In twonderland, users don’t just share messages—they influence each other’s moods.

Here, you can explore how Alice and Bob affect one another. Each has a sentiment slider, and the white zone around Alice shows the range of sentiments she’s open to. If Bob’s sentiment falls within that range, his message can shift Alice’s mood closer to his.

Try adjusting their sentiments, then press the button to see how Alice reacts to Bob’s message. Will she move closer to his mood?

As the platform designer in twonderland, you decide how users see the world.

Each user receives messages from all their friends—but only the top one makes it through. That’s where your ranking algorithm comes in. Should messages be shown at random? Prioritize similar sentiments? Highlight the most aggressive or the calmest voices?

Click the button to assign random sentiments to Alice’s six friends. Then, choose two ranking algorithms and compare how they sort the messages.

Assume Alice has a sentiment of zero. The connecting lines show how the same friends shift positions depending on your design.

Your choices shape what Alice sees—and how she feels. What kind of world will you build?

Now, let’s see the world through Alice’s eyes.
How does your algorithm choice mess with her mood?

Remember that Alice only pays attention to messages that feel familiar — ones that land inside her white bubble of influence. And she only reads the top-ranked post the algorithm serves her.

Explore how Alice’s model is affected by your algorithm choice. Start with similarity and move Alice’s mood around. Next, see how picking a different algorithm changes her mood shifts.

The simulation breaks this down into steps. Here, you see what happens in a step.

It’s like a never-ending carousel of influence:
Each user reacts, shifts (or not), and then passes the vibe along to others.

But take your time, explore, play. Get familiar with what happens in a single timestep in twonderland.

Welcome to warp mode. 🚀

Everything moves faster. You’ll begin to see how sentiments shift and spread over time. The network has also grown—more users, more connections, more complexity.

Choose a ranking algorithm and start the simulation a few times and watch what happens. You won’t catch every detail, but you’ll start to see patterns emerge.

One Run? Fun.
Many Runs? Science!

Every time you run the simulation, things might turn out a little differently. Why? Because there’s always a bit of randomness sprinkled into the system.

• The order of users? Random.
• The messages they see? Depends on the moment.
• The mood swings? Sometimes subtle, sometimes dramatic!
• Even the network differs slightly.

So don’t trust just one run. Only by repeating the simulation can you spot the real patterns hiding in the noise.

Here, you can run 12 simulations in parallel. Do you see patterns?

Now you know everything to run a scientific experiment and compare two different ranking algorithms starting from the same initial network.

Here’s your challenge:

• Pick two algorithms.
• Run the 12 simulations for each one.
• Compare the results. Spot the differences.
• Experiment! Try maximize or minimize the polarisation.

• Each user has just one trait.
• They read only the top-ranked message.
• The algorithms? Basic by design.

Yet even in this minimal setup, the dynamics are anything but simple.

Now imagine adding more traits, more complex networks, smarter algorithms...
💥 The dynamics could explode in complexity.
That’s why we need formal tools to truly understand these systems.

You’ve seen how even the tiniest design choices can ripple through a digital society. They shape what people see, how they feel, and how they influence each other.
But here’s the twist:
The consequences of these choices are often hidden, complex, and hard to predict. That’s why we need tools like twonderland. Not just to play — but to experiment, observe, and understand. The twon project is an attempt to do exactly this.