Live data from Hacker News

TikTok and the Sorting Hat

eugenewei.com

41–50 of 233 posts

Re: TikTok and the Sorting Hat

#41
post #5

Earlier quoted context omitted.

Which other chinese startups with machine-learning as its core? Do you mind listing some examples?

I think he might be referring to companies like SenseTime which were in fields like image/video recognition. I think for fields like this there is a cultural barrier which prevents companies like these from being successful in US. Say, if FB starts using its AI chops to build smart video analysis tools which can be used by law enforcement, you would probably have employee mutiny at your hands. I was seeing a Vice vid…

You also have face recognition used in law enforcement in Britain too, though. And from top of my head, Britain was the first country to mass roll out CCTV system as well.

https://www.nytimes.com/2019/09/04/business/facial-recogniti...

Re: TikTok and the Sorting Hat

#44
post #31
post #21

Earlier quoted context omitted.

The point is not media format (short video clips) but the "sorting hat". That it breaks the traditional social media model with a news feed, discussions, a social graph of friends. I think it is completely meaningful to call it next generation (or second generation where the social graph model is the first generation). Also notice that this model can be applied to other media formats: Text, pictures, audio ...

>That it breaks the traditional social media model with a news feed, discussions, a social graph of friends. don't most social networks have the "sorting hat" in the form of algorithmic feeds?

The article goes into this - for people just jumping into the comments:

The author states that while western algorithms are based on your follow graph (e.g. Instagram is relatively useless until you follow someone and even then your feed is based on your follower graph, like what people you follow like), TikTok builds this data on video features. This increases TikTok's stickyness because you don't need to do anything other than use the app for it adjust to your tastes. There's no need to "import" your contacts or suggest people you should follow, it just "knows" after you watch a couple videos.

Re: TikTok and the Sorting Hat

#45
post #36

Earlier quoted context omitted.

If we were really seeing the effect of monopolies how would Tiktok hold ground in US. FB could have actively gone after it and not allowed FB authentication or ads of Tiktok on FB. In fact, as mentioned in the article Tiktok grew by leveraging the dominant platforms like FB, Instagram. I would still wait and watch on Tiktok phenomenon. It is a viral thing but has no moat as such and only $0.5 billion in expected reve…

I think OP's point is that the American startup that could have turned into what TikTok is today (Vine) was bought only so it could be shut down.

Vine was acquired pre launch. It was launched and allowed to run for a good amount of time before being shut down.

Twitter isn't an example of a well run company considering the alienating things they have done to third party devs but I wouldn't call this an effect of monopoly. Even ByteDance acquired Musical.ly which became Tiktok, they just had a clear idea of what to do with it.

Re: TikTok and the Sorting Hat

#46
Modern day tobacco. 100% unregulated addiction.

FB, TWTR, GOOG, INSTA, everyone switched from timeline to personal recommendation engines and TikTok is just doing it so much better. Where will things be in 5, 10, 50 years? Terrifying and exciting.

Do these tech companies sit on research to suggest that these algorithms can induce behavior-changing tactics through trial-and-error? Will they all have their tobacco-causes-cancer moment?

Old-Age Fear: YouTube algorithm sending people down the conspiracy rabbit hole

New-Age Fear (Hyper-Efficient): TikTok secretly radicalizing individual in X?

The big question strikes: "what is free will?"

Amazing read

Re: TikTok and the Sorting Hat

#47
post #7

TikTok really is next-generation social media. Interesting that this should come out of China and not the west. Maybe we are seeing the effect of monopolies stifling innovation.

>> the effect of monopolies stifling innovation.

Less monopolies, more so western culture. Many of TikTok's algorithms would be subject to great scrutiny if run by a western company. YouTube would face riots if it were caught openly supressing or hiding videos of disabled, unattractive, or simply poor people. But such "innovation" is fine for a Chinese company.

https://slate.com/technology/2019/12/tiktok-disabled-users-v...

https://www.theguardian.com/technology/2020/mar/17/tiktok-tr...

Re: TikTok and the Sorting Hat

#48
Eugene mentions that the algorithm is the one that built this latent interest graph and this algorithm itself is worth a lot of money. Given that most personalization is using ML/Deep Learning and various video features like scene descriptions, object identification in each scene etc, Is it really that proprietary to warrant this level of premium. Once the awareness of interest graph could be built to this level of efficiency using usual techniques, Is the algorithm and approach exclusive to TikTok ?

Awesome Insights from Eugene !!

Re: TikTok and the Sorting Hat

#49
post #31

Earlier quoted context omitted.

>That it breaks the traditional social media model with a news feed, discussions, a social graph of friends. don't most social networks have the "sorting hat" in the form of algorithmic feeds?

The article goes into this - for people just jumping into the comments: The author states that while western algorithms are based on your follow graph (e.g. Instagram is relatively useless until you follow someone and even then your feed is based on your follower graph, like what people you follow like), TikTok builds this data on video features. This increases TikTok's stickyness because you don't need to do anythin…

Youtube has been doing this for years. Clear your cookies then start watching youtube videos. Youtube will immediately begin tuning video suggestions to what you watch, no contact importing or channel subscribing necessary.

Re: TikTok and the Sorting Hat

#50

this doesn’t surprise me at all. I work at a consumer company and we use neural networks to personalize feeds. Thousands of features feed in and each member gets a different feed. It doesn’t take much / long to personalize. What I am surprised by is that YouTube or others dont use as effectively. Neural networks are very well known. or, is there something “next level” that TikTok is using that isn’t well known? Also…

My gut instinct is that Netflix can't use the personalized algorithmic approach that TikTok succeeds with because Netflix has to pay dearly for its content, where TikTok does not. Because of the cost structure of licensing, Netflix cannot afford to have the diversity of content that TikTok's user-generated approach enables. A small content library makes it impossible to build a meaningfully personalized feed.
Post reply on HN