Gossipping has demonstrate to be an efficient mechanism for spreading information among P2P networks. Within the context of P2P computing, we propose the so-called Evolvable Agent Model for distributed population-based algorithms which uses gossipping as communication policy, and represents every individual as a self-scheduled single thread. The model avoids obsolete nodes in the population by defining a self-adaptive refresh rate which depends on the latency and bandwidth of the network. Such a mechanism balances the migration rate to the congestion of the links pursuing global population coherence. We perform an experimental evaluation of this model on a real parallel system and observe how solution quality and algorithm speed scale with the number of processors with this seamless approach.