US electoral fraud - could it swing things?

I'm not going to report yet on the second US presidential debate, as I haven't had time to watch it yet. For what it's worth, post-debate polling numbers look very similar to the first debate, i.e. Obama a modest winner. And that wasn't good enough for McCain, who needed a big win.

Or did he? Last night's newsnight had a piece by Greg Palast, author of The Best Democracy Money Can Buy, examining the extent of electoral fraud in the US. Really frightening stuff, including such horrors as the following:

• huge numbers of (mainly low income and/or African American) voters turning up at the polls only to find that they weren't registered to vote. This was especially likely to happen in states like Colorado, with a Republican Secretary of State.
• Voters whose houses have been foreclosed (i.e. repossessed) being disenfranchised
• Voter intimidation at the polling station (which is apparently legal in some forms in the US - what the hell?)
• shortages of voting machines in polling stations in low-income neighbourhoods
• rigging of electronic voting machines (many of which have no 'audit trail', i.e. no means of establishing whether the votes counted tally with the votes actually cast.)
• new legislation requiring photo ID for people to be able to vote - people in the US without any form of photo-ID (e.g. no passport or driving licence) are overwhelmingly from the poorest section of the population.
  

Greg Palast has several earlier short films for the BBC on his website, so hopefully this one will be put up there soon, as it's a must-view. The big questions arising from this are:

1. How much worse is voter fraud now than in 2004, when it is widely acknowledged to have cost John Kerry the election?
2. Given that it looks like (hopefully) Obama will be a lot further ahead this time in terms of voting intentions than Kerry was in 2004 (or Gore in 2000) how much fraud would there need to be to 'flip' the result to McCain?

In the comments on the fivethirtyeight website I found a link to this site which claims to reveal "a [Republican] plan to steal the next election by 51.2% of the popular vote and three [electoral college] votes."

Also something even more extraordinary: an interview with a Republican "cyber-security expert" who claims that foreign hacking of voting machines could decide the election... watch that one here. I haven't had time to watch it yet but will do soon.

One thing seems obvious: the exit polls from the 2004 election was pretty close between Bush and Kerry - and although it was surprising that a number of swing states went to Bush, the overall reported share of the vote was not that far out of line with what was predicted. It was a real bummer, but it took some time for the truth to emerge that the election was rigged. And by then it was too late.

By contrast, if Obama is (say) 8% in front in the polling in the run up to election day, but the actual votes show a 2% lead for McCain, it is very likely that all hell will break loose. At the very least you could see large-scale civil disobedience, and in the most extreme scenario, civil war. I hope that Greg Palast Newsnight film gets shown in the US before polling day, so that people know what they're up against.