The entropy of things increases over time provided their size remains constant. This says nothing about the complexity of phenotypes, but it does suggest that it is unlikely for the same phenotype to be reproduced by a "devolving" process in which all the physical things on which the phenotype are based revert.

Because life isn't a closed system (there is always energy in and out) the second law applies with a caveat that the flow energy can be used to maintain the state of the system. However, unless there is infinite energy available this won't be a perfect process and things will tend to change.

I think we are talking past each other because I am focusing on genomic evolution and you are thinking about it from a phenotypic perspective. If we ignore that genomes determine phenotypes, and look at evolution as a progression through various phenotypic states then it is more comfortable to say that there is no direction to the process. To me, evolution is a physical process, but I think you have a different perspective.

I also don't think you're wrong or I'm right--- my focus here was just to describe why "devolve" is not a technically rigorous concept!