Baran described a distributed network with about 3-4 links per node as a useful level of redundancy. That design insight does not by itself guarantee survival of any particular fraction of node failures. In a separate simplified model with n independent paths, the probability of at least one path delivering a message is:
P(delivery)=1−i=1∏n(1−pi)
With 3 independent paths each having 90% reliability: P=1−(0.1)3=99.9%. Shared failures or correlated paths would lower the real-world result.
Interactive Network Reliability Calculator: Explore how redundant paths improve reliability:
html`<div style="background: var(--bs-light, #f8f9fa); padding: 1rem; border-radius: 8px; border-left: 4px solid #2C3E50; margin-top: 0.5rem;"><p><strong>Network Delivery Probability:</strong> ${(delivery_probability *100).toFixed(2)}%</p><p><strong>Single Path Reliability:</strong> ${(path_reliability *100).toFixed(1)}%</p><p><strong>Improvement over Single Path:</strong> ${improvement_over_single.toFixed(1)}% better</p><p style="margin-top: 0.5rem; font-size: 0.9rem; color: #7F8C8D;">Formula: P(delivery) = 1 - (1 - p)^n, where p is individual path reliability and n is number of paths</p></div>`