Pitch
We in the US are unable to have conversations across the political aisle, in large part because we can't agree on the basic facts related to the issue at hand. People on both sides of the aisle are convinced that the other side is mistaken, duped, or, at worst, lying, about the “facts” underlying their positions. Confirmation bias - the selective search for facts that support our pre-existing beliefs - is a powerful force that prevents agreement on the facts. There are several organizations trying to mediate this problem by providing "better" or "unbiased" facts, but these organizations have not been able to earn bipartisan trust. Without the ability to agree on the facts, we have no common foundation on which to base political discussions or debates, which has led to the extreme polarization we are faced with today.
Rather than tackling polarization head on, we'd like to reframe the problem. There are still Facts (capital ‘F’) on which we can agree. The text of the US Constitution is one. The full unedited presidential debates going back to Carter, as well as other presidential speeches, are another because they are matters of public record. While you may not agree with the positions stated by a president or a presidential candidate, the fact that they made the statements are Facts.
We want to create a platform (similar to ‘Medium.com’) that provides the user with links to these undisputed, unbiased Facts, which they can use to support their position. The final article that an author would produce is essentially a tweetstorm or Facebook essay, but with links to evidence. There are opportunities in the design of the editor to help the author find evidence – using snazzy timelines, browse speeches by person, etc. The goal is to provide the author with some undisputed truths and let them focus on just thinking, analyzing, and presenting their thoughts to the reader. Readers, in turn, will be able to provide their thoughts in a similar fashion, and will be provided with opportunities to see how others, including those who are identified as belonging to different schools of thought, have used similar Facts.
If we are able to build an engaging authoring experience, then we also have an opportunity to build an engaging reading experience in which all of the content is based on evidence that the reader trusts.
There will always be disagreement about "What should we do next?" But as partisans have tried to build the case for why theirs is the path forward, we have reached a point where we disagree even about the physical reality of our present moment. There are lots of facts, but very few Facts.
One of the few Facts we do have is that the common ground between the parties is too small. Traditionally, we've built our foundation of Facts with mainstream journalism, but there is no mainstream anymore – it has cleaved into two mainstreams.
One approach to referee these two mainstreams is to inject Facts from a "neutral zone" - an organization will claim that it is not red or blue, and so its facts should be Facts – Politifact and Grasswire are examples of this method. Another approach is to mix the red and blue facts up (e.g. mix partisan Facebook feeds), and hope that we lose track of biases and muddle our way into agreement.
Attempts to mediate the Fact zone have not been effective thus far. And why would they be? The diagram at the top does not represent the present day. Our psychological reality right now is this:
It is difficult at this point for any evidence introduced by "the media" to become a Fact. Even if you remove the color from a fact by delivering it in a jumbled newsfeed or from a "neutral" arbiter, confirmation bias will cause a quick decision as to whether it is a Fact or a fact. Because there is such tiny overlap, the color of the fact also will be determined right away. The media are seen as biased, in one camp or the other, and all evidence from these sources are doomed to be facts.
A story told by a person you trust on a particular topic can be very persuasive. If you have the same taste as I do in beer, but not in politics, then I'll take your beer recommendations and ignore your political opinions.
How can friends in life, who disagree in politics, build trust with each other on specific issues? The same way we build trust in our shared taste in beer – physical evidence. I put the beer in my mouth, tasted it, and found that I shared your opinion. Similarly, our political discourse needs a way for friends to use physical evidence to build trust on political topics.
The first thing we need to establish is a "foundation" of shared facts – press conferences, debates, interviews, speeches, the raw text of laws. It is critical that the facts in this "foundation" are indisputably accurate and unbiased. When in doubt, leave it out.
The primary ways that engaged citizens publish their views are Facebook posts, tweetstorms, and chain emails, with a few power users employing Medium to do longer-form articles. Twitter is too short to capture the complexity of the issues we face, Facebook makes it impossible to support a statement with multiple video sources, and Medium's focus on text is less engaging than something based more on video. MyTake.org allows authors to write a few sentences of text, include a short video clip, add a few more sentences, embed another video clip, etc.
On our site, authors can only post links to urls within the established foundation of shared facts. By having a transparent and irrefutable foundation of facts, we can earn readers' trust that the citations they see are accurate. Because the foundation of facts will contain only complete sources, without cuts, the reader will always be able to see for themselves whether a quote or fact referenced by the author was cherry-picked or taken out of context. To ensure that our foundation isn't contaminated by selection bias, we only include complete sets of evidence (e.g. every Presidential debate, instead of just the ones that make party X or Y look good).
Currently, most individuals' opinion on an issue is determined more by their party than by their own feelings. We want to build a platform that makes it hard for groupthink to start, so that the experience, moral fiber, and reasoning of each individual can come through. One of the challenges here is that the name for a topic may be disputed – e.g., "Islamic terrorism" vs "radical jihad", "pro-life" vs "pro-choice". Any form of user tagging system will identify these groups immediately.
To address this problem, an article will be identified only by its content – no hashtags, no keywords. By discouraging viral memes, we leave more space for original thought and analysis. You can leave feedback on an article with thumbs up and thumbs down; that's it. Discussion will be hidden until you open it, and article authors will have full control over moderating their discussion sections.
As the corpus of content grows, users will naturally segregate themselves ideologically. Conservatives will share articles in their social network, liberals in their social network, etc. Rather than relying on user tagging, we will use cluster analysis to identify groups that agree with each other. After reading an article, the site will then suggest ‘Read Next’ articles on the same topic written by people from different groups than the author of the original article.
This is the hardest part of the whole experiment. By limiting our foundation to only complete sets of official documents (all presidential debates, all census results, etc.), we are excluding important documents such as eyewitness video and peer-reviewed studies.
We're attempting to disrupt political punditry in the Christensen sense. There are many political debates in which it seems to not be possible to have a productive debate. We aren't distracted by that - we're focusing only on those issues for which a provably unbiased evidence database can be helpful.
The goal of our site is not to discuss every political topic – the goal is to discuss whatever can be discussed using the limited set of mutually-trusted evidence. We can safely expand the foundation only to the limits of mutual trust.
Depending on user feedback, we could introduce an "expanded foundation" later, which could include peer-reviewed studies, eyewitness footage, etc. We could use visual cues such as a sloppy font to indicate that an article relies on facts outside of the core. It will be very important to conduct this expansion gradually, transparently, and with user-governance. Because this expansion will be experimental, we will terminate and, if necessary, reverse the expansion if it begins to conflict with our core goal.
A potential future vision is to help fund local journalism. It would be critical that such journalism would be limited to fact-finding, leaving the analysis to ordinary citizens.
Notably absent from the foundation of facts is "the media". A huge amount of political discussion centers around the discussion itself. Our site's goal is to let individuals express their opinions on policy directly – full stop.
In order to accomplish our goal, we must be verifiably neutral. We must never ask a user to take our word for anything – they need to be capable of quickly verifying all of our claims themselves. That means several things:
Every transaction we make will be public. We will initially be funded by donations, and eventually plan to become advertising-supported. All financial documents, including bank statements, will be publicly accessible.
Our code must be open source, GPL. Our primary mission is running the website itself, but if anyone wants to ‘fork’ it and build their own community, they will be free to do so. We anticipate that a key feature – curation of the "Read Next" suggestions – will be hotly contested. It is important that the algorithm be completely transparent, especially the decision-making which will go into the design of the algorithm.
As mentioned above, the "Read Next" algorithm will be hotly contested. It will be impossible to analyze its design without also having all source data, which requires that our "user privacy policy" be essentially "none". We cannot provide transparency and privacy at the same time, and we choose transparency.
The people who build our site will obviously have political opinions. Pretending that they don't would not be transparent. ‘Dogfooding’, where one tests and uses one’s own product, is a valuable process of which we will take advantage. Rather than pretend that we don't have an opinion, we need to maintain a diverse team with varying opinions who want to help build this ‘DMZ’.