Over at Debunking Christianity, apostate Hector Avalos says:
One should not let these apocalyptic interpreters forget that it is atheists who have been 100% correct in predicting that those end dates will fail, while it is believers who have been 100% incorrect. In other words, atheists (and other skeptics) have been the best "prophets" when it comes to these end dates.
To simplify, let's use round numbers. Suppose Jesus was going to return before 3000 AD. That means the odds are about 11,000 to 1 that he won't return on any particular day during that interval. Any astrologer with half a brain could safely predict that any given date for his return will be wrong. Mathematically speaking, the odds are overwhelmingly against the accuracy of any date you pull out of the hat. And that will be the case even if Jesus was, in fact, going to return before 3000 AD.
Predicting when something won't happen can be infinitely easier than predicting when it will happen. If it happens on one day, it won't happen on all the other days. The days when it won't happen outnumber the day when it will happen by many orders of magnitude. Don't pat yourself on the back when you accurately predict a nonevent. In general, it takes no foresight to predict the nonoccurence of an event on any given date.
Some events are predictable if they fall under human control. Likewise, some events are predicable if the outcome is connected to an observable a chain of causes, like running out of fuel, or the trajectory of a hurricane.
But in the abstract, a one-time event is unpredictable in the sense that it could happen at any time, yet it won't happen most days, weeks, years, centuries–even millennia (or more).
For instance, some scientists theorize that a vast asteroid struck the earth about 65 million years ago. Let's grant that for the sake of argument. Consider all the days when it didn't happen, both before and after. Yet impact craters bear witness to such events, however rare or isolated.