Why don't people trust the media?

Strategypage explains How the Lancet Cooked the Numbers:

The recent survey, published in the British medical journal, "The Lancet," claiming over 650,000 civilian deaths due to the liberation of Iraq, was quickly labeled propaganda, not science. Is the survey accurate? The answer is, apparently not. The survey is widely out of sync with casualty counts by other organizations, and by a wide margin. A 2004 study by the same authors claimed 100,000 civilian casualties – a survey at odds with one done by the United Nations at the same time (which estimated 18,000 to 29,000 deaths). To compare this with other studies – the group Iraq Body Count only claims 49,000 civilian deaths, the Brookings Institution reports 62,000, and the Los Angeles Times has reported 50,000 civilian deaths since the liberation of Iraq...

Another sign this number is off is the fact that it implies that, on average, hundreds of civilians per day have been killed since the liberation of Iraq. Not even the mainstream media has reported death tolls that high. One recently reported "surge in violence" involved a total of 110 people killed in a two-day span. That is an average of 55 people killed per day – which would imply a total of roughly 72,000 civilian deaths (somewhat higher the Brookings Institution estimate).

As Dan Rather might counter: the facts are "Fake but Accurate".