None, just lots of smoke.
In conspiracy-theory thinking, one allegedly rigged precinct means they must
all have been rigged. Hence not only would Pennsylvania have gone to Romney, but also other states and their electoral votes.
However no one is seriously contesting the precincts where voting was not unanimous. My state, Utah, was understandably the most ardent supporter of Romney, with urban counties splitting the vote relatively evenly (or slightly in favor of Romney) and most other counties turning in 85 to 90 percent in favor of Romney. These highly lopsided results are still considered credible because Utah is a Republican stronghold state. In fact I believe in this election we earned the distinction of being the reddest state on the map.
That leads to the question of wanting to
credibly steal an election. Knowing that each precinct, county, and state must report its results publicly, why would a hypothetical vote-rigger make an entire precinct
unanimous and thus be sure to attract the attention of people whose intuition would tell them the vote should be at least somewhat divided? A credible plan to rig an election (or do anything illegal) generally comes accompanied by a plan to conceal one's actions and not get caught. Criminals who know what they're doing know how not to be noticed or seen. But as usual, the hypothetical perpetrators in conspiracy theories are bumbling and inept. "Hey, we need to make sure it looks like
someone in Philadelphia voted for Romney, otherwise it will look suspicious." "No, let's make it unanimous. We need to be sure."
And keep in mind for those precincts there's still no boots-on-the-ground evidence. It's all a conclusion inferred from the precinct reports. As was reported in the cited article, only a few voters were registered Republicans in these precincts, and at least one of them didn't realize he was. Given the demographics of the region, that's not surprising. The whole reason we employ statistical probability over intuition is that life often surprises us.
The Law of Averages leads to many misconceptions in electoral politics. The one I'm thinking of is a fallacy of decomposition, or the fallacy that says that the behavior of the whole should be individually reflected in the behavior of the constituent parts. That we went into the national elections with opinion polls reading 49%-49% or some such deadlock doesn't mean those proportions hold at each polling granularity. As mentioned, in many Utah counties 4 out of 5 or 9 out of 10 voters voted for Romney. That's offset (and more) by liberal counties where similar turnouts occurred in Obama's favor.
But unanimous? Yes, that happens. It happens very infrequently, which is why the unanimous precincts comprise only a tiny outlying part of the distribution. If you graph the party swing by precinct, it will follow a normal distribution. That means precincts with pronounced swings occur with little frequency (if at all) and occupy the narrow tails of the distribution -- in this case well outside two standard deviations. The Law of Averages fools us into believing that improbable events never occur. They do -- just improbably.
Let's go back to back to the computer programming. Let's say you planned to rig the election, and you planned to do it, as our "auditor" alleges, by changing the programming in the voting machine. If you have access to the programming, why would you do something so monumentally stupid and
reveal the changed vote to the voter in the user interface!?
Again, our hypothetical vote-riggers are bumbling and inept. They might as well right large bells and flash big red lights and say, "We're changing your vote!"
If you control the programming, you control the user experience. One part of the computer program presents the graphical interface, driven by data stored in its memory about candidates and issues. Another part of the program writes the votes to the Smart Card, the network, or data transmission medium. The central data structure is a table of voter issues and offices, associated with digital "check marks" representing the voter's choice in each, or that the voter has made no choice.
During the voting session, the UI portion of the program manipulates this data structure. When "Cast Vote" is finally selected, the other portion of the program takes over, reading the data structure as previously composed by the UI and rendering it in a form acceptable to the transfer medium. Error- and tamper-check data blocks are added. The place to change the voter's vote is there. The
only thing that matters in terms of the vote tally is what goes on that Smart Card. If you're going to change a vote, change it as it's being prepared for storage. There's no inherent condition that requires any sort of thing to be displayed on the screen while a voter's votes are being tallied, recorded, and (allegedly) tampered with.
The silly presumption that software-doctored votes
must be visible to the voter as he looks at the ballot is evidence of just how ineptly and inexpertly considered these vote-rigging conspiracy theories are formulated. The alleged actors must always operate in the stupidest and most detectable ways possible, with no thought toward concealment or best-practices. As such they are not credible.
