Here's an example of a man turned loose for a murder conviction after 10 years. He had two eye witness testimonies against him who later admitted to lying, and he's now filing a suit claiming in part that prosecutors suppressed evidence that would have acquitted him. This kind of crap happens every day.
http://www.cnn.com/2014/03/11/justic...guson-lawsuit/
- - - Updated - - -
Uh, yes it does. Take for instance something with a 99.9% accuracy rate. That means for every 1000 people, one of them will be wrongfully held guilty. Take a city with a large population, say Dallas. That's 1.3 million people. 1,300,000,000/1,000 will give you *multiple* people in that city for which the evidence would be wrong.
Furthermore, the evidence is handled by people, and people can ere. Samples can get mixed up. Tests can be botched.
Add to that a not so distant case of a doctor rapist who beat the dna tests by surgically inserting a sealed tube of someone else's blood in his forearm.
The simple fact of the matter is that no evidence is bullet proof. And yes, it can lie if used incorrectly.