So - if your a campaign strategist this is EXACTLY what you count on happening.
Then you alter those inputs that will skew the results but knowing that there is no real correlation of (input x - say its phony voter registration) to a real change to probable outputs.
I'm not necessarily saying your wrong, I'm pointing out (in light of the evidence of this note ostensibly written from within the 0bama campaign) that a stragey like this is a real possibility given the assumptions that you point out.