Home ›News ›Meta's reported Takes trial explores forecasting without cash

Meta's reported Takes trial explores forecasting without cash

Meta employees are testing a prediction app called Takes using virtual points rather than real money, Reuters reported on 13 November, citing The Wall Street Journal. The points are called cred. The report describes an internal product without a confirmed public release timetable.

In this article
Illustrative photograph of two fictional adult friends discussing approaching rain on an open city terrace; one holds an unbranded phone with its back visible. Not Meta employees, a Takes demonstration or a real product-testing event.
Illustrative editorial image.

For the gambling and prediction-market industries, the interesting question is what a company might gain from forecasting as a social activity. An experiment can explore whether people enjoy making predictions and returning to their outcomes before a business decides how, or whether, to build a wider service.

That is a different proposition from demonstrating demand for cash trading. The available reporting supports a limited product story, leaving its eventual audience, operating model and commercial prospects open.

Facebook has explored social forecasting before

There is historical context within the company. In a 20 October 2020 post on LessWrong, Rebecca Kossnick identified herself as a member of the team building Forecast within Facebook's NPE labs group.

She described a community where people used virtual points to trade on future outcomes, with moderated questions, a leaderboard and incentives to explain their reasoning. Her account presented discussion as an important part of the experience alongside the predictions themselves.

That earlier experiment shows why a technology company might be interested in the format beyond financial transactions. Asking someone to make a forecast can give a conversation a defined question, an eventual outcome and a record that can be revisited.

It does not establish that Takes inherits Forecast's technology, team or commercial objectives. The two products should be treated as separate episodes unless further evidence connects them.

The points system shapes the evidence

A points system gives people a way to express confidence and compare results without staking cash. Competition, curiosity and reputation can all provide reasons to participate.

Those motivations also affect what a test reveals. A participant may be willing to make a playful forecast while having little interest in putting money at risk. Another may value the discussion more than the score. Treating both behaviours as evidence of demand for a financial product would miss that difference.

Any scoring system also needs rules that participants understand. If repeated activity earns an advantage, a high score might partly reflect persistence. If a few lucky outcomes dominate, it may reveal less about judgement than the ranking suggests. These are general design questions, rather than findings about the unpublished workings of Takes.

A useful evaluation would ask what the product rewards and whether those rewards support its purpose. Entertainment and forecasting accuracy can each be legitimate goals, but they require different evidence.

An employee trial has a narrow audience

Internal testing can help identify confusing screens, ambiguous questions and problems with recording or resolving outcomes. It can also reveal whether the basic experience is enjoyable enough for people to return.

The audience, however, is selected by employment rather than by a representative sample of potential users. Staff may share interests, familiarity with technology or reasons to try a colleague's project. Those characteristics can influence participation.

A public product would face people with different expectations and much weaker reasons to tolerate inconvenience. It would also need to sustain interest after the novelty of trying a new app had passed.

Our reading is that an internal trial can produce useful evidence about a concept while leaving its market size unresolved. Usage becomes more informative when its circumstances are clear: who participated, what encouraged them to join and how their behaviour changed over time.

Three evaluation questions: is the concept clear and engaging; would a wider audience choose it beyond novelty; and would its benefit justify development, moderation and support costs? An employee trial does not establish public adoption or a sustainable business. Evaluation framework, not disclosed Takes results.
Evaluation framework, not disclosed Takes test results.

Meta's scale changes the business context

Meta's third-quarter results reported an average of 3.64 billion Family daily active people in September 2026. That is a measure across its established Family of Apps, rather than an audience for Takes.

The distinction matters when assessing the opportunity. A large existing business may have resources and distribution options that a start-up lacks. A separate product still needs a reason for people to choose it and return.

Meta's advertising-supported business also provides one possible economic context. A free forecasting experience could, in principle, create value through engagement or improvements to the wider service without collecting a fee on every prediction. That is an analytical possibility, rather than a monetisation plan established by the reporting.

Any integration with Facebook, Instagram or another service would require its own announcement and evidence. The parent's user base should not be converted into a projected adoption figure through an assumed percentage.

The questions and their resolution are part of the product

For any forecasting community, a prediction needs a clear subject, deadline and method of deciding the outcome. Otherwise, participants can agree to take part while understanding different questions.

Resolution becomes particularly important when events are disputed, postponed or reported differently by competing sources. A service needs a consistent way to handle those situations and explain its decisions.

The same applies to the conversation around a forecast. Encouraging people to give reasons can make a discussion more useful, but a popular explanation is not automatically an accurate one. Moderation and evaluation therefore remain part of the operating cost even where no cash changes hands.

The earlier Forecast account is relevant as a record of how one team approached these questions. Its description of incentives and discussion provides historical context, without demonstrating the performance of a product being tested six years later.

Cash markets raise different operating questions

Our coverage of Kalshi's state-access disputes concerns event-contract businesses and their legal operating framework. That background helps explain why a report about prediction software should be read carefully before being treated as news of another financial-market entrant.

The distinction is commercial as well as legal. Cash trading involves money moving between participants and a business model for supporting that activity. Virtual-point participation measures something else. The two cannot be compared by treating points as pounds of trading volume.

Similarly, our examination of Polymarket's organisational changes considered the resources needed to support a developing business. Visibility and product interest are only part of that operating picture.

Any future change involving payments, redeemable rewards or a commercial launch would need to be assessed on the terms actually announced.

What would make the experiment significant?

The next useful evidence would concern the product's purpose and results. A small test could be valuable if it answers a specific question about how people engage with forecasts, even if it never becomes a standalone service.

AreaQuestion for evaluation
ParticipationDo people return after the initial novelty fades?
Forecasting qualityAre questions clear and outcomes evaluated consistently?
CommunityDoes discussion improve understanding and remain manageable?
EconomicsDoes the benefit justify development, moderation and support costs?

Our judgement is that the appeal of a points-based prediction product lies in making opinions testable and participation social. Whether it can do that well depends on its incentives, question design and community, rather than the familiarity of the company behind it.

For now, Takes is a reported internal experiment. Public release, meaningful adoption and a sustainable business would each be separate developments requiring their own evidence.

Topics and categories

Countries
United States

Background

Related reporting for comparison.

All coverage: Meta Takes prediction app