In order to improve democracy, there are many ideas of how to combine direct and representative democracy (liquid democracy) and even, recently, augmented democracy using AI. However, it’s difficult to experiment with software which must guarantee that the votes are private and secure.
In this post, I’ll explain how we could divide the voting process into 2 steps where each step could be managed by different providers and each one would focus on completely different requisites. One of them, privacy and security and another, experimentation, engagement, etc.
On the one hand, 2nd-step voting providers would allow users to directly vote each decision like normal secret votings. However, it would also allow users to delegate to a web address (url) or voting feed. If this url or feed includes a position for a concrete proposal, and the user hasn’t directly voted that proposal in this 2nd-step provider, this user would vote as the url/feed indicates.
On the other hand, 1st-step voting providers would generate these url/feeds of positions. I don’t call them “votes” as in the 1st step because it’s not required that they are secret. I could have an account in a 1st-step provider and treat it like my Twitter account. Everyone could know that I manage that account and that what I state there is my opinion.
However, despite the fact that people know which one is my official account in a 1st-step provider, they couldn’t be sure that the url or feed of this account is the one I have configured in my 2nd-step voting provider. My votes and the url/feed in this 2nd-step provider must be secret to avoid coercion – and this means that not even the administrators of the voting system can know my direct votes or url/feed.
Moreover, although all 1st-step providers must adhere to a minimum standard to return positions given proposals, there is a lot of flexibility to create functionality on top. For example, a 1st-step provider could be focused on liquid democracy and implement features such as multiple representatives. All this could be done within a single 1st-step provider or spread across multiple ones. Therefore, 1st-step providers could have url/feeds from other 1st-step providers as inputs. Thus, Malala, Bill Gates, a friend of mine and I could use different 1st-step providers and I could still configure that if I don’t directly state my opinion on a topic where either of them have expressed their position, I want my position for this topic to be what any of these other people I trust say. If they state different positions on a proposal, I would like to be notified to directly state my opinion. Another option for this 1st-step provider could be implementing hierarchical delegation so I could define the order in which I would like to follow the opinions. E.g. if my friend shares her position, I’d like to copy her position. If not, Malala’s. If Malala doesn’t share her opinion either, then Bill Gates’. All these different options could be implemented within the same provider or across multiple ones.
Similarly, 1st-step providers could crowdsource opinions from well-known people. This would be similar to what my side-project Agreelist does for topics such as basic income or whether “AI will pose a risk to society in 50 years” Then, Agreelist, or another third party, could also crawl the web to get all texts written or spoken by people, use natural language processing and predict whether they are positioned in favour or against. This could also be used as a url/feed for a 2nd-step voting or as an input for another 1st-step voting, in case I’d like to delegate to Malala and Bill Gates’s crowdsourced opinions. That way people could delegate in Malala or the Pope even if they haven’t heard about any of these 1st-step providers.
Another possibility for a 1st-step provider could be to focus on creating “digital twins” as the MIT professor Cesar Hidalgo describes in his TED talk “Augmented democracy“. In this case, an AI would predict what you think. If the confidence of your digital twin is high enough on a given issue, it would return a position. But, of course, if you disagree with your digital twin, you could always override the position in this 1st-step provider or in your secret vote in your 2nd-step provider.
To sum up, there are many possibilities to explore in liquid and augmented democracy and splitting voting systems into 2-steps with a small common standard could allow different providers to experiment and focus on completely different requisites.
Categories: Inspiring ideas
Hector Perez Arenas 