Unexpectedly Useful: Convergence Bounds And Real-World Distributed Learning