This paper explores the implementation of ternary (base-3) logic systems using global internet latency as a novel computational substrate. We present a theoretical framework, experimental methodology, and detailed results for a Ternary Internet Computer (TIC) that encodes computational states within the temporal behavior of networked IP addresses. Our findings demonstrate that this approach can reliably simulate basic logic operations and implement a simple perceptron capable of learning the XOR function. Through rigorous statistical analysis across multiple trials, we show that while network latency introduces variability, our redundancy mechanisms yield reproducible computational results. This work contributes to the growing field of unconventional computing by demonstrating how distributed network properties can be harnessed for non-binary computation.