Jump to content

Future Audiences/World of Wikipedia

From Meta, a Wikimedia project coordination wiki

World of Wikipedia, the experimental card game created by the Future Audiences team, sprung from a question: since Wikipedia belongs to everyone, wouldn’t it be fun if everyone could hold a little part of it in their hands? We wanted to build an experience that lets users do exactly that. World of Wikipedia gives players a way to build personal, ownable connections to our collaborative encyclopedia. Our hope is that by playing and sharpening your intellectual instincts, you can earn your own little piece of the internet, one card at a time.

Overview & Key Facts

[edit]

What is the World of Wikipedia game?

[edit]

World of Wikipedia is a card game made by the Future Audiences team in which players sort between cards made from real Wikipedia articles and cards made from made-up articles. Players earn cards by correctly identifying real cards. Each card is unique and one-of-a-kind, so owning one means owning the only copy of it in the world.

Why did we decide to build a card game?

[edit]

Card games have become more popular recently. The global trading card games market was estimated at $7.4 billion in 2025 and is projected to grow to $16.9 billion by 2035 at a compound annual growth rate of 6.9%. Digital platforms are pulling in new players, especially through mobile. This shift is what influenced our decision to build a casual, mobile-first experience for collecting articles.

Here’s how we got here:

  • In February 2026, a Japanese developer created WikiGacha, a project that lets users collect cards based on Wikipedia content.
  • We posted about WikiGacha on our TikTok account and saw a strong response from the community, which led us to reach out to WikiGacha’s creator to learn more about his thinking and development philosophy.
  • We'd also been thinking about the rise of misinformation and whether a game built around distinguishing real from made-up content could double as a lightweight way to develop that skill. Our thought was that we could teach media literacy by providing opportunities to practice it.
  • We made a few prototypes and shared them with WikiGacha’s creator before iterating based on his feedback.

We also wanted to introduce a new and engaging way for people to learn about what goes into an article. Articles are carefully crafted by over 250,000 active volunteer editors who write, cite, revise, and patrol articles as volunteers who believe in free knowledge as a public good. We wanted World of Wikipedia to showcase some of the reasons readers can trust these articles, such as editor and citation count. These are the same kinds of figures we include in our attribution framework, which helps reusers of Wikipedia credit our content properly. Good attribution helps increase trust in content by giving readers a way to trace what they're seeing back to our volunteers and sources behind them.

Hypothesis & Success Criteria

[edit]

Our research and related thinking led us to this hypothesis: “If we implement a proof-of-concept card game to extend our presence beyond Wikimedia sites and reach 10k card pulls per day, we'll learn whether this is an effective way to generate new engagement.”

Our project will be considered successful if:

  1. We can release a card game that lets players collect unique Wikipedia articles,
  2. Players pull at least 10k card pulls/day in at least 1 week in the 30 days post-launch,
  3. We are able to receive qualitatively positive media attention,
  4. This card game receives qualitatively positive feedback from players, and
  5. We are able to use this game as an example of attribution in our attribution framework.

Mechanics & Methodology

[edit]

What does our game do?

[edit]

In World of Wikipedia, players test their knowledge and fact finding skills across categories such as sports, science, history, and video games. Through play, players earn unique cards that are created from real Wikipedia articles. Each round presents players with a stack of 10 cards. Of these 10 cards, 5 are built from real Wikipedia articles and 5 are built from invented ones. Players sort the stack to correctly identify which cards are real.

Players have 3 lives per round, meaning they can guess wrong twice before a third incorrect guess ends the round and their ability to keep collecting cards for that session. Players who complete a round by making it through all 10 cards without guessing wrong 3 times can log in to claim real cards.

At the end of each successful round, players are presented with results, which include a breakdown of star ratings and another one of correctly answered vs incorrectly answered cards. Claimed cards live in a digital binder on player profiles, which come with shareable links for card collection showcasing. The binder supports a few customization options (username, description, binder color) and automatically sorts cards by category. Players can also share individual cards from their binder or their round results.

How do World of Wikipedia cards work?

[edit]

Our cards are made from Wikimedia content:

  • Each card's image comes from the article’s main image. We used related images from Wikimedia Commons when needed.
  • We took care to select only images with appropriate licenses and recorded which license each image carries at the bottom of each card.
  • Each card’s description text is drawn from the article's first sentence. Sentences are occasionally trimmed for brevity and readability.

Players can flip over cards in their results or profile binders. Doing so reveals a set of stats and trust signals about the underlying article, such as:

  1. Views over the last month,
  2. Number of contributors,
  3. Number of revisions,
  4. Number of sources cited, and
  5. Last edit date.

These trust signals are meant to give players a sense of how much work and dedication contributors put into each article so that readers can trust what they’re reading. They are also tracked in our attribution framework, which guides reusers on how to cite Wikipedia content. Since users trust Wikipedia content, reusers who surface similar signals when attributing content stand to benefit from increased user trust.

We gave each card a star grade out of 5 based on the card’s traffic. The probability of drawing each card is completely random, and is not based on the card’s star level. The star tiers are:

  • 5 stars: Top 10% of traffic
  • 4 stars: 11%–30%
  • 3: stars: 31%–60%
  • 2 stars: 61%–80%
  • 1 stars: 81%–100%

How did we build the World of Wikipedia game?

[edit]

We began by gathering the top 90k articles by recent traffic. From there, we set out to ensure that content was appropriate and SFW. Defining that filter was difficult. We erred on the side of caution and aimed to over-filter rather than risk leaving inappropriate content in.

  • We aimed to scan article text for a systematic set of keywords. However, terms that seem like clear red flags in isolation often show up in acceptable contexts.
  • For example, completely excluding articles mentioning any violence removes most country articles, and a word like "execute" carries a very different meaning depending on an article’s context.

Given the scale, we imagined that some articles might slip past multiple rounds of automated review, so we undertook manual spot checks. Our team went through portions of the card list by hand to catch and remove anything that shouldn't be there. If you come across a card that seems inappropriate, we'd appreciate you flagging it through this form, which helps us keep refining the filter as we go.

To divide cards into categories, we used the Wikimedia Foundation's Machine Learning team's ORES article topic model to determine each article's topic. We then mapped those topics to a set of categories spanning sports, media, history, TV, and more. This was to respond to feedback from our multiple rounds of user testing that topic selection was especially important.

How did we attribute and credit content?

[edit]

We worked with our internal attribution team to determine which fields of data were best for proper attribution. This resulted in capturing:

  • The URL, license, image credit, and attribution details for every card's image.
  • As mentioned above, on the back of each real card is the article's viewcount for the last month, along with its contributor count, revision count, source count, and the date it was last edited.

Release & Results

[edit]

How are we sharing this game and what do we want to learn from this?

[edit]

Since we’ve conducted several rounds of internal testing and user research, including interviews and surveys, we want to bring World of Wikipedia to the public and see how people respond. To distribute our prototype, we're working with our short video team to post a walkthrough of World of Wikipedia across social channels (TikTok, Instagram, and YouTube). We’re also working with our PR team to explore other ways to get our prototype in front of more people.

Quantitatively, we want to learn if our game is interesting enough such that 10k articles are “pulled” each day. Qualitatively, we want to know:

  • Do players feel more connected to Wikipedia after playing?
  • Does owning cards make players feel like they own a piece of Wikipedia?

The goal, above all, is feedback. We want to see how people actually respond to the game. This part of the project is still in progress, and this MetaWiki will be updated with results.

Get Involved

[edit]

We are shifting towards holding calls on Discord to discuss our projects with the Wikipedia community. Please join the Wikipedia Discord server here to stay updated.

If you'd like to provide your thoughts on this project, please feel free to fill out our feedback form. Every response helps us build better, more useful products.

If you have any further questions/inputs please get in touch on the talk page!