Each Thursday at 8:30 a.m. EST, the government tells us about how many people went through one of the most unpleasant experiences of their lives: filing for unemployment help for the first time. It's essentially a survey, since state unemployment is managed by your state, not the federal government.

The report runs like clockwork, but it¿s notoriously inaccurate. For one thing, the number often has wide swings from week to week, so it's a rare event for the figures to come in exactly as economists predict. Second, it is very seasonal. Folks like school bus drivers often file claims when summer comes around, and other people get retail jobs as the holidays approach. Some economists like to use it to handicap the big monthly employment situation report, but they often do so at their statistical peril

Sometimes, weekly jobless claims make political, rather than economic, noise. If there¿s a big spike in claims, some politicians will often cite the number as a sign the economic sky is falling. But, it's important to remember what the weekly jobless numbers don't tell you: you don't know how long these folks stay unemployed, how long they've been out of work in the first place, or even if they're truly out of work and not just trying to scam the government.

Because it's so unreliable, economists usually put the past four weeks together and look at a moving average. That gives a little better picture of the overall trend, but it's still not a great indicator.