That's an utter non-sequitor. It has nothing to do with either our fossil evidence or with various models of cars buried in junkyards. Further, I'm not saying that natural selection is impossible. Your argument acts as if I did, though, and that's one of many reasons why it is a non-sequitor (i.e., has nothing to do with the debate at hand).
What I am saying is that natural selection can't produce out of chaos the level of order required for either useful DNA or useful computer programs. Once DNA and computer programs are here, natural selection can concievably reduce the number of computer programs or DNA life forms, but that serves no use toward creating distinct new, improved, more complex computer programs or DNA life forms (sans intelligent intervention, anyway), for that we've only seen evidence of designers giving us those new creations.
The genome of every living thing contains a record of every change that has ever happened to the species since life first formed on this planet. That is why more than 95% of every genome is just junk. The only exception is for deletions which are discarded. Sequencing the junk permits a kind of mapping that sequencing of functional genes does not permit because the junk is not constrained to function by natural selection and so can mutate without effect. This serves to provide a molecular clock that can be calibrated against the fossil record and radioisotope dating to confirm dates of separation for related (all life on the planet is related phylogenetically) species. Intelligent Design hypotheses are strangely silent on the question of this genetic detritus of the evolutionary history of the species.
You are ignoring the fact that sharing of genetic material among all species is possible through the agency of viral transduction. Bacteria share genes through plasmid transfer between species. Again, you are ignoring the fact that the experiment of life is conducted on such an extraordinarily massive scale that there is more than enough opportunity for the opportunistic origination of novel useful sequences by entirely random processes.