Death of The Creator
Sora marks the beginning of the third (and final) phase of social media.
The first phase of social media was defined by social graphs. Creators amassed followers, and when they posted content, it would be distributed to their followers’ newsfeeds. This guaranteed that at least some followers would consume a creator’s content, even if it wasn’t perfectly tailored to their interests. It wasn’t very efficient, but it worked.
Then came the second phase, defined by “the algorithm,” popularized by TikTok, and copied by Reels and every other platform. I called it “Recommendation Media” instead of social media (and wrote about it in The End of Social Media). This phase was far more efficient: creators no longer needed to build up a follower count to get distribution. They simply had to make good content and let the algorithm do the rest, finding the perfect consumer for any given video. This made content distribution dramatically more efficient, and platforms were rewarded with much higher engagement (a.k.a. addiction).
Ben Thompson summed it up perfectly in a conversation I had with him on this topic:
“The other angle is that to succeed with recommendation media, to be a recommendation engine, which I think is honestly another way to say to an aggregator, you need volume, you need huge amounts of stuff on your platform on both sides, both on the consumption side and the creation side. How do you get big volume? You get big volume through innovation, through doing something unique and so it touches on both sides of it in a way.”
We’re now entering the third (and final) phase of social media. Platforms love distribution efficiency because it drives consumption, and in this third phase, they’re guaranteed to get it. As generative models improve, platforms will pair on-the-fly content creation with algorithmic distribution. Instead of sourcing from creators, they can generate content tailored to individual preferences at scale. Production shifts from human to synthetic. With these models (like the ones we’re seeing in Sora and Vibes), platforms will increasingly generate content dynamically to perfectly match a consumer’s interests, rather than selecting from an ocean of human-made videos.
The good news is that people will keep creating, because it’s what we do. But the incentives will be very different.
On one end, we’ll see creators moving toward differentiated human experiences that resist commodification. Live events, community, and shared experiences are already on the rise, and that trend will accelerate. On the other end, we’ll see platforms rewarding uniqueness instead of eyeballs, likely through direct payments to creators when they feed the models (but skipping the distribution part). People will no longer be incentivized to simply reach as many viewers as possible, because platforms will be able to do that without them.
What platforms’ AI won’t be able to generate is something truly unique: a human’s name or likeness, the feeling a revered brand evokes, or a net-new meme or joke that only a human could dream up. We’re all going to help the models get smarter, until they no longer need us.
If there’s any silver lining to this third phase, it’s that some form of creativity will continue to be rewarded, at least for the foreseeable future.
Overall, the system will keep optimizing for distribution efficiency, and we’re fast approaching the point where platforms no longer need creators to feed the machine.