It's written tools to let participants answer open questions, it's read tools to let citizens decide

Modern Democracy is based upon the american and the french revolution. From there came the idea of representative democracy, as well as the research on voting theory. After all modern states were too big to all fit in a room, communication too slow to let everybody participate, so it was unavoidable that few people needed to decide for the others. And if this was the case, voting theory tried to solve the problem of finding the most fair way to chose among a set of given alternatives the best solution, based on the preferences of a number of participants. Although this might sound very democratic it has a big drawback. Who chooses what are the possible alternatives?

This is not just a theorethical sterile question. Especially if you consider that a common criticism given toward modern democracy is that our choises when we vote are among parties that are more and more similar to each other. A new form of decision making is necessary. Something that let people not just chose among a series of alternatives, but also insert new solution themselves. And build up, upon existing solutions, to create something truly innovative.

In other terms we need to move from fair decision making among given alternatives, to fair decision making over an open question.

This is not a simple transition, as the tools are all but built, and the theory behind it hasn’t yet been studied. I have tried to frame the question in a previous talk.

But the research we developed was not just trying to frame the question, but also trying to answer it. For this we developed two different tools. Both work with open questions, and both are being hosted at The name, Vilfredo, being a recognition to Vilfredo Pareto, the Italian Economist who developed the Pareto Front. A mathematical tool used extensively on

The two tools developed are somehow complementary, and although they don’t cover the whole range of the necessities, they do cover quite a bit. Both work on open questions. The first tool is Vilfredo Genetic Algorithm. It is a Human Based Genetic Algorithm where users go through a series of round, In each round they alternatively read a selection of the proposals of previous generation, present new proposals, and vote on the proposals made from others. The system then use the votes of the participants to chose the new selection of proposals in a way to be both exclusive (trying to eliminate as many proposals as possible), and inclusive (making sure each participant point of view is represented in the selection). The result is a series of further approximation where different ideas are investigated in parallel, integrated, voted, and finally when (or if) a comprehensive proposal is presented approved by all. But experience showed that this process worked well with some questions and not others, and if the group of people participating are of a certain size (and not too big or too small).

The second tool is usually referred to Vilfredo Bubbles. In this tool users are invited to present and vote proposals that are then presented as bubbles in a sort of tank. Bubbles enter from the left and move to the right. Bubbles above are considered better than bubbles below. And bubbles start always from the top row. But each new bubble that arrives makes it harder for the old bubbles not to drop. Eventually the highest row will always contain the best and newest proposals. Most similar system tend to show either the newest or the most votes proposals. With the unavoidable result that eventually proposals in the “most voted” list will have gathered so many votes to make it impossible for new proposals to ever enter into that list. Practically closing the system to new proposals.

As mentioned before the two system seem to work on mutually exclusive systems. Vilfredo bubble seem to work well for “what” questions. Situations where an open questions is fully open to receive many unrelated proposals. And more than one can in parallel be implemented. “What shall we do about the economy/immigration/education”. All those are what questions. And all are open to different answer, which often can be implemented in parallel. Instead Vilfredo Genetic Algorithm tends to be more useful for “how” questions. “How shall we build this website?”, “how shall we organise the edgeryders conference?”. While “what” questions are open to multiple indipendent answers, the “how” questions usually are best answered by a single answer. Multiple answers need to have their differences smoothed out to reach an integration that is valid for all the participants.

Also while Vilfredo Genetic Algorithm tend to work best for small group of people (the limit being that each participant MUST reas every proposal, and the filter process become inefficient when too many point of view are considered). Vilfredo Bubbles can be used by any number of people.

As we explained before those are tools. But then the importance is how can those be used in e-Government, and e-Democracy. Vilfredo bubbles was integrated as a Facebook application, and was then used to let people decide how to gather the sorted garbage in Trastevere. An authentically impossible task for one of the most turistic district in Rome. A place with small streets, many turists (that thus will have to figure out the rules), restaurants that remain open until late, and a living community that seem to always be getting more annoyed with the situation.

You can see the proposals presented here. You will have to register in the Vilfredo application to do so.

Hot topic!

I am quite comfortable with online open collaborative environments to explore and organize information related to governance: this is, after all, what Edgeryders is doing. Online communities are very good at that. Decision making, however, is a totally different ballgame. Despite your and others’ excellent work in the area, I feel these tools are still very much in the testing phase, not quite ripe for large-scale rollout on real governance decisions.

Having said that, this is clearly a need that humanity feels is urgent right now, at least in the West. The Pirate Party in Germany earned a lot of attention because of its use of some kind of algorithmic decision making tool for its internal democracy - and let’s face it, Occupy’s 1970s -style assembly are just not a good tool, they don’t scale, they are too time-intensive and (at least to me) mindnumbingly boring.

Have you looked into the Pirates’ Liquid Democracy tool? What do you think of it? What is its mathematical basis?

Liquid feedback: good but not good enough

Dear Alberto,

I think you are absolutely correct both in saying that online collaborative decision making and online collborative gathering of information are different, and in saying that online collaborative decision making is still not ready.

I see the relation between those fields as the relation between a map and the track a person follows on it. One will present you with all the possibilities, the other says, “right we shall go here, passing through here, here and here”.

For information gathering I feel wiki as still being mostly unsurpassed. Even with all their shortcomings when trying to integrate information among people coming from very different experiences.

Speaking about the discussion tool itself, the problem is that we have different principles, and it is not possible to honor them all at the same time. We also have some unspoken assumptions, and some problems.

Among the unspoken assumptions I see:

  1. There more (and often better) ideas in the wider community.
  2. if we integrate different ideas the result will generally have a wider support.
Among the problems the biggest is that many topics are too complex for most people to participate in. Although most people are experts in something, no one is expert is everything. So if we want to create a system that permit to anyone to participate, will this not exclude the people who do not understand the issue? Or who understand in general the issue, but are not technically advanced enough to participate in the discussion?

This is a serious question, and the answer to this question defines the type of system you will have.

This have several connected issues: (1) if only some people can, meaningfully, participate in a discussion would this not impoverish it? (2) Would the result still be representative of the will of the majority of the people if only the people who are able to understand it participate in it?

Liquid feedback position is that it is important that everybody participates. So when they have not the technical knowledge to do it they are to chose a person to represent them. But the representation is not limitless. In any moment they can change their mind and ask someone else to represent for them from now on.

The positive element of this is obviously the fact that everybody can participate.

There are many problems though.

  1. if people vote for a person, people with charisma will have a high chance of reaching the end. Actors, singers, (accordion :wink: ) musicians, comedians, anyone with a wide public profile will be favorited at this game.

  2. people can ask a person to make changes to his proposal. But if he does not want to do it, they can only start a new proposal. Having to start from zero to gather votes. (I just discovered this recently). In short a proposal is in some sense owned by the persone who starts it. It strike me as ironic that this is the piraten party system ;-).

  3. although people can delegate their power, and recover it in any moment, many people will probably not do it. (we) People tend to be quite lazy, especially on topics that fall out of our direct area of interest or expertise. So until a person does something really bad, he might still have the support from a lot of people who are just too uninterested to bother to look for a better candidate.

But what are the alternatives?

There are many possible alternatives. Let me list some:

  1. ignore the problem (so let only the people who the topic participate) at the early stage where people write and integrate all the possible ideas. Then once a final set of ideas are presented that cannot be further integrated are defined, everybody should vote on them (using voting theory, Shulz system like PP is ok, but others are also possible). Although the people might not be able to understand that inner workings of each of them, it is possible to add to each idea a small video presentation that explains it in a non technical language. If we accept that the vote of the people who do not understand is given out of personal favoritism for the person, and not out of technical analysis, we might consider it just as a form of noise in the system.

  2. [and this one is my favorite] use internet to educate the public, while explaining them the ideas. There will still be people who are not interested to participate in the discussion, but maybe it is possible to permit to all those who are willing to do a little effort to participate.

By asking people not just to rate how much they like an idea, but also to rate how much do they understand it, we can gather important data that can then be automatically used by the system. Here is how: ( I presented this idea -developed with Cyril Velikanov- at the ePart 2011 conference).

With this information would then be possible not just to see the most supported ideas, but also the clearer ones. Not only but it will also be possible to see who are the people able to write in a more comprehensive way something. With the possibility for the system to ask them to rewrite an idea making it easier to understand for the others. Alberto, you remember how in Vilfredo the system is able to find out which proposal you should work on to better help the convergence process? Well here the system could do something similar, albeit simpler. Just asking people to rewrite the best ideas (to fully understand this idea, please see the video above). At the end the final ideas would could (and should) still be evaluated using a general closed voting procedure. But the participation on the planning stage would have been much wider without requesting any representation procedure. Also as people participate in the process they would become more and more aware of the different issues of their society. Not a bad side effect.

This idea was part of a more general system that was presented last year.

So, in brief, what I think of liquid democracy? It is surely a step in the right direction. But I think it is really not using the internet at its best. We can do much much better. And I am afraid if the tool raises too fast we might find out the problems once it is too late. What we need is that those tools should be used, and tested online by many communities.