Key takeaways


  • The group inertia that is your enemy before you’ve coalesced a community is your friend after it’s formed. Anyone who’s ever hosted a party and provided booze knows it’s often hard to get the last stragglers to leave. We are a social species. (View Highlight)
    • Note: Inertia works both ways, both acceleration and deceleration.
  • I tried switching to the Following from the For You feed, but it seems the Following feed is strictly reverse chronological. This is a serious regression to the early days of Twitter when you had to check your feed frequently to hope to catch a good tweet from any single person you followed. We tried this before; it was terrible then, it’s terrible now. (View Highlight)
    • Note: He really dislikes the strictly chronological feed, but some people love it.
  • ut what if there was a way to build an interest graph for you without you having to follow anyone? What if you could skip the long and painstaking intermediate step of assembling a social graph and just jump directly to the interest graph? And what if that could be done really quickly and cheaply at scale, across millions of users? And what if the algorithm that pulled this off could also adjust to your evolving tastes in near real-time, without you having to actively tune it? (View Highlight)
    • Note: How TikTok is built. You don’t need the social graph anymore because the algorithm is so efficient in selecting content.
  • Twitter doesn’t see a lot of passive negative sentiment; it’s a structural blind spot. In a continuous scrolling interface with multiple tweets on screen at any one time, it’s hard to tell disapproval from apathy or even mild approval because the user will just scroll past a tweet for any number of reasons. (View Highlight)
    • Note: Compare this to TikTok where you see one video at a time, and the algorithm can learn about your disapproval much more efficiently.
  • Here’s another way to think about it. The difference between Twitter and an algorithmic entertainment network like TikTok is that you could fairly quickly reconstitute TikTok even without its current graph because its graph is a much less critical input to its algorithm than the user reactions to any random sequence of videos they’re served. (View Highlight)
    • Note: TikTok has tried to add some social graph features but with little success. It simply doesn’t need them.