There's been a lot of talk in the Berkman group about using technology to help voters make more informed decisions. As I was reading this article from The Nation (thanks Jim Moore), I got to thinking: if Dean knew that the mass media were going to turn on him, why didn't he plug his website at every opportunity? Well, it wouldn't have been very useful because it is mostly a rah-rah brochure with lots of padding in between actual position statements.
Let's face it: when a candidate is on TV or in a debate, they give canned, generic responses that I'll call brochure answers. These are designed to make everyone feel good about the candidate, without giving away enough actual policy to turn anyone off. They're fluff.
Now, if a candidate actually did stake out some positions in any detail, we would have a much more honest and straightforward election. It would become feasible to know what policies Mr./Mrs. Whoever actually advocates. Right now the best place to get candidate position information is perhaps from the League of Women Voters' DemocracyNet, but the candidate-supplied statements there are incomplete and vague. They're statements manufactured during the campaign to have the same brochure appeal as all the other garbage we see. DemocracyNet gives you power to compare, but what you're comparing is sales pitches. We'd like to compare facts about political issues.
So how do we compare facts? First, I'd like to see major issues (e.g. "Abortion", "Gun Control") broken down into specific multiple-choice questions ("Roe v. Wade: should we keep or overturn it?", "Assault rifles: allow or ban civilian purchase?"). Each candidate would then be paired with each issue. Clearly, the reductionist way that these questions are stated and structured makes it difficult to guarantee objectivity. Therefore, the Voter Support System would have to make available the identity of the researchers who selected the issues and interpreted the candidates' positions. Like a good academic paper, all references must be cited. Any visitor to the site should be able to contribute to the vetting process by issuing specific references that support or challenge the validity or applicability of a reference. And very importantly, there must be no aggregate or statistical reasoning. The system shouldn't tell me "266 people agree with this statement and 127 disagree," because that doesn't tell me if it's true or not. Majority opinions, especially on a website, do not correlate to factual accuracy and leave open a huge window for abuse.
Here's a mockup:
(Hover over a link to get more information on what that link will do.)
The Voter Support System would involve three major roles:
- readers, who view the position information for candidates and offer new resources that support or challenge certain interpretations of a candidate's position
- moderators, who screen out reader suggestions, but only ones that are obviously off-topic or spam
- researchers, who are tasked with verifying the accuracy of reader suggestions and, if accurate, classifying them on the candidate-issue tree
Readers can be anyone, including known biased sources such as competing campaigns; we evaluate specific arguments, not their messengers. Moderators are a mostly administrative function necessitated because of the web-based nature of this system, but still all their decisions should be subject to public review. Researchers must be a disciplined group devoted solely to evaluating the accuracy of claims from readers.
The reputation of the voter support system is staked primarily on the quality of its issues breakdown and the objectivity of its research. I wouldn't use a system that told me if candidates were for "tax relief", because "tax relief" is an abstract term with a loaded meaning—what kind of cruel person could be against relief?! The key is to stick to specifics and avoid opinion, and gather measurable data. To initially populate the issue tree with candidate stances, you'd want to do a vast media review. You might want to enslave a large cadre of grad students for this purpose.
The key assumption here is that voters want to vote based on rational facts, not hype or emotion. This may not be true for all voters, but it's how I'd like to vote, and I think there should be a system to help me do so.
- break down issues into specific, multiple-choice questions
- research positions of each candidate on each specific issue
- publish rationales, not statistics, for all position judgments
- solicit vetting of those judgments from readers
- research the accuracy of reader arguments and publish that research
Or in short, reduce to specific issues in a systematic, verifiable, open way.
Finally, a Solicitation
If you'd like to work with me on a system like this, I can do the technical side of it. But I'd need help with the issue structures and ongoing research. Email me.