This upset is not really surprising given the proliferation of un-auditable "Evoting" machines in the swing states.
My state has the new Diebold machines - which in 2003 was publicly discredited by the Johns Hopkins Information Security Institute and Rice University...to no avail. Here's the press release:
Electronic Voting Security Flaws:
Johns Hopkins Researchers Respond to Diebold AnalysisAugust 1, 2003
http://www.jhu.edu/news_info/news/home03/aug03/e-vote.h... Anyhow, exit poll MOE is typically 3% in the national elections (as well as the larger states) and has been relaible in analyzing election in both industrialized and third world nations for the past 25+ years. So the exit polling methodology is mature, but the 4-5% variance reported yesterday b/t the 5pm exit polling data and final votes for Florida and Ohio suggest either 1) the people were lying to the pollsters about who they just voted for -unlikely, 2) fewer Bush supporters were being exit polled- GOP answer, or 3) the voting machines are lying - whcih is my analysis of the anomalies.
Indeed, since the 2000 election, the exit polls have become curiously unreliable, with Georgia in 2002 being the serious example where Cleland was polling ahead - but lost by a strange 10% swing to Chambliss in the final vote totals. (Diebold DREs were first rolled out in that election)
I'm not a 'tin-foiler', just a student of history and I do have a Masters in Informatin Technology, with a concentration in Information Security (INFOSEC), along with CISSP certification, and I am in good company those like Bev Harris who understand the security flaws/hackability of the Diebold central vote tabulators (which count up to 2 million votes per server - thus gaining access to just one of these servers is enough to sway a state's vote). Examples of security flaws reported last year included things like the password default on the *central tabulator* was "1111" or "11111" - until exposed on the Internet, and modem access #s were also revealed for anyone curious...thus hacking in from the inside and changing the total would be easy - and very tempting for political operatives.
Indeed, I am becoming convinced that security flaws in the DIebold DRE-paperless/auditless/'proprietary code' voting system is simply too tempting of a target to ignore for political operatives - especially when the 2 monopolies (Diebold and ES&S) are heavily-aligned/invested with the GOP.
IMO, the only way to spot systemic fraud would be a technical review of the server's internal audit logs and all modem/uplink activity of the central tabulators along with the source code in the harddrives of the voting machines themselves, plus the source code of the central storage servers/vote tablulators in both Florida and Ohio...which would require a small army of network and/or INFOSEC forensic experts - not lawyers who would not know what to look for. Well, that ain't gonna happen b/c Diebold has successfully argued these are "Trade Secrets" issues...so, we are firmly in the Orwellian world of black-box voting.
On that note, it's ironic that I received a paper receipt out my bank's Diebold ATM yesterday when I withdrew $20, but I did not receive anything when I voted yesterday on Maryland's shiny new Diebold touch-screen voting machine. I suppose a responsable Congress would mandate that Evotes be treated in the same way electronic financial transactions have been transmitted/protected for the last 20 years...but that would of course would require campaign finance reform to sever the corporate influence, and a Congress that actually wanted the People to have faith in their voting system. Sadly, that is too much to ask.
Regrettably, given the 2000 debacle in Florida, I think we are a supremely naive and foolish nation for allowing an electronic voting system to be implemented in numerous states that does not provide any sort of paper audit trail to verify the voting data.
Bottom Line: Unless a credible whistleblower appears, or a court order allowing the seizure of any and all Diebold central vote tabulators in both Ohio and Florida - and the
internal audit logs had not been shrubed - I don't see how evidence could be collected to prove or disprove what appears to be a case of hacked election results. Of course the problem is this is an INFOSEC issue and most people will not understand the complexity involved in doing an Information Technology-based forensic investigation.