Mark Zuckerberg has apparently decided that scrolling through the present is no longer enough. Meta now wants people to predict the future too – ideally through an app built by the same company that has spent two decades turning attention, identity and social interaction into an endlessly optimised product.

The project is reportedly called Arena, a standalone prediction-market app designed to let users wager points on yes-or-no questions about real-world events. At first, there may be no real money involved. But if this sounds like a harmless little game, that is probably the point: gamify the instinct to forecast everything, build a massive audience, then see what becomes possible once people are hooked.

There is an obvious joke here about Zuckerberg appointing himself supreme leader of the universe, standing above the timeline with a giant Meta-branded crystal ball, asking the public to place their bets on what happens next. Ah, yes: the internet was not chaotic enough. It apparently needed a live odds board for reality.

But underneath the absurdity is a serious and distinctly Meta-shaped idea. If the company gets this right, Arena would not just invite people to make predictions. It could turn future events into content, social currency and, eventually, another stream of behavioural data.

Meta has not publicly detailed the product, and it reportedly may never launch. Still, the outlines are revealing. Arena is said to be built separately from Facebook, Instagram, WhatsApp and Messenger, initially using video game-like points rather than cash. Reuters reported that Meta is also exploring potential partnerships with established prediction-market companies Polymarket and Kalshi, while the company allegedly wants to attract 100 million monthly “predictors,” especially among 18-to-34-year-olds.

That is not a small experimental side project. That is a company with more than 3 billion people using at least one of its apps each day, looking at one of the internet’s hottest new behaviours and wondering how to make it social, sticky and global.

Prediction markets are, in theory, simple. A question gets posted: Will a candidate win an election? Will the Federal Reserve change interest rates? Will a film cross a box-office threshold? People buy into “yes” or “no” outcomes, and the market price becomes a rough, constantly changing measure of what participants think will happen.

Supporters describe this as a powerful form of collective intelligence. Rather than relying on one pundit, one pollster or one overconfident person with a podcast microphone, prediction markets pull together the views of many people who have something at stake. If someone is convinced an outcome is likely, they can put points, or money, behind that conviction.

The complication is that the future is not a football score. Politics, wars, public health emergencies, corporate decisions and celebrity scandals are not merely trivia questions waiting to be resolved. When huge platforms turn them into markets, they can change how people behave around the event itself.

That is where Meta’s version could get especially strange.

According to internal documents reviewed by NPR, Arena could use Meta’s Llama AI model to generate questions from trending topics, recommend markets to individual users and resolve those markets in near real time. In other words, Meta’s AI may decide what future people should speculate about, steer them toward particular questions, and then determine whether the thing actually happened.

Take a moment with that.

An AI system trained and deployed by Meta could notice a topic catching fire on Instagram or Facebook, turn it into a prediction market, recommend it to users most likely to engage, and later rule on the outcome. The company would not just host a conversation about reality. It could package reality into a competition loop.

That does not mean the AI would literally predict your next move, despite the deliciously dystopian framing. But Meta already knows a great deal about what captures attention: which posts people linger on, what they share, who they follow, what they watch, what kinds of messages and media keep them returning. Combining that kind of engagement machinery with a system built around forecasting events makes for a potent feedback loop.

The question is not simply whether Meta can ask, “Will this happen?” The question is whether it can learn to ask the question most likely to make you come back tomorrow.

This is familiar territory for Zuckerberg’s company. Facebook began as a social network, but Meta’s larger talent has always been converting human impulses into product mechanics: friendship into a graph, popularity into likes, visibility into reach, insecurity into ad targeting, and short bursts of curiosity into a business model.

Prediction markets fit neatly into that worldview. They transform uncertainty into an engagement format. Every breaking-news event becomes a prompt. Every rumor becomes an opportunity to check the odds. Every political development can become a scoreboard, complete with winners, losers and people who want to prove they knew it first.

Meta has tried a version of this idea before. In 2020, it launched a crowdsourced forecasting app called Forecast, where users could make predictions about events including the Covid-19 pandemic. The company shut it down two years later, with internal documents citing the operational burden of manually curating questions. Arena appears to be a rebuild designed to use AI to remove that human bottleneck.

That explains a lot. The old product had editors and manual work. The new one has generative AI, personalised recommendations and a company that sees AI as the way to scale almost everything. In Meta’s version of the future, the difficult part – deciding what is worth asking, what is true and how to settle a dispute – may become just another automated layer.

It is also important not to confuse the current plan with a full-fledged gambling app. The reporting so far says Arena would use virtual points, which may help Meta avoid the immediate legal and regulatory headaches associated with real-money prediction markets. Gaming lawyer Daniel Wallach told NPR that a points-based launch could give Meta time while the legal status of prediction markets remains unsettled.

That legal uncertainty is not a footnote. Prediction markets have become a contested space, with questions around gambling law, financial regulation, manipulation and insider trading. Reuters noted that the sector has drawn scrutiny over unusually well-timed trades ahead of major policy developments, while NPR reported that the industry is facing more than 30 pending lawsuits over its legality.

And this is where the “it’s just points” argument starts to look a little thin. A non-cash system can still create social status, compulsive participation and reputational stakes. People do not need to win money to become obsessed with being right. Spend five minutes on any social platform during an election, product launch or major sporting event and that much is obvious.

The monetary prize may be absent, at least initially. The dopamine is not.

Arena could also run into a deeper problem: prediction markets are only as trustworthy as the rules governing them. Who writes the question? What counts as an outcome? What happens when the wording is vague, a result is contested, or an event unfolds in a way no binary “yes” or “no” can capture?

Those issues already exist on existing platforms. Handing more of that work to an AI system brings a new set of questions. If Meta’s AI resolves a market incorrectly, who appeals? If the model misreads a developing news story, whose account is affected? If a question is generated from a misleading trend, does the app amplify misinformation before anyone has a chance to correct it?

The product pitch might be collective intelligence. The risk is collective fixation.

There is also an uncomfortably Meta-specific angle: influence. A company that owns major social platforms has unmatched power to direct traffic toward a new product. Reuters reported that Meta may eventually integrate parts of Arena into Facebook and Messenger. Imagine checking a news story in your feed and seeing, right alongside it, a market asking whether the next development will happen by Friday.

That could be fun, in the way a pub quiz is fun. It could also turn every serious event into a speculative side hustle for attention.

A hurricane approaches. What are the odds it makes landfall in a certain area? A company plans layoffs. Will a particular division be cut? A government is considering military action. Will it happen before the month ends? These questions may generate engagement, but they also create an ugly incentive to treat other people’s uncertainty, fear and loss as a game.

Existing prediction markets have already invited concern for exactly this reason. Their defenders say markets can surface information faster than conventional forecasting. Their critics worry that markets can reward insiders, encourage manipulation, and blur the line between observing events and betting on them.

Meta does not need to invent those tensions. It only needs to put them in front of billions of people.

Perhaps that is the most Zuckerbergian part of the story. The company is not trying to become a god in the mystical sense. It is doing something more conventional and, arguably, more consequential: attempting to build the interface through which a vast number of people interpret what is likely to happen next.

First it was the social graph. Then the algorithmic feed. Then short-form video, creators, AI chatbots and AI-generated media. Now comes the possibility of a future feed – a permanent, personalised dashboard of things you are invited to care about before they happen.

The immediate product may be a points game. The larger ambition is harder to miss. If Meta can make forecasting social, addictive and algorithmically tailored, it will have found another way to keep people inside its ecosystem – not just sharing their lives, but constantly speculating on everyone else’s.

The future, apparently, is not something we will experience. It is something we will refresh.