The Traveling Salesman Problem (TSP) is a important optimization problem in computer science, mathematics and logistics. It belongs to the class of NP-Hard problems and can be very time consuming to find solutions to large instances with guarantee optimality. As number of city-nodes in the graph increases, the amount of valid route tours also growths rapidly and thus requiring considerable time to evaluate and classify each permutation. The objective of the heuristic process is to search the solution space for the optimal solution while maximizing the attached utility-cost function (i.e. finding the shortest euclidean distance tour) and minimizing the computational time complexity of the algorithm.Many complex real world scenarios can be reduced to a simulation of a salesman trying to find the shortest (length) Hamiltonian (cycle) route in a euclidean super-graph G*. If each city-node is modeled as a input symbol in a communication channel represented by an output pair with consistent probabilities distribution thus an polynomial-time probabilistic algorithm can use this information to improve the solution quality at the same rate of transmission of information over the channel.In this paper we explore an quantitative stochastic process based in Algorithm Information Theory and the Shannon-Kelly criterion to find valid near optimal solutions using a new growth- optimal strategy applied to the TSP problem that have statistically significant transmission rate even when no encoding scheme is available, regardless of time-complexity of the problem.Previous heuristics such as 2 opt, Genetic Algorithms (GA) and Simulated Annealing (SA) approach’s the TSP problem by relying on a priori knowledge about the data distribution in order to reduce the probability of error in finding the best candidate solution tour.In this work we propose a method that models the solution space boundaries of the TSP problem as a communication channel by means of Information Theory. We describe a search algorithm that check for patterns (i.e information content) in the elements of a constrained solution space modeled as messages transmitted through communication systems. The boundaries of the search space are defined by the Kolmogorov complexity of the candidate solutions sequences. We conclude with an discussion about the quality of the results and implications for general decision problem in Turing machines.