The American people are truly fed up with politics. There is constant arguing about the direction of the country, and very little space for forward progress to be made. It is commonly believed that voters in this country have nothing in common, and that reaching agreement is impossible. We are here to prove otherwise.
On other social media, nothing gets decided. There is constant arguing, but no conclusions are ever reached. Social media websites are not built to reach consensus, instead they prioritize engagement and outrage.
Pollsters usually focus on a small handful of popular issues, and rarely measure people's opinions in depth. Nuance and detail are compressed into yes-or-no, without regard to the complexities of the issue. "Do you support abortion?" is not a meaningful question: Very often people's opinion is some variant of "sometimes" or "it depends." There are too many possible opinions to have to compress into simple answers. How should regulations around second trimester abortions be different than the first? No one asks.
Politicians are supposed to negotiate for us, representing their constituent's interests and finding a reasonable compromise on their behalf, but they don't. Politicians instead prefer gridlock, bad faith negotiation, and using blind opposition to convert everything into bargaining chips. They are not required to represent their voters at all. The laws that do get passed are often prewritten by lobbyists, and presented directly to lawmakers for support. The system is wildly broken, and it has gotten to the point where voters increasingly prefer a dictator to cut through the problems.
This would be a disaster. Democracy is the best form of government.
Here's how we fix it: We, the people, negotiate those compromises for them.
Agreed Upon Solutions is an experimental social polling site, engineered to solve the problems of social networks, pollsters, and traditional political representation. We have designed what we call the twothirds system, an attempt to apply the modern mathematical understanding of fault-tolerant systems to politics. We envision a lightweight, expressive, and comprehensive citizen democracy running on top of the pre-existing government. In our implementation, we are not an official government actor - We cannot make laws, but we are able to run the system and see the laws we would be determining if we were. If politicians need their laws prewritten for them, we can do that.
We believe that by systemically searching for agreement on the widest possible range of issues, high consensus areas can be found. You don't have to believe us, we've gone and done it: You can check out the Open Discussion on our homepage to see our proof of concept in action. Our undirected search has discovered extremely high agreeability positions on several rarely polled issues like "plastic pollution," "food waste," and "unwanted advertising," as well as a few silly things: People generally believe big burgers should be wide, not tall.
The same mechanisms can pick through more contentious topics, like climate and health care, in exactly the same way. We're running discussions on Every Thing: over 157,000 topics extracted from Wikidata. All we need now is to reach more people.
We want to build the best possible voting system, run the largest and most comprehensive democratic discussion in the world, and put together the most complete record of popular consensus ever assembled. Come vote on our site, and we'll handle making the world better for you.
We have developed an algorithm we call the twothirds system, which seeks to find opinions with high consensus. The core idea is that a clear finding in a "reliable enough" online poll can be used to predict real world majority support. By making our threshold higher (67% instead of 51%), our strategy bakes in robustness to factors distorting the vote - Bots don't need 1% to swing the vote, they need 33%. This threshold comes from the mathematics of Byzantine Fault Tolerance - It balances bot & troll resistance with the ability for unanimous votes to always make forward progress. You can see more about exactly how it works here.
We've pulled over 157,000 common nouns from Wikidata, the most comprehensive database of things in existence. If Wikipedia knows about it, it's on this list (or it's a very specific chemical, disease, protein, fruit fly gene, or handful of other things with no conceivable political relevance). We've filtered out the slogans, celebrities, and political bias, leaving only the meat of the topic - We're discussing infrastructure, not Build Back Better; billionaires, not Jeff Bezos. This list is comprehensive and extremely specific - We have separate topics for "immigration," "illegal migration," "immigration officers," "sanctuary cities," "border checkpoints," and more.
Every topic is given an explicit definition, helping ensure everyone is talking about the same thing. What does "socialism" mean? Socialism is a "system of government where the means of production are socially owned," not a totalitarian movement to ban private property. (That would be "all things in common," also discussable on the site!)
We've implemented quite a few measures already to prevent bots and spam, and we have more on the way.
All this is just the beginning. We know fighting robots is a constant battle, but we also have a pretty good idea of how to fight them.
Every political organization needs a set of colors, but we felt like the usual choices are too limiting. We aren't red or blue, or green or yellow or purple. We're organizers and summarizers, and we take from everywhere to find a baseline almost everyone can support. We think 90s geometric patterns summarize that perfectly. It's fun, and color is an important part of our world. Our commitment to color extends throughout the entire site, from our comment colors to our visualizations. It's important to us to have a personality.
We've also attempted to build a page that avoids the overengineered design of the modern internet, building our page with simple web technologies for more efficient pages. We also allow you to customize the look and feel of the site, including light\dark modes & springy\solid scroll physics. Check out the account settings page, there's a lot there!
We're here to be the calm eye of the political storm, not another data broker selling a mailing list. We use your information strictly for finding consensus. We've done the following things:
On most social networks, posts are rewarded by going viral. While this is great for engagement, it is terrible for finding out what people support. Instead, you get whatever is most reaction provoking, which is just as likely to be a terrible hot take, an irritating troll, or callout dogpiling. You are given no sense of the average person, only whatever happens to be engaging to a particular social circle.
We take the opposite approach. Our goal is to examine as wide a range of opinions as possible. Comment weights are capped, ensuring there is no way to manipulate a comment to the top of everyone's feed. Getting your voice heard is no longer a matter of how many followers you have, all that matters is how much you contribute to the consensus. This is bad for engagement, but that's ok. We don't want to monopolize your attention, all we ask is five minutes a day.
We believe that for democracy to exist, you must be able to discuss any topic. There were times in history where things now seen as fundamental rights were considered undiscussable. These include interracial marriage, decriminalization of homosexuality, and even the idea of democracy itself. This list is not just limited to "unpopular liberal ideas," there are also reasonable objections to policies like affirmative action.
It's true there will always be some people who want horrible things. But their voice is a genuine human voice, coming from someone who lived a life and came to a conclusion. They should be allowed to express themselves, even if wildly disagreed with - A loss by the ballot box is fair, a loss by moderator takedown is not. We want our site to be a total democracy, so how do we handle them?
The twothirds system is designed to handle this problem. Because of the twothirds threshold, opposition from minority votes is doubled: they need to be outvoted 2-to-1. Anyone attempting to reach majority approval has a much harder problem. Proposals that meet with nearly universal disapproval are filtered out quickly, usually with only a handful of votes. Their reach is self-limiting, because the more people they encounter, the more quickly they are downgraded. The twothirds system is surprisingly aggressive: even ideas with weak majority support will eventually be filtered out!
We've been testing this idea using the spam comments that appear in the Open Discussion. So far, it has worked well: Keysmashes lose agreeability quickly, reasonable non-English comments become neutral instead of highly disagreeable, and off-topic comments (a large portion, by design) are generally filtered downwards. The political items with >90% agreeability have generally been well received in offline discussions, and those reaching 100% have almost always gotten extremely positive reactions. We believe the system works.
The twothirds system is designed to handle a lot of crap. Because of the wide margin for error, as long as we can verify voters are human "well enough," we don't need to deal with exceptionally complicated forms of identification. This means simple online voting becomes possible! The results of the polls as predictors of simple majority opinion, all that is officially needed for real world political change, remain valid.
We're not limited to the same dozen or so issues being polled by the major surveying companies. Our top 1% list includes "animal consciousness," "video game monetization," "sperm theft," "the eradication of suffering," and more. You can give your answers as freeform text, with as much or as little detail as you like. If you want to say something, but don't know what to talk about, we'll let you browse the entire list. If you want to make your opinion heard, no matter how strange or obscure, you can.
Furthermore, if your opinion is unpopular, you will not be canceled for it. There is no harassment because no personal information about you is shown, you have no inbox to flood, and direct responses are not possible. Responses to comments (not topics) are limited to Yes, No, or Pass. It makes voting as quick and easy as possible, while preventing mobs from forming against anyone.
Everyone has opinions, including us. In fact, we are political zealots of a certain type: you need to be, in order to build a project like this. What sets us apart is that our politics are outcome-neutral. We believe democracy is like driving a car. If you decide to turn down the wrong road, the car should do it. It is not the job of the steering wheel to second guess the driver. We believe the people should steer their government, and our job is to accurately determine where they want to go.
The tyranny of the majority is a real problem, but the premise of democracy is that an overwhelming enough vote should become law. The twothirds system is designed to offer stronger built-in protections for minority rights than any other voting system currently in existence. As a result of its increased threshold, the twothirds system weights minority voices twice as heavily as those in the majority. If being outvoted 2-to-1 isn't good enough to win, then what is? Eventually, you are no longer a democracy.
But let's say you did want to raise the threshold much higher, say to 99% of voters. This actually makes the problem worse! Now, any progress for minorities can be blocked by a group of 1% - You've given control to an even smaller group! The twothirds threshold is mathematically optimal, it is the point at which making and preventing forward progress have the same difficulty.
This does not completely solve the problem of the "tyranny of the majority," but this is only a surface level analysis of the problem. Our perspective is influenced by mathematical advances made since the 1700s, and coming to group consensus in the presence of malicious or unreliable actors is now a well-studied problem. A longer analysis is available on our website, but this kind of thinking is the key to the modern perspective - all groups are political, and movement in all directions must be balanced.
Politicians are not required to do anything for their voters. An elected representative is able to totally betray their electorate, with no consequences until an election years away. It happens depressingly often, lobbyists and the pressures of party politics often completely override the will of their voters. This leads to a situation where the only way voters can influence policy is through increasingly extremist positions. They elect extremists to represent them, which in turn saves merely bad politicians, who can now win elections without meaningful policy. The only issue becomes whether or not their opponent is "the greater of two evils." This makes all the previous problems worse, and the situation spirals.
In the twothirds system, your opinion matters; and when people's opinions matter, it helps bring them back to Earth. There is no incentive to be an unhinged lunatic, because they are not trying to change an unaccountable human person's mind. By allowing people to vote for their true and honest opinions, we can reduce extremism and bring politics back into the realm of reasonable.
It's the early days for us, and we have lots of things we want to improve, from features to statistics to usability. We believe in learning by doing - While we think what we have is good, we're a long way from being the Best Possible Voting System. Here's what's on our TODO list:
Our dream for the future of democracy is a world where making a change is about convincing other people, not deciding the winner of a popularity contest. It is within reach.