Die Zombie-Simulation, die niemand erklären kann

Scientific concepts / nature phenomena presented

1) Cellular automaton / grid-based spread models

• The world is modeled as a 2D grid (like graph paper), where each cell is either:
  • tree (susceptible) or
  • bare earth (empty)
• Spread rule (forest fire / infection analogue):
  • Fire/pathogen spreads only to directly neighboring cells (up/down/left/right).
  • This corresponds to having no diagonal contact in the basic model.
• The grid is randomly generated by assigning trees/people with a given probability (e.g., 50%).

2) Percolation theory and the “tipping point”

As the density (probability of occupied/susceptible cells) increases:

• Connected clusters form and eventually connect into a system-spanning continuous path.
• The system shows a sharp transition between:
  • Subcritical regime: spread stops (fire/disease fails to reach far)
  • Supercritical regime: spread becomes effectively unbounded across the grid
• Key concept:
  • Percolation = spread that forms a connected path across the entire playing field.
  • A percolation threshold is identified at about 59.27% for the specific 2D nearest-neighbor grid setup described.

3) “Universal” threshold value claimed for the basic 2D model (59.27%)

The video claims the threshold appears reproducibly at ~59.27% in simulations of:

• forest fire across randomly placed trees
• fluid flow through a coffee-powder cavity network analogy

It emphasizes:

• No elegant exact mathematical formula is known to derive this specific number.
• The value is known approximately via large-scale computation (supercomputer brute-force simulation).

4) Coffee filtration analogy (connected cavity networks)

• A compressed coffee bed is treated as a random network of pores/cavities.
• Water can percolate only if there exists a continuous connected route from top to bottom.
• As the “cavity probability” (or looseness/density) changes:
  • water first penetrates shallowly but stops
  • then abruptly transitions to continuous flow (percolation threshold)
• The video asserts the threshold matches the same ~59.27% value as the forest fire model.

5) Zombie/infection metaphor mapped to percolation

• Zombies are modeled like an infection that spreads to adjacent occupied cells:
  • first zombie infects neighbors standing directly next to it
• A sharp density threshold separates outcomes:
  • Below threshold (e.g., ~58–59%): infection dies out due to missing continuous paths (“last saving gaps”)
  • Above threshold (e.g., ~60%): a single infection triggers a connected path and spreads across the whole grid
• The video also claims the threshold becomes sharper for larger grids (random outliers become rarer).

6) Cluster formation and why transitions are abrupt

• “Clusters” form as occupied cells connect.
• With increasing density:
  • small clusters merge
  • eventually a single additional cell can create a bridge connecting large regions
• This yields a nonlinear tipping point rather than a gradual increase in total spread.

7) Model variations change the threshold

The video describes how altering rules changes the percolation threshold:

• Allow diagonal spread (8-neighbor connectivity instead of 4-neighbor):
  • barriers are easier to bypass
  • threshold drops to about 40.7% (as claimed)
• Move to 3D grid:
  • infection spreads in more directions
  • threshold drops to about 31.2% (as claimed)
• Main takeaway: thresholds depend on geometry and connectivity rules, though the existence of a threshold mechanism remains robust.

8) More realistic simulation (moving zombies + defense)

A referenced Cornell study simulates:

• zombies moving spatially
• humans defending themselves
• real USA population geography/data as the map/grid

Reported outcome:

• spread correlates strongly with population density
• cities are overrun quickly, while some rural regions remain zombie-free longer.

9) Percolation in financial networks (systemic risk contagion)

• Banks are nodes; lending/connection structure forms a network.
• If a bank fails:
  • it cannot repay debts → losses propagate to connected banks
• Contagion condition:
  • losses exceeding equity buffers cause additional bank failures
• The video connects this to the 2008 global financial crisis, framing contagion as percolation through the network.
• Response described:
  • post-2008 regulations requiring higher buffers and more network-risk control.

10) Implication: critical thresholds can prevent (some) global collapse

The video frames percolation as a lesson for real systems:

• many systems look stable until a threshold is crossed
• resilience can be improved by:
  • immunizing only to a critical fraction (not necessarily 100%)
  • adding safety buffers to economic systems to prevent network-wide contagion

---

Methods / modeling steps outlined (as presented)

1. Construct a grid world
   • Represent the area as a 2D (or modified) lattice.
   • Randomly assign each cell as occupied/susceptible (tree/people) with probability (p).
2. Define the adjacency rule
   • Basic case: spread only to up/down/left/right neighbors.
   • Variant: include diagonals (more neighbor connections).
3. Initialize an event
   • Choose one occupied cell as the “source” (lightning ignition / first zombie).
4. Iterate spread
   • Let “infection/fire” jump from an occupied neighbor to its allowed adjacent occupied cells.
5. Test outcomes across densities
   • Repeat simulations for different values of (p).
   • Determine whether spread reaches the edges (system-spanning path) or dies out early.
6. Identify the percolation threshold
   • Find the critical (p) where a connected path across the grid typically appears.
   • Report the claimed thresholds for each setup (about 59.27% for basic 2D; lower for diagonal/3D).

---

Researchers / sources featured (as stated in the subtitles)

• Researchers at Cornell University
  • simulating spatial zombie scenarios with defense and real USA population data
• Emma Laboratory / Emma
  • sponsorship mentioned; not claimed as a scientific contributor to the percolation research (as stated in the video)

---