This is one of an occasional series of fundamentals in election reform/integrity.
There is a long history of voting fraud in North America. Most of the fraud has been of the retail variety. However, retail fraud is now functionally obsolete and I will disclose one of several methods how wholesale fraud works and discuss why is is much more likely.
There is a long history of voting fraud in North America. Most of the fraud has been of the retail variety. For example, there is the classic case of 'voting the graveyard' where proxies come in and pretend to be voters that have actually died. My grandfather in law, an elected sheriff in Pennsylvania after WW-I told a terrific story about elections in the little town he lived in. The town fathers asked him to run for sheriff and he expressed concerns that he had never ran for office before and was concerned whether or not he'd be elected. They told him not to worry about that and showed him how the main ballot box actually delivered votes into a room in the basement of the polling place. Ballots were inspected and those that didn't meet the town councils requirements went into the fire box of a nearby boiler. Another story he told was about school board elections. The school board was able to persuade the voters to build a new, modern school with lots of windows. In the next election that came around for the school board, the opposition saw them through the windows of the room in the new school substituting premarked ballots for the votes they didn't like. They got into quite a bit of trouble. Old Lou loved to tell stories and wasn't above a little creative embellishment now and then but there's truth there somewhere.
Many of you have heard or can even tell stories of similar frauds. Our fine brethren south of the border have a wonderful tradition of things like; the pregnant urn, the taco and the wild mouse. Brevity forces me to direct you to Vote Fraud. However, these frauds are all pretty much what is called Retail fraud.
Retail fraud usually involves a ballot per ballot substitution or similar technique. Over the years, registrars of voters had developed techniques to defeat most of these retail frauds. For example, ballots are carefully counted before going to the polling place and are counted the morning and evening of election day to make sure ballots haven't 'disappeared'. The first voter of the day is shown the ballot box before it is sealed to demonstrate that it starts out as empty. Also, candidates or allowed to place observers in the registrar of voter's office during the night tabulation of votes to observe the process. While these are not perfect mitigation efforts, they do help. So, to corrupt an election by traditional means takes a lot of effort and generally requires a large number of people to be involved. In short, it is a retail operation just like McDonalds. With the large effort and the large number of conspirators involved, the difficulty of doing it in the first place and the chances of detection afterwards increase almost geometrically with the number of participants involved.
However, there are now new and improved wholesale methods to commit fraud. Poor people and most of the rest of us are generally too busy to go out and buy an election. However, the rich usually have the time and the resources to buy an election. Now rich folks don't get and stay rich by paying top dollar for everything. Often as not they buy things at a discount or even wholesale. One wholesale method of stealing elections focuses on where there are the most votes, the most confusion and where people aren't always paying close attention. In short, they buy their elections where the votes are being counted. Stalin is reputed to have said that, "It's not the people who vote that count. It's the people who count the votes." While you may have in your mind some 1930's movie version of how votes are counted with stacks of paper, chalk boards and men with cigars and suspenders, that's not the way that it's done today.
Instead votes are added to a database system commonly called a tabulator (please, now tabul-now jokes). The data is either read off of memory chips (like you might use in a digital camera) taken from touch screen machines or polling place paper ballot scanners or someone sits in front of a computer and keys in the data. Or, if they use a central paper ballot scanner (increasingly common), it may be directly networked to the database. So, you've got this choke point called the tabulator that all the votes have to pass through before the totals are displayed. What better place to modify the results and hide the manipulations than among the great mass of data handled by the tabulator?
My pet method, which is nominally thought of as a 'man in the middle' attack, consist of a simple, hard to detect, program installed on the tabulation system before or during the election would 'watch' the vote totals in the election contest(s) of interest. I use the word 'contest' to cover both the elections of people and the deciding of proposals submitted to the voters. If polling data showed that the contests was likely to be 'close' only a few votes would be swapped between yea and nay from a few precincts, randomly chosen, more or less in real time, as the votes accumulated. However, these small changes would be able to change the overall results. Many elections are decided by surprisingly small amounts. Your typical registrar of voting get down on their knees before an election and pray for a landslide, regardless of for whom. These small changes would be hard to notice during the controlling chaos of vote tabulation. Now if the polls showed your contest would be rather unlikely to win, you might have to swap around votes in nearly every precinct to get your desired results. Still we're talking a fairly hard to detect manipulation.
You may reasonably ask, "Why is this manipulation hard to detect, Ed?" It only takes one corrupted insider to slip the program into the tabulation system. The major parts of the program could be resident on the tabulation system from day one of use. Or it could be resident on the electronic voting machines at the polling place and would be transmitted to tabulator using the previously mentioned memory chips or, skipping the voting machine, written on the memory chips prior to use in the voting machines. Then, all that would be needed would be a little data file with the desired result. The critical information for the vote flipping could be added to the tabulation system over a network connection or even wirelessly with something a simple as a Palm Pilot. Another wrinkle is that the vote flipping program could have a date checking routine. The software code, which could be parked anywhere on a computer, would only install itself to the tabulation system when it 'knew' it was election day. Then it would uninstall itself (or even erase itself) the day after the election. Before you start measuring me for a tinfoil hat, please be aware that this scenario is not mine and is from or has been reviewed by a number of highly respected experts in computers, cryptography, and elections.
The final reason why this sort of vote flipping would be difficult to detect has to do with the normal numbers of errors that crop up in most every election. People incorrectly mark (spoil) ballots, show up at the wrong polling place, clerks have trouble adding up numbers, it's oh dark hundred in the morning, people just want to go home to bed after getting to the polling place at 5:00 in the morning and there are plenty of other opportunities for error. The number of vote flips would typically be below the normal error threshold or if higher than that would still as likely would be excused as errors in the process. Remember, your typical Registrar of Voting is an adminstrator. They don't have backgrounds in cybernetics, cryptography, statistics and all the rest of the hard and soft sciences needed to competently run today's elections. You are lucky if the Registrar has a MBA. Please be aware that I respect how difficult running elections can be and I don't wish to cast aspersion on Registrars of Voting.
There are other ways elections can be corrupted but I only promised you one example so that's that. Wholesale fraud involves corrupting a few or even one person in a critical location in the tabulation process. Or the corruption could occur at the factory where the equipment is built and configured. I admit that my first reaction in 2002 when I got involved with election reform/election integrity issues was to focus on retail fraud. It's easy to understand, it appeals to the'...we wuz robbed!....' sentiment and it has a somewhat romantic history. I hope the above example shows that concentrating on retail fraud is a harmful distraction when working on election reform/integrity issues.
When the notorious bank robber Willy Sutton was asked, "Why do you rob banks?," he is reputed to say that he did so because that's where the money was. Willy wasn't going to walk around sticking up individual people when there was a more efficient method of stealing money as in robbing banks. Similarly, robbing election one vote at a time is inefficient when you can do it wholesale at the tabulator or similar 'pressure' points in the process.