Two New Approaches to Understanding the Second Law

Stephen Wolfram has recently presented a new approach to understanding the 2nd Law of Thermodynamics based on (i) computational irreducibility and (ii) our nature as computationally bounded observers supplemented by the remark: 

• Other observers—or even our own future technology—might see things differently.

The 2nd Law is to Wolfram not a law of physics per se without observers, but something in the mind of human observers with all their limitations, which may not be shared by any form of observer. 

Myself I have over a long period of time been searching a different understanding based on (a) finite precision computation and (b) stability, as developed in detail in The Clock and the Arrow; A Brief Theory of Time and Computational Thermodynamics.

Here (a) connects to Wolfram by seeing the evolution in time of a physical system from one time instant to the next as a form of computational process, where forces at current state are evaluated and so form the dynamics over a small time step creating a new state.  

But (b) has no counterpart in Wolfram's understanding. The essence of stability is that it makes a distinction between a step (or sequence of steps) forward in time and a step (or sequence) backward in time, as an expression of the direction of time expressed in a difference between forward stability and backward stability.  

Here, forward stability means that a step forward in time is possible under finite precision computation (chopping of decimals), while keeping the coherence of the state. The next state thus does not critically depend on the number of decimals kept in the forward computation. The typical case is turbulent flow in e.g. a water fall, where gross features of the water fall are maintained under turbulent fluctuations. 

The essence is here the fluctuating nature of turbulence allowing forward in time computation of mean values under chopping of decimals. This connects to the ageing of a photo with small details being lost while main features remain,  like in Photoshop smoothing of a photo by averaging over local details corresponding to chopping decimals. 

On the other hand, backward in time computation corresponds to unsmoothing, which requires very high precision or simply is impossible. The decimals that have been chopped during forward computation/smoothing are lost and cannot be retrieved and so an original photo cannot be fully recreated/computed from a blurred one. 

Wolfram puts forward the limitations of an observer as an explanation of impossibility of time reversal. 

I seek an explanation in different stability aspects forward and backward in time of inherent physical nature the same for all observers. 

Your choice.