"hand count matches the machine count"? (Score 3, Insightful)

Where do you get that from? Looking at the actual results, I see numerous instances where the initial count differs from the recount. More often than not it's just a 1 or 2 vote discrepancy, but it is pervasive. During the early stages of the recount it was being reported that getting on for 1% of the votes hadn't been counted; either the machines would ignore every 100th ballot or so, or they'd ignore a whole batch because pens with the wrong kind of ink had been used. Most of the batched errors were caught on the day of the count, but it took a recount to show up the sporadic individual errors.

Why is this important? One reason is because there is a demographic difference between hand-counted and machine-counted areas, so if you're going to disenfranchise 1% of the population in machine-counted areas, even if it's done entirely at random, that reduces the elective power of the demographic in those areas, which can tip a close election. Besides, isn't every vote supposed to be counted? Isn't that why you go and vote, because you believe that your vote will be counted? How many voters wouldn't bother if they knew it was some sort of lottery?

Another point: a partial recount will never disclose a fraud if the people choosing the areas to be recounted are also the people behind the fraud. They will simply leave the areas where the fraud took place until last, secure in the knowledge that the recount sponsor's finances or resolve will run out before the recount gets that far. And it certainly was the case that Kucinich couldn't specify exactly what was recounted. His requests for a tally of the uncast ballot papers, for instance, fell on deaf ears. So what happened to those uncast ballots? Did they get cast after all?