challenge of this algorithm is very big.

What have we brought to the industry? One more channel for distributing content. Based on users' interests, every day we can see real reposts, reads and click-through rates, and we've brought users' attention to many partners. We're not a media outlet that directly creates content; we distribute content, and the content shared out from us is all links to the original media.

My view is that technology should do well the things it should do itself: we should do well at understanding content and distributing data, and help media monetize. Some ask, what will become of media people? I think media people will do very well in the future, because content creation will get easier and easier; media and content creators will keep producing quality content, keep building brands, and interact better with users.

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.