TikTok – Remote control lobby.

TikTok lobbies for an extension

  • TikTok’s transparency briefing actually explains very little about how the algorithm works and looks to me to be an attempt to get the press to lobby the US administration for an extension to the September 15th deadline after which TikTok will be blocked.
  • TikTok held a virtual media event on Wednesday that attempted to explain how the recommendation algorithm works and to demonstrate how it uses the data it gathers with a virtual tour of its “transparency center”.
  • This strategy works perfectly for TikTok because the vast majority of the lay press has little to no understanding of artificial intelligence meaning that it is ill-equipped to critique the commentary.
  • TikTok gave a lot of details around how the process works including how new users are profiled and placed into preference clusters and how it avoids leaving users stuck in bubbles and seeing the same sort of content over and over again.
  • What I think the press has missed is that what TikTok explained here is the easy bit which everyone knows how to do and in many cases is already doing.
  • The difficult bit is how its algorithms are able to perform at a level which is much higher than anyone else which I have long believed is the critical key to TikTok’s success.
  • TikTok’s algorithms have an uncanny ability to characterise uploaded videos which comes not from metadata and labelling but from analysing the videos themselves.
  • This is very difficult to get right and is an area of intense research in both the corporate and government sectors in both China and the West.
  • This functionality is based on deep learning and ByteDance has worked out just how to train and tweak its algorithms to provide a level of performance that is better than anyone else’s.
  • The problem with these types of algorithms is that they are black boxes where it is almost impossible to know exactly how they work and even ByteDance will have only a sketchy understanding of how they are able to perform to the level that they do.
  • Hence, ByteDance’s core IP lies in how these deep learning algorithms are structured, trained and what data to feed them and in what format to produce the right results.
  • This is the bit of which there was no mention because it is central to ByteDance’s value proposition but this is the bit where all of the security concerns are.
  • If the value-creating methodology was as simple and clear as TikTok seems to claim, there would not have been serious wrangling in the acquisition talks with Microsoft over ownership and usage of the algorithms.
  • The company also discussed its data security and privacy concerns and here the focus is on keeping US data within the USA and being able to deal quickly with “incidents”.
  • I think that this event was designed to give the impression of openness without having to give away any corporate secrets in order to generate positive press as the deadline approaches.
  • Next Tuesday 15th September is the deadline when TikTok must be no longer Chinese owned or face being blocked, and I think that ByteDance needs an extension.
  • This is because discussions with acquiring companies have proved to be more complicated than expected and because the new Chinese regulations (see here) have thrown a spanner in the works.
  • The net result is that I don’t think that ByteDance will make the deadline and it faces being blocked.
  • However, a block may take a little time to enact meaning that there will be a little leeway before users start to suffer.
  • Either way a resolution needs to happen fairly quickly as the example of India demonstrates that users could quite quickly move over to rival services should TikTok no longer be available.
  • There are no shortages of innovative apps waiting in the wings to fill any gaps left by TikTok should it not move fast enough to prevent a ban.
  • Instagram and Snap are two that leap immediately to mind but there is no shortage of others.

RICHARD WINDSOR

Richard is founder, owner of research company, Radio Free Mobile. He has 16 years of experience working in sell side equity research. During his 11 year tenure at Nomura Securities, he focused on the equity coverage of the Global Technology sector.