Replication Arena
How the arena works, and what to try
The dish holds loose cells (the small dots) and a few parent bots. Bots swim around and push the cells ahead of them, so cells collect into ridges and piles. A pile that holds enough cells, is roughly round and is left untouched for a few seconds hatches into a new bot, one generation on from the bot that built it. Bots and the chart are colored by generation. The line under the dish counts time, loose cells, bots, the highest generation reached, and the largest pile against the size needed to hatch.
The presets recreate the 2021 experiments and push past them: spherical parents, as in the paper, which rarely build a pile big enough; C-shaped “Pac-Man” parents, which scoop cells into their mouths and go further; a Pac-Man dynasty, where offspring inherit their parent’s shape; a starved dish with too few cells; and a flood with plenty.
The controls. Settings change the number of loose cells and parent bots (both restart the dish), the size of the bots’ mouth, how sticky cells are, how many cells a pile needs to hatch, and the speed. The checkbox decides whether offspring are spheres, as in the lab, or copy their parent’s shape. Click the dish to drop fresh loose cells; Restart and Step are under the settings.
Things to try. Run the spheres preset, then Pac-Man, and compare how many generations each reaches on the chart. In the starved dish, click to drop more cells and see whether replication restarts. Turn the mouth up or down and watch the largest pile change. Then try the dynasty, where shape is inherited, and see how deep the generations go.
Spherical parents push cells into ridges but rarely gather a pile big enough to hatch, so replication stalls early.
Loading…
Starting…
Click the dish to drop fresh loose cells.
Generations
Colored areas: living bots per generation (left axis). Dashed line: loose cells left (right axis).
Settings
Changing loose cells or parent bots restarts the dish. Hatch size is how many cells a pile needs; it must then sit untouched for a few seconds to mature.
The science
In 2021, researchers at Tufts and the University of Vermont found that xenobots swept loose frog stem cells into piles, and some piles developed into new moving xenobots. This is kinematic self-replication: building a new body from surrounding material instead of growing one. Spherical xenobots usually stopped after one round. An AI-designed C-shaped body, like Pac-Man, extended replication to four rounds.
This simulation is a 2D version of that process. Real cells aren’t particles, cilia push through fluid, and a real pile takes days to develop. Here a pile hatches once it holds enough cells, is roughly round, and sits untouched for a few seconds. What the sim can show is how bot shape and the amount of raw material decide whether replication keeps going.
One thing the lab organisms lack is heredity: a pile rounds up into a sphere whatever shape its parent had, so nothing about the parent passes on. The “offspring copy the parent’s shape” setting switches that on, to show what inheritance would add. It is a thought experiment, not a lab result.
What’s real
- Xenobots gathered loose frog stem cells into piles, and some piles became new moving xenobots.
- C-shaped parents kept replication going for more rounds than spherical ones.
- Offspring round up into spheres: nothing about the parent's shape is inherited.
- Replication needs a supply of loose cells, and stops when it runs out.
What’s simplified
- Cells are 2D dots with simple stickiness; real tissue is 3D.
- Bots swim at a constant speed and turn at random; real ones are pushed by cilia through fluid.
- A pile hatches by a rule (size, roundness, left alone for 3 seconds); real piles take days.
- Bots never wear out; real xenobots run on energy stored in their cells and eventually stop.