Matthew Ingram's article on GOGAOM reviews a study done by IBM on Twitter 'coverage' during the evolving reporting in the immediate aftermath of the Boston Marathon tragedy. It is no surprise that much of the early Twitter feeds were wrong. Some were rumors, but some were reporting information derived from police scanners, which were also wrong. I tend to follow CNN and they got it initially wrong.
During the reporting of the Benghazi attacks, I followed coverage via BBC radio. Their reporter was on the ground in Benghazi and got in touch with the landlord of the U.S. Consulate. He lived next door to the consulate and watch the whole thing from his window. His accounting was that a peaceful crowd of demonstrators gathered in front of the consulate and shortly after an armed group in pick-up trucks ran them off and began attacking the building. Never heard anymore about his guy and his account. I don't know if his witness was accurate or not as things began spinning off on who was at fault.
Anyway, in this day of instant reporting I, like Mr. Ingram, suggest that whatever source you follow, take it with skepticism and maybe even follow multiple sources. In any case it may take one or two days before an accurate report begins to form from the myriad sources we must deal with today.