It’s important to remember that bias isn’t evil. Everyone has biases and it doesn’t make us bad people. We don’t want to punish people for their biases, we just want to factor it out of objective content moderation.
Before we can do anything about bias, we need to measure it.
We want the ability to estimate the likelihood that any two users will agree (or disagree) with each other, even when we don’t know the topic at hand. Thankfully, social media is full of signals of agreement, the simplest of which is the “like”.
Podium implements this a little differently — both to maximise the value of the data we receive and undo some of the toxicity that evolves when “likes” are perceived as a “score”.
Instead of “liking” content, Podium users “react” to it — giving a nuanced response equating to a value between -1 and +1. This gives the bias map more information to work with while also allowing users to express strength of feeling.
You’ll notice a common feature in Podium’s design is our focus on quality over quantity.
Reactions get surfaced to other users as an aggregate score — so instead of 1.3k likes, you’ll see +70% support. This makes the overall volume of responses less relevant, without reductive measures like Twitter’s reported option of hiding likes entirely.
So, we now have an ocean of ±1 reactions. How do we use this to predict agreement?
Imagine we randomly scatter 100 people across a field and have each of them shout out their opinions at random. Whenever someone hears an opinion with which they agree, they take a step towards that person. Conversely, when they hear an opinion with which they disagree, they take a step away from that person.
After a while, people with similar views will group themselves together, and groups with opposing views will have pushed themselves apart. People with more nuanced views will be scattered between the groups, depending on how strongly they agree/disagree with each.
Note that the field itself is immaterial — if we ran the experiment again, people would group in the same way, but probably in wholly different locations in the field.
This is Bias Space and the location of each person represents their bias.
Whenever a Podium user reacts to some content, they are moving themselves towards or away from that content’s author. We can then measure our chance-of-agreement between two users as the distance that separates them in bias space.
In practice it gets more complex than that — Do we need more than 2 dimensions? How do we stop groups moving ever further away from each other? How do we stop people’s old opinions outweighing their current ones? How do we stop people faking their bias?
The answer to these (and more), will be covered in future articles. For now, we’ll focus on a more important question:
How do we get enough data?
The POD Token
A clear exploit in this system would be a for a bad actor to just never react to another post. We’d never measure their bias and so would be unable to factor it out.
So we need to require all users to reveal their bias, but in a way that doesn’t disrupt the experience or introduce different loopholes instead.
To do this, we created the POD token.
POD is the currency of Podium. You earn it by reacting to content and spend it to create content. This guarantees that we know everyone’s bias and comes with many other benefits as well.
It allows us to incentivize brevity in communication — but without the hard friction of the Twitter character limit. It allows us to introduce soft communication limits for mild infractions — by simply increasing the per-character cost for those users. And it gives us the means to incentivize other platform actions that are beneficial to the community.
POD has all the protections of any cryptographic currency (think Bitcoin), but with the key difference that it cannot be traded or sold (as that would rather defeat the point).
Podium does have a tradable token called Audium (or AUD), but that’s a post for another day. In the meantime, please:
- Check out our website at Podium-Network.com
- Follow our Twitter Trojan Horse at @ PodiumNetwork
- Give us your feedback in the comments below
We want you to set the agenda for these blogs — so help us decide where to focus first.
Podium is actively seeking angel and pre-seed investment to help build our MVP. You can start a conversation by emailing email@example.com.