Travesty of justice and statistics

In a development that boggles my mind, the U.S. court system is completely ignoring a basic tenet of statistics and (wrongly?) convicting people because of it. Police are running partial DNA samples through databases of millions of people, and claiming that the chances of anyone in the database matching is the same as the chance of one person matching. That's like saying that the chance that someone has hazel eyes is 1 in 20, so the chance that anyone in the world has hazel eyes is 1 in 20. A man was recently convicted of a 38-year-old rape-murder in San Francisco based on a degraded DNA sample. The jury was told only that the chance of a match was 1 in 1.1 million, and not that, given the size of the database, the chance of a random match was 1 in 3. The judge prevented the defense from challenging this number, and even from saying that the defendant was a suspect only because of this match. The Washington Monthly and LA Times have articles on it. The SF Chronicle apparently couldn't be bothered to do any research, just putting out the prosecutor's line. The Washington Monthly article even points out that someone was fingered using this technique for a crime she could not have committed, since she was in prison at the time. How the courts could so willfully misunderstand statistics, and how the FBI's scientists can have the gall to testify to false odds, really baffles and enrages me.