Mark Pincus was asked what he would build if he lost everything and had to start again from nothing. His answer was a social network built to help people leave it. He would "reimagine and reinvent our online social experience," he said, and "leverage AI and agents and gamify how little time you can spend there." To gamify something is to turn it into a game, with points or rewards. In his version, the prize would go to spending less time on the platform, not more.
Pincus is best known as the founder of the games company Zynga. He gave the answer on stage at a live event for the Build with Gemini XPRIZE, a $2 million competition for AI businesses, on Moonshots with Peter Diamandis. The episode was recorded on 25 September 2026 and published three days later. Diamandis, who founded XPRIZE, put the same question to the competition's judges. Palmer Luckey, who founded the virtual-reality company Oculus and the defense company Anduril, was one of them. Both men were on the judging panel, according to XPRIZE's announcement of the judges.
From cocktail party to "bad calories"
Diamandis framed the question as a thought experiment. The judges had just graduated, or a bad bet had wiped them out. They had no network and no wealth. What would they do today?
Pincus started with a caveat: the answer might sound "really obvious," just what people would expect from him. He then explained why he would take on social media. He finds today's experience "so toxic," he said, calling it "such bad calories" and "so high in noise to signal." In other words, there is a lot of clutter and little worth having.
His view comes from someone who was there early: his background includes the social-networking venture Tribe.net as well as games. "As one of the early people in social," Pincus said, he remembers it starting "as this magic, kind of magical cocktail party." It has since become something he thinks "nobody feels good about." Nobody is proud of how long they just spent on Instagram or similar apps, he said.
The trouble he describes is with what these products are built to reward. Mainstream feeds are ranked by recommendation systems, software that predicts what a user is likely to engage with. X, for example, says its "For You" timeline learns from interactions such as likes, replies and reposts, and combines posts from inside and outside a user's network, according to X's own documentation. Pincus would flip the goal, so that the product counts less time spent as a win. Agents here means AI software that can carry out tasks for a user. He gave this as his answer to a hypothetical. He did not describe a product, a company or how the agents would work.
Luckey: right on average, but not about everything
Diamandis noticed Luckey nodding and asked for his view. "I think that you're right on average," Luckey said. Then he qualified it.
Luckey said he has "managed to curate a pretty good feed." He also pointed to the many online communities built around "very, very niche things." In his view, some of these communities only work because they sit on the major social platforms. He liked the idea of keeping people out of the app, he said, but only when it comes to "the unintentional" side: "endless scrolling, doom scrolling."
To draw the line, he went back to his own history. Luckey grew up on internet discussion forums and watched many of those communities move to the big social platforms. Time spent there is different, he argued: "that's intentional time." His example was his own hobby. "I want to see what people are doing with software defined radios today," he said, and noted that he belongs to a Southern California group of software-defined radio enthusiasts. These are radios where much of the signal handling is done by software instead of fixed hardware. "That stuff is good," Luckey said.
He reused Pincus's food metaphor with a twist. Pincus called social media "bad calories." Luckey called the useful communities "the high calorie stuff" and said the goal should be to "shift from the crappy stuff" toward it. So the two did not disagree about whether social media has value. They differed on what an app built to reduce time should protect. Luckey's answer is the time people choose to spend, as opposed to the scrolling that simply happens to them.
The exchange also put a rough number on the problem. X was described as a social network that delivers "tons of value," but, honestly assessed, that value came from "probably like 10 % of the time I'm on X."
What existing feed controls do, and don't do
The big platforms already offer ways to see less of the feed their algorithms choose. In 2022, Instagram introduced a chronological Following feed and a Favorites feed limited to up to 50 chosen accounts. It described these as tools to help people use the app more intentionally. In the same announcement, it said it expected to add more recommendations to the main feed. X also offers a Following timeline that shows posts from followed accounts, newest first.
These controls roughly match Luckey's idea of curating your feed: users choose whom they hear from. As the platforms describe them, they govern which posts appear and in what order. Pincus's proposal, as we read it, aims at a different lever: the goal the product itself is built around. For now it is a proposal: the company he said he would start from zero, one that would count time saved as a win. Luckey's condition was that it keep the "high calorie stuff," like the software-radio group, while cutting out the doom-scrolling.