To begin with, the fact that one is practicing the prior art doesn't prove they're not also doing something that infringes. For example, if I design and build a car that uses a gasoline engine and has LED headlights, I'm practicing the prior-art inventions of Daimler and Benz for a gas-powered automobile. If a patent holder claiming to have invented the combination of LED headlights sues me, the fact that I'm practicing the prior art makes no difference. The real question is, do the patent claims cover the use of LED headlights on an automobile--and does my car infringe those claims. An infringement analysis is determined by comparing the claims of the patent to my device, "not by comparing the accused device to the prior art."
What people really mean when they say they are practicing the prior art is that they are not doing anything new. In other words, if the patent claims cover what they are doing, then they must be invalid because they read on the prior art. So why does it matter whether the issue is infringement or validity?
The answer lies in the burden of proof in patent cases. To prove infringement, a patent owner must convince a jury that, in light of all the evidence, it is more likely than not that the defendant infringes. At trial, the jury will see the patent and the accused product, and will hear testimony from experts on both sides of the case to inform its decision. Jurors realize that most of the witnesses they hear from have some bias in the case, so they rely on their judgment to decide which side is telling the truth. On the other hand, when the validity of the patent is challenged, the jury is asked to question the judgment of the Patent Office--which the jury knows is unbiased. To prove invalidity, the defendant must prove by the much higher "clear and convincing evidence" standard that the patent is invalid. In other words, the jury must decide that the Patent Office clearly made a mistake.
During the past 12 months, the U.S. Patent office issued approximately 182,000 utility patents containing millions of patent claims. Many of them are very likely invalid. But, they are all presumed valid until proven otherwise. Given the choice between convincing a jury that the U.S. government got it wrong in any particular case and proving that the plaintiff has it wrong, most of us would rather take on the plaintiff, which is why lawyers and their clients continue to argue the "practicing the prior art" defense.