Evolution doesn't have a direction so surely it is possible for it to go backwards?
I had a thought about this. You seem to be thinking of forward, and backward as directions evolution can take. This is, of course, not true. If you want to consider an evolutionary change as having a direction, you'd need to consider it in a (very) multi-dimensional space. For example, if you had DNA coding for just this short section of protein LIAGMP, a change in a single one of those amino acids is effectively a direction evolution can take, so with just six amino acids you have six directions! But, wait, that's not right because there are 19 different possible other amino acids each one could change to.
As you can imagine the number of "directions" rapidly becomes bewilderingly high. So, in fact, "backwards" is not a choice of two options; it's a choice of billions so the chance that a new mutation will directly reverse a prior mutation is
tiny*.
* - unless we're considering bacteria which have small genomes, huge population sizes and tiny generation times meaning the probability of any change occurring is very high indeed.