That's an unfriendly comment

I don't think you understand the second law. Boltzmann already showed in 1877 that entropy is statistical in nature, which means that it theoretically could spontaneously decrease, it's just extremely unlikely.

Your statement "The total entropy change of the system plus the surroundings (the universe) can never decrease" is false

The second law of thermodynamics is also statistical in nature, so the best way to describe it is to say that entropy is unimaginable extremely likely to never decrease regardless if the system is open or closed.

It could still polarize

I made a mistake, I can't edit it anymore, but it should be $ \delta L = L_1 $. We have a product of 2 infinitesimals, so that is an infinitesimal to order 2; you can neglect these in the first approximation. The first approximation only takes into account infinitesimals to order 1.

In a Taylor expansion, $\delta \mathbf{L} $ is the first-order term representing the linear response of the angular momentum $ \mathbf{L} $ to a small perturbation. The Taylor series is $ \mathbf{L}(\epsilon) \approx \mathbf{L}(0) + \epsilon \frac{d\mathbf{L}}{d\epsilon} \Big|_{\epsilon=0} + \frac{\epsilon^2}{2!} \frac{d^2\mathbf{L}}{d\epsilon^2} \Big|_{\epsilon=0} + \dots $. Here, $ \delta \mathbf{L} = \frac{d\mathbf{L}}{d\epsilon} \Big|_{\epsilon=0} = \epsilon \mathbf{L}_1 $ is the first-order change in angular momentum. Does that make sense?