{"id":4010,"date":"2022-04-04T19:48:43","date_gmt":"2022-04-04T19:48:43","guid":{"rendered":"https:\/\/linksus.net\/how-tiktok-reads-your-mind-the-new-york-times\/"},"modified":"2022-04-04T19:48:43","modified_gmt":"2022-04-04T19:48:43","slug":"how-tiktok-reads-your-mind-the-new-york-times","status":"publish","type":"post","link":"https:\/\/linksus2.linksus.net\/index.php\/2022\/04\/04\/how-tiktok-reads-your-mind-the-new-york-times\/","title":{"rendered":"How TikTok Reads Your Mind – The New York Times"},"content":{"rendered":"

Advertisement
Supported by
the media equation
It\u2019s the most successful video app in the world. Our columnist has obtained an internal company document that offers a new level of detail about how the algorithm works.
Send any friend a story<\/strong>
As a subscriber, you have 10 gift articles<\/strong> to give each month. Anyone can read what you share.
By <\/span>Ben Smith<\/a><\/span>
There are four main goals for TikTok\u2019s algorithm: \u7528\u6237\u4ef7\u503c, \u7528\u6237\u4ef7\u503c (\u957f\u671f), \u4f5c\u8005\u4ef7\u503c, and \u5e73\u53f0\u4ef7\u503c, which the company translates as \u201cuser value,\u201d \u201clong-term user value,\u201d \u201ccreator value,\u201d and \u201cplatform value.\u201d
That set of goals is drawn from a frank and revealing document for company employees that offers new details of how the most successful video app in the world has built such an entertaining \u2014 some would say addictive \u2014 product.
The document, headed \u201cTikTok Algo 101,\u201d was produced by TikTok\u2019s engineering team in Beijing. A company spokeswoman, Hilary McQuaide, confirmed its authenticity, and said it was written to explain to nontechnical employees how the algorithm works. The document offers a new level of detail about the dominant video app, providing a revealing glimpse both of the app\u2019s mathematical core and insight into the company\u2019s understanding of human nature \u2014 our tendencies toward boredom, our sensitivity to cultural cues \u2014 that help explain why it\u2019s so hard to put down. The document also lifts the curtain on the company\u2019s seamless connection to its Chinese parent company, ByteDance, at a time when the U.S. Department of Commerce is preparing a report on whether TikTok poses a security risk to the United States.
If you\u2019re among the billion people (literally!) who spend time on TikTok every month, you\u2019re familiar with the app as 2021\u2019s central vehicle for youth culture and online culture generally. It displays an endless stream of videos and, unlike the social media apps it is increasingly displacing, serves more as entertainment than as a connection to friends.
It succeeded where other short videos apps failed in part because it makes creation so easy, giving users background music to dance to or memes to enact, rather than forcing them to fill dead air. And for many users, who consume without creating, the app is shockingly good at reading your preferences and steering you to one of its many \u201csides,\u201d whether you\u2019re interested in socialism or Excel tips or sex, conservative politics or a specific celebrity. It\u2019s astonishingly good at revealing people\u2019s desires even to themselves \u2014 \u201cThe TikTok Algorithm Knew My Sexuality Better Than I Did,<\/a>\u201d reads one in a series of headlines about people marveling at the app\u2019s X-ray of their inner lives.
TikTok has publicly shared the broad outlines of its recommendation system, saying it takes into account factors including likes and comments as well as video information like captions, sounds and hashtags. Outside analysts have also sought to crack its code. A recent Wall Street Journal report demonstrated<\/a> how TikTok relies heavily on how much time you spend watching each video to steer you toward more videos that will keep you scrolling, and that process can sometimes lead young viewers down dangerous rabbit holes, in particular toward content that promotes suicide or self-harm \u2014 problems that TikTok says it\u2019s working to stop by aggressively deleting content that violates its terms of service.
The new document was shared with The New York Times by a person who was authorized to read it, but not to share it, and who provided it on the condition of anonymity. The person was disturbed by the app\u2019s push toward \u201csad\u201d content that could induce self-harm.
The document explains frankly that in the pursuit of the company\u2019s \u201cultimate goal\u201d of adding daily active users, it has chosen to optimize for two closely related metrics in the stream of videos it serves: \u201cretention\u201d \u2014 that is, whether a user comes back \u2014 and \u201ctime spent.\u201d The app wants to keep you there as long as possible. The experience is sometimes described as an addiction, though it also recalls a frequent criticism of pop culture. The playwright David Mamet, writing<\/a> scornfully in 1998 about \u201cpseudoart,\u201d observed that \u201cpeople are drawn to summer movies because they are not <\/em>satisfying, and so they offer opportunities to repeat the compulsion.\u201d
To analysts who believe algorithmic recommendations pose a social threat, the TikTok document confirms their suspicions.
\u201cThis system means that watch time is key. The algorithm tries to get people addicted rather than giving them what they really want,\u201d said Guillaume Chaslot, the founder of Algo Transparency, a group based in Paris that has studied YouTube\u2019s recommendation system and takes a dark view of the effect of the product on children, in particular. Mr. Chaslot reviewed the TikTok document at my request.
\u201cI think it\u2019s a crazy idea to let TikTok\u2019s algorithm steer the life of our kids,\u201d he said. \u201cEach video a kid watches, TikTok gains a piece of information on him. In a few hours, the algorithm can detect his musical tastes, his physical attraction, if he\u2019s depressed, if he might be into drugs, and many other sensitive information. There\u2019s a high risk that some of this information will be used against him. It could potentially be used to micro-target him or make him more addicted to the platform.\u201d
The document says watch time isn\u2019t the only factor TikTok considers. The document offers a rough equation for how videos are scored, in which a prediction driven by machine learning and actual user behavior are summed up for each of three bits of data: likes, comments and playtime, as well as an indication that the video has been played:
Plike X Vlike + Pcomment X Vcomment + Eplaytime X Vplaytime + Pplay X Vplay<\/em>
\u201cThe recommender system gives scores to all the videos based on this equation, and returns to users videos with the highest scores,\u201d the document says. \u201cFor brevity, the equation shown in this doc is highly simplified. The actual equation in use is much more complicated, but the logic behind is the same.\u201d
The document illustrates in detail how the company tweaks its system to identify and suppress \u201clike bait\u201d \u2014 videos designed to game the algorithm by explicitly asking people to like them \u2014 and how the company thinks through more nuanced questions.
\u201cSome authors might have some cultural references in their videos and users can only better understand those references by watching more of the author\u2019s videos. Therefore, the total value that a user watches all those videos is higher than the values of watching each single video added up,\u201d the document says. \u201cAnother example: if a user likes a certain kind of video, but the app continues to push the same kind to him, he would quickly get bored and close the app. In this case, the total value created by the user watching the same kind of videos is lower than that of watching each single video, because repetitiveness leads to boredom.\u201d
\u201cThere are two solutions to this issue,\u201d the document goes on. \u201cMake some assumptions, and break down the value into the value equation. For instance, in terms of repeated exposure, we could add a value \u2018same_author_seen,\u2019 and for the boredom issue, we could also add a negative value \u2018same_tag_today.\u2019 Other solutions besides value equation may also work, such as forced recommendation in users\u2019 for u feed and dispersion etc. For example, the boredom issue can be solved through dispersion.\u201d
Another chart in the document indicates that \u201ccreator monetization\u201d is one of the company\u2019s goals, a suggestion that TikTok may favor videos in part if they are lucrative, not just entertaining.
Julian McAuley, a professor of computer science at the University of California San Diego, who also reviewed the document, said in an email that the paper was short on detail about how exactly TikTok does its predictions, but that the description of its recommendation engine is \u201ctotally reasonable, but traditional stuff.\u201d The company\u2019s edge, he said, comes from combining machine learning with \u201cfantastic volumes of data, highly engaged users, and a setting where users are amenable to consuming algorithmically recommended content (think how few other settings have all of these characteristics!). Not some algorithmic magic.\u201d
Mr. McAuley added that he was a bit perplexed about why people were always asking him about TikTok.
\u201cThere seems to be some perception (by the media? or the public?) that they\u2019ve cracked some magic code for recommendation, but most of what I\u2019ve seen seems pretty normal,\u201d he wrote.
And indeed, the document does much to demystify the sort of recommendation system that tech companies often present as impossibly hard for critics and regulators to grasp, but that typically focus on features that any ordinary user can understand. The Journal\u2019s coverage of leaked Facebook documents, for instance, illustrated how Facebook\u2019s decision to give more weight to comments helped divisive content spread. While the models may be complex, there\u2019s nothing inherently sinister or incomprehensible about the TikTok recommendation algorithm outlined in the document.
But the document also makes clear that TikTok has done nothing to sever its ties with its Chinese parent, ByteDance, whose ownership became a spasmodic focus at the end of President Donald J. Trump\u2019s administration in 2020, when he attempted to force the sale of TikTok<\/a> to an American company allied with his administration, Oracle.
The TikTok document refers questions to an engineering manager whose LinkedIn biography says he works on both TikTok and ByteDance\u2019s similar Chinese app, Douyin, offering a glimpse at the remaining global element of an increasingly divided tech industry, the engineering talent. According to LinkedIn, the engineering manager attended Peking University, received a master\u2019s degree in computer science at Columbia University and worked for Facebook for two years before coming to ByteDance in Beijing in 2017. The document is written in clear, but nonnative, English, and comes from the perspective of the Chinese tech industry. It makes no references, for instance, to rival American companies like Facebook and Google, but includes a discussion of \u201cif Toutiao\/Kuaishou\/Weibo have done something similar, can we launch the same strategy as they have done?\u201d
Marriages and divorces. <\/strong>Faced with a soaring divorce rate, China introduced a rule forcing married couples to undergo a 30-day \u201ccooling off\u201d period before formally parting ways. The move seems to have been effective at reducing divorces<\/a>, but is unlikely to help with a demographic crisis fueled by a decline in marriages.<\/span>
China Eastern Airlines crash. <\/strong>Emergency workers found no survivors<\/a>\u00a0after a Boeing 737 plane carrying 132 people crashed in the southern region of Guangxi. The crash of China Eastern Airlines Flight 5735 is the country’s worst air disaster in more than a decade<\/a>.<\/span>
The war in Ukraine. <\/strong>Despite calls from other world leaders for China to play a more proactive role in pressing Russia to negotiate an end to the war, Beijing has instead tried to keep its distance<\/a>. The result has left China, diplomatically, on the sidelines of the conflict<\/a>.<\/span>
Omicron surge. <\/strong>As China grapples with its worst Covid-19 outbreak since the beginning of the pandemic, Beijing is trying to fine-tune its \u201czero Covid\u201d playbook<\/a>, ordering officials to quash outbreaks<\/a>\u00a0but also find ways to limit the economic pain<\/a>\u00a0involved.<\/span>
TikTok\u2019s development process, the document says, is closely intertwined with the process of Douyin\u2019s. The document at one point refers TikTok employees to the \u201cLaunch Process for Douyin Recommendation Strategy,\u201d and links to an internal company document that it says is the \u201csame document for TikTok and Douyin.\u201d
TikTok employees are also deeply interwoven into ByteDance\u2019s ecosystem. They use a ByteDance product called Lark, a corporate internal communications system like Slack but with aggressive performance-management features aimed at forcing employees to use the system more. There is, for instance, a graphic that tells you whether you have performed actions \u2014 like opening messages \u2014 more or less than your co-workers, according to screenshots I was given.
Concern about Chinese consumer technology is bipartisan in the United States. President Trump\u2019s executive order attempting to ban the app in August 2020 warned that TikTok\u2019s \u201cdata collection threatens to allow the Chinese Communist Party access to Americans\u2019 personal and proprietary information.\u201d The Chinese government could \u201cbuild dossiers of personal information for blackmail, and conduct corporate espionage,\u201d it said. That ban stalled in court and faded after the presidential election. President Biden rescinded the executive order, but his administration then announced<\/a> its own investigation into security threats posed by TikTok, with an unnamed senior administration official telling<\/a> reporters that China was \u201cworking to leverage digital technologies and American data in ways that present unacceptable national security risks.\u201d
In an emailed statement, Ms. McQuaide said that \u201cwhile there\u2019s some commonality in the code, the TikTok and Douyin apps are run entirely separately, on separate servers, and neither code contains user data.\u201d
She also said, \u201cTikTok has never provided user data to the Chinese government, nor would we if asked.\u201d
TikTok, whose chief executive lives in Singapore, hired a raft of well-connected American and European executives and security experts as political pressure intensified under Mr. Trump. It says it has no formal headquarters. It has sought to soothe American concerns by storing user data in the United States, with a backup in Singapore.
The American government\u2019s security concerns come in two forms. The first, as Mr. Trump suggested in his executive order, is whether the vast trove of data TikTok holds \u2014 about the private sexual desires of fans of the app who might end up becoming American public officials, for instance \u2014 should be viewed as a national security issue. There\u2019s no evidence the data has ever been used that way, and TikTok is hardly the only place Americans share details of their lives on social media. The second concern is whether TikTok censors politically sensitive posts.
A report<\/a> this year by Citizen Lab, the cybersecurity watchdog organization in Toronto, suggested that both of these concerns are, at best, latent: It did not find any indication that TikTok was either censoring sensitive topics or transmitting data to China.
But TikTok\u2019s glimpses of people\u2019s inner lives are unusual. Another screenshot shared with me indicates that its content moderators have access not just to videos posted publicly, but also to content sent to friends or uploaded to the system but not shared, a difference from apps like WhatsApp and Signal that provide end-to-end encryption.
The second question is whether the Chinese government could use the platform to spread propaganda. After getting caught<\/a> censoring a video condemning the mass detention of minority Muslims in China, TikTok has allowed criticism of the country\u2019s government. For instance, the hashtag #whereispengshuai, a reference to the Chinese tennis star who accused a top Chinese leader of sexual assault<\/a>, autocompletes in the system, though TikTok videos with that hashtag have few views. There is no independent way of telling whether the company is suppressing the search, which has far more engagement on Twitter but similarly little on Instagram.
Some American analysts see TikTok as a profound threat; others view it as the kind of clueless panic that Americans now approaching middle age faced when their parents warned them that if they shared details of their lives on social media, they\u2019d never get a job. Many, many other products, from social networks to banks and credit cards, collect more precise data on their users. If foreign security services wanted that data, they could probably find a way to buy it from the shadowy industry of data brokers.
\u201cFreaking out about surveillance or censorship by TikTok is a distraction from the fact that these issues are so much bigger than any specific company or its Chinese ownership,\u201d said Samm Sacks, a cybersecurity policy fellow at the research organization New America. \u201cEven if TikTok were American-owned, there is no law or regulation that prevents Beijing from buying its data on the open data broker market.\u201d
One thing that reporting this column has reminded me: The menace that TikTok poses to American national security appears to be entirely hypothetical, and depends on your analysis of both the U.S.-China relationship and the future of technology and culture. But the algorithm\u2019s grasp on what keeps me hooked \u2014 between trick tennis shots, Turkish food videos and all the other things it\u2019s figured out I like to watch \u2014 did pose a clear and present danger to my ability to finish this column.
Advertisement<\/p>\n

source<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

AdvertisementSupported bythe media equationIt\u2019s the most successful video app in the world. Our columnist has obtained an internal company document that offers a new level of detail about how the algorithm works.Send any friend a storyAs a subscriber, you have 10 gift articles to give each month. Anyone can read what you share.By Ben SmithThere […]<\/p>\n","protected":false},"author":869,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[],"tags":[],"_links":{"self":[{"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/posts\/4010"}],"collection":[{"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/users\/869"}],"replies":[{"embeddable":true,"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/comments?post=4010"}],"version-history":[{"count":0,"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/posts\/4010\/revisions"}],"wp:attachment":[{"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/media?parent=4010"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/categories?post=4010"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/linksus2.linksus.net\/index.php\/wp-json\/wp\/v2\/tags?post=4010"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}