The thing with ERV's (Endogenous retroviruses) being evidence for common ancestry has a lot to do with the occurence of the same ERV's at the same characteristic genomic sites in both Chimpanzees and humans, the null hypothesis being that such ERVs could occur by the separate integration of the same retroviruses in both chimps and humans, but since this is extremely improbable, the only other possibility for explaining the aforementioned statistic is common ancestry, that is how it works.
TalkOrigins have a section on this, and a lot more literature on the topic can be accessed using Pubmed.
Coming to Mr.Skeptic's comment about how the same evidence could be used for common design, papers such as this http://www.ncbi.nlm.nih.gov/pubmed/20345981
illustrate that any use of genes that would specifically cause disease would automatically indicate that the designer, if there was one, would be quite unintelligent, thus leaving evolution as the only explanatory candidate.
Merged post follows:
Consecutive posts mergedComing to the information canard that keeps being trotted out with astounding regularity, it never really gets defined properly, does it?
Also, examples of de novo gene creation (new information) are here at http://www.ploscompbiol.org/article/info:doi%2F10.1371%2Fjournal.pcbi.1000734
and
http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0000096
You may also want to read about copy number variations.