If a gene is deleterious to reproduction it's failing to meet the gold standard. What I mean by a lack of template is there is not some immutable gene that can't remain changed even if for the better, because there is something in nature that is going to intentionally change it back to meet some ideal. That would amount to a mechanism preventing micro-evolution from becoming macro-evolution.
This isn't to say there is not some type of correction system for DNA. If I remember correctly there is: It can check itself against the other strand. Correct me or fill me in on this too, if you will.
I'm no expert but I'll jump in here. Mostly to beat others so I can be corrected myself.
There is a DNA repair mechanism that keeps the rate of mutations in check. I'm not sure you can be correct that it uses the "other strand" because it comes into play when it is replicating a strand. But I don't know.
There isn't anything trying to force DNA back to "some ideal". What does exist is selection. Using the oft bandied about statistic that about half of human conceptions fail we see a mechanism weeding about very bad changes.
There are some basic genes that are highly conserved. We share them with all (or almost all) life forms. These are those that influence fundamental processes like cell division. It appears that most modifications to them means you get a gene that doesn't work. This may generate some of the failed human conceptions for example. This means that the gene is held in a stable form for a long time. It isn't anything magic it is just the constant removal of anything that doesn't work and it appears that there are few variations on some of these genes that can work.
This could be true even if there could be a "better" form of this gene. The gene's environment includes other genes and the chemical processes they end up producing. If a very basic, core gene is changed and it doesn't fit with others then it is removed even if it somehow did it's own specific thing "better". "Better" is always environmentally dependent.