You absolutely can buy a database of voter data from vendors who specialize in this and use it for whatever purpose. I did it when I worked on a committee supporting a local referendum. We used the data to identify voters likely to support the referendum and made sure to get in contact with them to encourage/remind them to vote. The referendum passed.
So, I think what poochyena is suggesting is:
- Buy voter database
- Use it to identify people who rarely vote (the db we used allowed us to query any household and find out in what elections they had voted in the past)
- Show up to their polling place, say you're them (with their address) and vote.
- Repeat until sufficient number of votes
As DarthSlack pointed out, very few races are decided by a handful of votes, and as I pointed out, any race decided by that few votes is going to be heavily scrutinized anyway, probably leading to this tactic being unsuccessful.
But plenty of races are somewhat close. Let's use the recent IL-9 House primary as an example. The winner beat second place by ~4,000 votes.
IL allows early voting up to a month and a half (ish) before elections, but in practice, most early voting places don't really open up until the two weeks before the election, and really only have full hours on the weekday.
So let's assume we're using a 10-working-day window of early voting to accomplish this.
Let's assume a person could visit 10 early voting locations on a day (aggressive, but possible). Each person could then cast 100 votes over the early voting period. Assuming everything goes well (they remember the address of the non-voter, the election judge checking them in doesn't get suspicious, they don't get stuck in traffic, the non-voter doesn't show up to vote, etc...).
You'd need 40 people. And in this scenario that would yield a very close win, which would probably be scrutinized. So you'd probably need 50-80 people to a) avoid the possibility of scrutiny and b) deal with attrition from the 100 votes/person (see the above reasons).
4-7 dozen people would be involved in this conspiracy. The most successful conspiracy in history is generally considered the Tuskegee Syphilis Study. Though a number of people were involved in various elements, the co-conspirators who knew everything likely numbered less than 10.
So, incredibly difficult, even in a reasonably optimal scenario (above).
Far easier, and more effective, one suspects would be things like
raiding election offices,
controlling voting machines,
controlling voting machines (again), and of course good old voter intimidation, voter suppression, and
aggressively purging voter registrations.