How Zeno Works
Zeno operates at the intersection of cryptography and distributed computing.
The Challenge
When a client initiates a protected action, the Zeno Server issues a Space-Time challenge.
- Space (Memory): Find a 42-cycle in a random bipartite graph determined by a seed.
- Time (Latency): Compute a sequential VDF (Verifiable Delay Function) on the result.
The Solver (Client)
The client (browser) uses WebAssembly to solve this challenge efficiently.
- Web Worker: The solver runs in a background thread to prevent UI blocking.
- Allocation: It allocates ~32MB of RAM (at default GB=18) to generate the Cuckatoo Graph.
- Graph Search: It identifies the required 42-edge cycle.
- VDF Squaring: It performs sequential squarings in a Class Group to generate the time-proof.
- Submission: The cycle indices and VDF proof are sent to the server.
The Verification (Server)
Verification is instantaneous ($O(1)$ relative to the work).
- Cycle Check: The server verifies the 42 edges form a loop (~42 hashes).
- VDF Check: The server checks the Wesolowski proof equation (2 modular exponentiations).
Why This Works
Asymmetry
| Operation | Time | RAM |
| :— | :— | :— |
| Solve | ~0.9s | ~32MB |
| Verify | ~1ms | ~1KB |
Creating a solution costs 1000× more than verifying it.
Memory Wall (Primary Defense)
| Attacker |
Hardware |
Max Parallel (32MB each) |
| Bot Farm |
8GB VPS |
250 instances |
| GPU Farm |
24GB RTX 4090 |
750 instances |
| ASIC |
Custom silicon |
Cannot add RAM |
To send 1,000 requests/second, an attacker needs 32GB of RAM dedicated to Zeno alone.
Time Wall (Secondary Defense)
The VDF is inherently sequential:
- 1 CPU core at 3GHz ≈ same speed everywhere
- A GPU with 10,000 cores cannot parallelize this
- Prevents pre-computation (VDF depends on challenge discriminant)
Scalability
The server is stateless during solving:
- Only validates the final result
- Does not track solving progress
- Can handle unlimited concurrent solvers
Attack Economics
| Attack Type |
vs SHA-256 PoW |
vs Zeno |
| ASIC purchase |
1,000,000× speedup |
No speedup (memory-bound) |
| GPU rental |
1,000× speedup |
Limited to VRAM ÷ 32MB |
| VPS farm |
Scales linearly |
32GB RAM per 1000 req/s |
The “Quantum Zeno Effect”
The system is named after the physics phenomenon where continuous observation freezes a quantum state.
Similarly, Zeno “freezes” bot throughput by forcing continuous resource expenditure:
- Each request costs real resources (RAM + time)
- Cannot batch or pre-compute
- Cannot parallelize beyond memory limits
- The attack is effectively frozen by physics