• The makers of TikTok, who mostly had never been outside of China, cracked the cultural code of Americans (primarily, the youth) based on a largely conventional video matching algorithm.
• The real question is how does the “creativity network effects” flywheel work between TikTok's video creation and distribution — from its origination to mutation to dissemination?
—Origination of a Creative Tool Suite: Underrated set of good camera tools, high quality filters, editing functions, all freely available to download by the public;
—Mutation: Creator content pushes in different directions because the app incorporates licensed music, meme trends, and gamifying the app with challenges from corporate sponsors, etc;
—Dissemination: TikTok records every single action you take in app - whether you like a video, how long you view a video, what actions you take afterwards. Eventually, it sees you for who you really are, even if the developers don't even speak or understand your native tongue.
"A machine learning algorithm can pierce the veil of cultural ignorance. Culture can be abstracted." —Wow! That is insane and a bit scary ("eerily perceptive').
"Culture Eats Strategy for Breakfast," but this algorithm seems like it's going to run culture over and not look back.
Finally, for readers who don't know what FYP is = "For Your Page" algorithm, which you refer to it as the "Magic" that powers TikTok.
Thank you Eugene I really enjoyed this. I think you nailed something that's been in my head a while which is existing strategy frameworks (i.e. Porter's 5 Forces, value chains) certainly have limitations when trying to diagnose industry economics and strategic positioning today. I was reading a really nice text on this called Economics of Strategy and they have one chapter on the economics of information and when I finished the book I thought: is this just too dated? The 5 Forces works well when industry boundaries are well defined, but abundance has made this difficult and blurred these boundaries. Declining or zero marginal costs (i.e. Ben Thompson idea) make bundling and abundance possible. Then, you have ecosystem based business models (i.e. Apple) can be involved in so many value chains and business models its difficult to parse. These frameworks are still useful, but have to be used more carefully. I hope someone takes a crack at it in a more immersive manner, but Byrne, Ben Thompson, and others have definitely laid a nice foundation.
Eugene, great post!
I particularly found your podcast with @Soonaorlater on "16 Minutes for the News" very helpful to flesh out your analysis on TikTok. I also appreciated the fact you two focused on the core technology issues and user base and not so much on the latest fashionable news. See https://a16z.com/2020/09/18/16mins-tiktok-seeing-like-an-algorithm-friendly-design-creativity-network-effects-video/
This was particularly insightful to me:
• The makers of TikTok, who mostly had never been outside of China, cracked the cultural code of Americans (primarily, the youth) based on a largely conventional video matching algorithm.
• The real question is how does the “creativity network effects” flywheel work between TikTok's video creation and distribution — from its origination to mutation to dissemination?
—Origination of a Creative Tool Suite: Underrated set of good camera tools, high quality filters, editing functions, all freely available to download by the public;
—Mutation: Creator content pushes in different directions because the app incorporates licensed music, meme trends, and gamifying the app with challenges from corporate sponsors, etc;
—Dissemination: TikTok records every single action you take in app - whether you like a video, how long you view a video, what actions you take afterwards. Eventually, it sees you for who you really are, even if the developers don't even speak or understand your native tongue.
"A machine learning algorithm can pierce the veil of cultural ignorance. Culture can be abstracted." —Wow! That is insane and a bit scary ("eerily perceptive').
"Culture Eats Strategy for Breakfast," but this algorithm seems like it's going to run culture over and not look back.
Finally, for readers who don't know what FYP is = "For Your Page" algorithm, which you refer to it as the "Magic" that powers TikTok.
Best,
Chris
(@ChrisHarveyEsq)
Thank you Eugene I really enjoyed this. I think you nailed something that's been in my head a while which is existing strategy frameworks (i.e. Porter's 5 Forces, value chains) certainly have limitations when trying to diagnose industry economics and strategic positioning today. I was reading a really nice text on this called Economics of Strategy and they have one chapter on the economics of information and when I finished the book I thought: is this just too dated? The 5 Forces works well when industry boundaries are well defined, but abundance has made this difficult and blurred these boundaries. Declining or zero marginal costs (i.e. Ben Thompson idea) make bundling and abundance possible. Then, you have ecosystem based business models (i.e. Apple) can be involved in so many value chains and business models its difficult to parse. These frameworks are still useful, but have to be used more carefully. I hope someone takes a crack at it in a more immersive manner, but Byrne, Ben Thompson, and others have definitely laid a nice foundation.