Popular bulletinboard site 4Chan was in the news last week, because a group of 4chan readers successfully pranked Time Magazine. Time was conducting an online vote to determine the "world's most influential people in government, science, technology and the arts." Some guy named "moot" got almost 17 million votes, and the rest of the list has a bunch of odd names on it too. Well, moot is the founder of 4chan, and the first letters of all of the names on the list spell out "Marblecake, also the game," which is some kind of 4chan inside joke. A cadre of relatively-technically-sophisticated and highly wired individuals were able to game the system that Time Magazine was using in order to manipulate the results for their own purposes. In this case, the stakes are low, because the poll was just for a magazine item and the nefarious purposes were the private entertainment of a group of pranksters. But what happens when the network effect is abused for more serious purposes?
I thought about this topic because this week I got an email from a friend of mine, who entered a contest sponsored by the Kona bike company to name their newest bike. She's one of the three finalists, and the votes are being tallied online. She sent out an email to all of her friends asking us to vote for her bike name (Cadabra), and if she gets the top spot, she wins a very nice bike. She said we should spread the word, and that got me thinking. Influence on the internet is distributed very unevenly. As the past week's hype over Ashton Kutcher and his race to get 1 million twitter followers before CNN points out, certain people have the ability to rope a lot of people into their schemes (stupid or otherwise) while others have a much smaller soapbox. This is, of course, the way it's always been, and the internet has actually made this situation better, because by and large the internet is pretty meritocratic. If you have something really interesting to day, consistently, you can start a blog or a twitter feed or something, and build an audience. If you're charismatic enough, you can even start a following. And because communication is so easy, you can call your readers or your followers to action.
Calling people to action used to be much harder for two reasons: First, getting the word out either required printing and distributing some kind of document or gathering your followers into an arena and giving a speech. With TV and radio, you had more options if you were already rich or influential enough, but the average person was blocked out of those media. Second, in the olden days the call to action usually involved some kind of physical, intentional act, such as holding a demonstration, going to the polling place, writing letters, or changing some kind of personal behavior. Online, calls to action are frequently more like "go to this link and click a button." What's changed is that both the communication with your following and the action that your followers must perform are now so easy that anyone can do it. And this isn't necessarily a good thing for democracy.
Back to the bike naming contest for a moment. I really think my friend deserves to win, and she's behind by a couple hundred votes. If only a fraction of the people who read this article were to go to Kona's web site and vote for "Cadabra," she'll probably win. This isn't even against the rules, or even the spirit of the rules (as long as every vote is cast by a real person and there's no shenanigans). My friend was encouraged to get people to the site to vote, because Kona's primary reason for having this contest is to promote their new bike, and the more people who know about it, the happier they are. It's a well-designed viral marketing campaign, really. But in this case, something is at stake: a valuable real-world item to be won or lost. Because I have more influence (simply by being able to write about it at OSNews) than the average person online, I can affect the tally more than the average person. A contest like this is really testing the power of an individual's network. We'll see how my hypothesis tests out when the votes come in.
There's a similar situation with all of the talent shows on TV these days, such as American/Pop Idol and Dancing with the Stars. In these shows, multiple votes are often allowed, so you're testing not just the popularity of a contestant, but the relative fervor of the voter, skewing the results toward the most fervent viewers with the most time on their hands, which I suppose this is true genius because that's also the prime TV-watching demographic. In these cases, having the contestants engage with their networks of fans in order to try to secure some kind of advantage over the competition is expected.
But what happens when the stakes in the outcome are a little more serious? What happens when there's an expectation that the standard polling rules will apply? A lower-stakes example from a couple years ago was the Washington State quarter design. The poll designed to judge public reaction to three proposed designs didn't prevent multiple votes from a single IP address, and someone set up a relatively simple ballot-stuffing bot and spoiled the results. That one guy really wanted one particular design on the quarter (or more likely just wanted to see what would happen if he tried to juice the poll).
We're probably all familiar with other examples of these kinds of events happening, either by one person exploiting security holes in a voting system and stuffing the box with fake votes, or people with influence mobilizing their followers to cast genuine votes in numbers far greater than those of other voters.
In each of those cases, we end up with what I'm calling tyranny of the minority. One of the dangers of direct democracy has always been that the majority of people can band together to persecute an individual or smaller group using legitimate voting, such as voting for confiscatory taxes on a wealthy individual, or restricting the civil rights of a minority ethnic group. This is called "tyranny of the majority." That's why no country practices direct democracy. There always needs to be a constitution to enumerate essential rights, a court to ensure that the constitution is obeyed, and a representative structure such as a legislature to insulate the nation's laws from the whims of the voters. A tyranny of the minority is when a vote is open to anyone, but because not enough people are engaged politically, or not enough people know about it, a small group can organize itself to make a surprise assault on the poll and exert disproportionate influence.
In the case of trivial votes, you can either embrace the possibility that someone will try to sway the vote using their network (like the bike naming contest or other publicity stunts) or you can give yourself a legalistic out, wherein you state that the vote isn't binding and you can choose the winner regardless of the vote, such as what happened recently when Stephen Colbert tried to have a NASA space station wing named after himself (when he won, they decided to name an exercise machine on the space station after him instead).
There are also technical and procedural steps you can take to prevent interference. Most basically, as I mentioned before, you can try to prevent multiple votes from a single source, whether it's an IP address, login account, or both. You can use a CAPTCHA or other tool to try to make sure that it's a human being that's voting, because a botnet could easily make votes from thousands of legitimate IP addresses. And you can make rules about who's eligible to vote (evidence of a huge botnet of almost 2 million PCs controlled by criminals was recently discovered). If we had a poll at OSNews, we could limit it to people who had user accounts before the poll started, making sure that people didn't sign up for new accounts to vote twice. There are also steps you can take that will thwart casual scofflaws but probably not a determined scammer, like geo-targeting or banning known proxy servers.
Reducing the likelihood of vote tampering will likely mean using a combination of all these tactics, and their effectiveness will depend on the nature of the vote. A good example from recently of a real-world high-stakes vote that was disrupted completely within the rules by use of advance planning happened in Singapore. A feminist organization saw a sharp increase in new members in advance of its yearly leadership election. When the results came in, most of the organization's leaders were replaced, and it turns out that a religious organization had made a coup, and essentially taken over the leadership and resources of an organization whose views it apparently disagreed with. In this case, all of the votes were legitimate, and the only way the old leadership could have prevented the takeover would be to have restricted voting by new members.
The issue of the security and integrity of online voting isn't serious because of magazine lists, bike contests, NASA modules, or even advocacy organizations. It's important because people have been talking about using the internet for civil voting. I can understand why people are tempted to pursue online voting. Greater participation in the democratic process is a worthy priority, and the easier it is for votes to be cast and counted, the greater number of people will be able to participate, and the less the election will cost the taxpayers. Easier and less expensive voting could even enable more frequent elections on more topics. That would strengthen democracy, right?
I'm not convinced that it would. As anyone who's been following the ups and downs of the electronic voting controversy over the past few years knows how difficult it is to create an electronic machine that will take a vote that we know we can depend on, whether Diebold is run by a shadowy Republican conspiracy or not. Multiply that by ten for remote e-voting over the internet. But let's set aside the technical aspects of a secure vote. That's been a topic of deep conversation for a long time, and it's pretty clear it's a pipe dream anyway. But let's assume for a moment that we could come up with a system that only allowed for legitimate votes, and we could have 100% confidence in that fact. Let's assume that this system enabled votes to be easy to cast and easy to count. This system would probably work fine for big, high-profile elections like the presidency and congress, because the candidates and the parties are already doing everything they can to mobilize their troops to vote for their person. Where the tyranny of the minority would come into play would be the smaller races, such as school board, county sheriff, and other local ballots. These are races that are much more easily swayed by an organized group that represents a small minority of the voters but can swing the vote their direction if they're determined enough. This is something that happens already every election, with manual voting, but with electronic voting, it would happen much more. I'm afraid that with remote e-voting, coupled with every more useful and popular regional and local social networks, Stephen Colbert would win every election in the country.
Update: My friend, who works with Cancer patients, just informed me that if she wins the bike, she's going to raffle it off to support cancer research. The voting has been re-set, and now they've reset the vote for the final two, so go vote again.