And before the recounts started, there was 1784 votes in favour of Bush. And if you red the bottom parts of your link, you'll see that Gore could've won depending on the standard of counting. Let me put it this way, Bush's victory was within the margin of error for an election and that's a problem, even if you do have a fair win.
What you do want is to get a system, where there's one standard, so that a recount does give the same result and not an official 1200 vote difference. You also want a system where you can follow the recount, rather than "this machine with substandard encryption says so".
1-2% systematic difference only done during close elections and you can do it for decades before you'll see a clear tilt in the margin of error for the polls.
Bookmarks