Jabble is a new kind of conversation platform, designed to make conversations on the Internet more intelligent and less polarized.
Read how Jabble makes conversations better below. And signup here to get notified when we Launch
Like many online social platforms, in Jabble users post things and other users can reply. Here's an example:
a post with a reply
Each post or reply then gets a score, shown as a percentage, based on how many upvotes and downvotes it receives.
a reply with votes
However, not all votes count equally! Two factors affect the strength of your votes:
-
You must earn reputation before your votes have any weight.
-
Your votes on a post get more weight if you read and acknowledge certain replies to that post made by other users.
When certain conditions are met, a reply to a post will be designated a critical reply. The critical reply will be indicated by a green dot.
a critical reply
If you have have voted on a post, but have not yet seen the critical reply, you will be notified.
Once you have voted on the critical reply, your vote on the post will get more weight.
a critical reply after vote
And of course, if the critical reply changes your mind about the post, you can change your vote on the post.
The idea is simple. If one user shares information that might influence other people's votes, then people should have a chance to consider that information before they decide how to vote.
Clickbait, disinformation, misinformation, etc. are all fundamentally the same thing: misinformed content. Misinformed content is stuff people wouldn't click on, upvote or share if they knew more.
The antidote to misinformed content is information. Jabble tries to promote more informed content based on the simple idea of looking at how people vote if they have more information.
Jabble identifies a reply as critical if it appears to be convincing: if it makes people more or less like to vote on the post.
Suppose that, in our example with the inauthentic earthquake video, initially 95 out of 100 users upvote the video. But suppose there were 20 users voted on the reply claiming the video was not what it claimed to be, and only 1 of of those upvoted the video. Since the reply appears to be very convincing, it is designated a critical reply.
But the conversation is not finished! Jabble informs users who upvoted the video and gives them a chance to change their vote or to respond. Maybe the video isn't fake after all! If anyone makes a convincing reply to the reply, then that reply will be designated as the critical comment, and the process will continue.
The thread of the most convincing replies, the most convincing reply to that reply, etc. is called the critical thread.
a critical thread
It might seem like this process could go on forever, but it probably won't. As we hope to demonstrate with early Jabble users, we expect most discussions will converge after the strongest arguments have been made on each side of an issue and nobody can find anything to say that further changes minds.
The votes of users who have voted on the critical thread are called the informed votes. The score on the post is just the probability that an informed vote is an upvote. We call this probability the informed upvote probability.
In our example with the earthquake video, informed users are much less likely to upvote the video, and since informed votes have more weight, as users start to vote on the critical reply the estimated informed upvoted probability falls quickly towards 5%.
the informed vs. uniformed upvote probability in a simulated scenario
This process works great if everyone is acting in good faith. But on the Internet, sometimes they're not. So how do we force people to be reasonable?
Well, it turns out you kind of can force people to be reasonable using game theory.
Every user gets a reputation, starting with 0. Until a user gains reputation, their vote has no weight.
Users gain reputation according to an ingenious scoring mechanism from MIT called the Bayesian Truth Serum. The formula is designed so that a user's expected score is maximized if they vote honestly -- as long as everyone else votes honestly. This is brilliant because, if people expect each other to vote honestly, then an equilibrium is established where users will continue to vote and expect others to vote honestly. This continues until users can get together and coordinate on voting dishonestly, which is unlikely to be a problem as we [write about here]. This kind of equilibrium at honesty is similar to the magic that powers blockchains, as we [write about here].
Now, Jabble only considers informed votes when calculating the BTS scores. This means that people earn reputation by voting honesty given the comments in the critical thread. Users most honestly consider the information in the critical thread, and decide whether they think that other people in this forum would honestly upvote it or downvote it given this information.
If the evidence proving the video is fake is strong enough, then not only will most everybody believe that the video is fake -- they will also expect that all the other voters who saw the evidence to believe the video is fake. So as long as there is an equilibrium at honesty, everyone will downvote the video.
So if there is an equilibrium at people voting honestly given the information and arguments provided, we can say there is an equilibrium at being reasonable.
In Jabble, you contribute to a conversation not so much by what you say, but what you vote for. Posts themselves are anonymous, and the karma earned by a post is divided among all the people who upvoted the post, with earlier upvoters receiving a larger share of karma than later upvoters. The author of the post themselves, being the first upvoter, receives the largest share.
But don't try to harvest karma just by upvoting everything. You also have a reputation score within each community. Before your votes have any weight, you need to build reputation by upvoting content that ultimately earns a high score from existing members of the community (or downvoting low-scoring content). And if you upvote too many things that receive a low score (or vice versa), you will lose reputation in that community.
On the other hand, you may gain reputation with the same content in a different community! So when upvoting or downvoting, consider the community. By upvoting, you are recommending content to that community, and staking your reputation on the belief not only that it will be popular in that community, but that it will stand up to scrutiny: that it is not fake, or misleading, and thus will still have a high score among users who saw the the critical comment thread.
If you are not sure, you can also just upvote content you like, without specifying a community, and Jabble will find the community for you! This is great for people with a variety of interests. You can upvote content related to your profession, as well as cute pet pics, without worrying about wasting the time and attention of people who aren't interested in one or the other.
Reputation has a cap, but you can earn unlimited Karma. Reputation is an estimation of the score a post will receive within a community given you upvote it. Karma is your total contribution to the community.
You can maintain a high reputation without accumulating much Karma by upvoting high-quality content, but infrequently. Or you can have a fairly low reputation and still accumulate a lot of Karma by upvoting a lot of mediocre content. But you earn Karma faster if you have a higher reputation.
A final piece of Jabble's scoring formula is bridging.
TODO: describe bridging and summarize.
A final piece of the Jabble process is the prediction market. At any time, you can make a prediction about what the final score of a post will be. Accurate predictions further boost your reputation. Predictions are a part of the Bayesian Truth Serum scoring mechanism.
TODO: describe prediction market