From Federated Learning to $mathbb{X}$-Learning Breaking the Barriers of Decentrality Through Rand