Well first, they are not genes, since they do not code for proteins. Secondly, although there may be no pressure to be deleted, there should certainly be no pressure to correct any mutations occurring on them. They are pristine. No changes. And the analogy is "good" in that programs don't mutate on their own. Darwinian evolution requires mutation.
> They are pristine. No changes.
I do not see that in the article.
> And the analogy is "good" in that programs don't mutate on their own.
Some do. Those meant to emulate the genetic process mutate quite nicely on their own.
And mutation most certainly does occur. But you are drawing conclusions from a tiny fraction of a percentage of "neutral" code which has not been affected by replication errors.
In the absense of a testable hypothesis, you cannot draw conclusions.
In order to make even a wild guess about the probability of this being adventitious, you would need to draw up a table of segment lengths of conserved code and see if the lengths can be placed in a normal distribution.