nonsense back to you.
genetic mutation comes in several forms, some are more common than others and sometimes the sequence being mutated can affect the rate of its own mutation (and even this can happen either biochemically or by following genetic instructions that affect mutation rate).
1. Point mutations.
Clearly you've heard of this--this is what you are colloquially referring to as "mutation". This is gain, loss or replacement of a single base. Due to the degenerate nature of the triplet code, most replacement point mutations that occur within a gene are effectively silent, not causing any change in the resulting protein. Of course, many more happen outside the coding regions of genes and typically do nothing. When the point mutation causes a change to the amino acid sequence, the protein still might not be affected. If the amino acid substitution of the same type (e.g., both hydrophobic or both acidic) the protein function might be unaltered. Or it might be impreceptibly altered. This creates very important variety in proteins that allows an individual in changing circumstances to adapt. Of course, sometimes, and maybe most often, the protein performs worse. If it's bad enough, that mutation is removed from the gene pool along with the individual possessing it.
Insertion and deletion of a single base is very similar to the next category, so I'll discuss them there.
2. Insertion and deletion of 1 or more bases.
This is a real problem for genes. If the insertion or deletion is a multiple of 3, then there's a chance nothing will happen but 1 amino acid will be gained or lost, but potentially not affecting the function of the protein. But there's also a good chance the protein will be severely altered, because from the mutation onward, the frame of 3-bases per amino acid will be off, and now you'll end up with a completely different string of amino acids. Chances are they'll do nothing, but sometimes they do something. If the protein is lost completely, then there may or may not be a problem to the organism. You've got two copies of most genes, and the second copy might be able to compensate for the bad protein. If so, this becomes what people call a recessive mutation, meaning if you've only got 1 "bad" allele, you're all right, but if both alleles are "bad" you suffer. Sometimes the truncated protein is worse than missing, though. Sometimes the shortened form works well enough to go through all the motions of being a protein (going where needed, binding to its partners, etc.) but then fails to carry out its job and at the same time interferes with the "normal" copy. This becomes what we call a dominant mutation and it's really bad news. Insertions or deletions are fairly common in certain types of sequences but for the most part, the quality control machinery can catch them and fix or destroy the aberrant cell.
3. Translocation and duplication events.
This is a specialized type of insertion, in which a whole section of DNA is either extracted and inserted elsewhere, or copied and the copy is inserted elsewhere. Extraction-insertion is not necessarily a problem, since you've got a net gain of 0 new sequence. The gene is still present and probably still functions normally, it just lives in a different location now. Sometimes this is a problem, if it is translocated without its regulatory machinery, so that now it doesn't activate at the right time or make the proper dose of protein. Sometimes the regulation is only slightly tweaked, again allowing for a slight variation of the protein function in an organism, which can be useful when the population is facing novel or changing conditions. Sometimes this causes cancer.
With duplication, evolution is most free to act in a good way. This is the wholesale copying and inserting of the copy in a new place. Often the new copy is put next to the original, but not always. With two fully functioning copies of the same gene, a rendundancy is built in, superior to the one created by having two alleles (one on each chromosome). Later on down the line, those duplicated genes will start to acquire mutations. If one gets mutated into a bizarre new form that behaves nothing like the original, there's still a copy able to take over the protein's regular function, allowing the copy to become fine-tuned to whatever new role happens to work for it. This is key when it comes to allowing one essential protein to evolve into another relate protein without losing the essential protein function.
Of course, if only part of a gene is translocated or duplicated, the gene would be torn in half and probably stop working, or result in a truncated form or something, or the copy could be a copy of only half a gene, possibly resulting in a dominant negative mutation. Alternatively, the insertion could interrupt some other, unrelated gene. This could be bad (destroying the function of that other gene) or just new (creating a fusion of two proteins that now end up sharing the same body. This can be really cool because it might allow them to continue to function properly, but they just both have to go everywhere together. This can be a real benefit if they work in the same pathway and can allow for better efficiency, or it can be a real disaster, as any mutation has the chance to be).
So, I hope you see that mutation is /not/ only about destroying information. There are many ways in which mutation (one part of evolution) can create new information or just change existing information in ways unpredicted. So far, the only argument in favor of irreducible complexity in the evolution of living organisms is our own inability to imagine how it all works. But just because we can't imagine it now has absolutely no bearing on whether or not it is true. And the evidence against irreducible complexity is vast, carried by the thousands of scientific papers that describe how something previously not understood has been worked out and now we know that one more thing, thousands of times over.
