You are completely missing what the Cambridge Analytica data is all about. It didn’t say Joe BillyBob in Precinct XYZ, county ABC, state DEF is going to vote for A, so you need to go into that county’s voting systems and change his vote to B. It said Joe BillyBob has network connections, profile data, viewing and like history, etc., that suggest he belongs to a demographic where exposure to messaging that says, say, white male coal miners that look like him at the nearby BigDig coal mine will benefit from Trump and hurt from Clinton, will be effective making him more likely to be influenced to vote for Trump and drive him, in turn, to likewise influence his connections.
Remember that the Cambridge Analytica data came from a platform designed and precisely engineered for exactly one purpose: generating data useful to more precisely and effectively direct the herd. It’s called advertising and what Facebook and Google and all the rest do is enable it to be vastly more selective, strategically targeted, and effective.
The regular customers for this data don’t use it to manipulate what Amazon and WalMart report selling or think they sold. They use it to manipulate what customers actually come to buy.
Inferring from lack of evidence the Russians manipulated the count that they had no impact on the election outcome ignores the effectiveness of the advertising model and its social engineering component. Countless billions of dollars are spent each year betting on that model’s persuasive effectiveness. Maybe it failed to shift any votes. I’m betting it shifted a lot more votes than were necessary without breaking into a single voting or vote tabulation system. It’s easier doing it this way and an already proven method.