Recommendation engines for proposing issues/delegates

[papadako]

Another interesting idea revolves around the creation of a recommendation engine that can stimulate political participation and collaboration in the vast amount of information (e.g. issues/delegates) that a democracy in scale would have . Such a system could be useful for making citizens aware of issues they are interested in or for proposing delegates. The important think here though is to do this in a transparent way, in order to avoid the introduction of any kind of bias, which I guess is not an easy problem to solve.

There is a PhD referenced below that proposes a fuzzy based recommendation engine that can assist: "In the area of eCollaboration, the platform can be used by governments or private sectors to find citizens interested in taking part in various projects based on their profiles. In the area of eDemocracy, the platform can be used to monitor, evaluate and provide relevant information on different political actors. In the area of eCommunity, the platform could provide tools for creating virtual communities"

SmartParticipation: A Fuzzy-Based Recommender System for Political Community-Building

https://doc.rero.ch/record/209654/files/TeranL.pdf

[bhack]

See also the already mentioned https://github.com/conversationai/perspectiveapi/issues/2

[metamerman]

Unfortunately, if you think about it, what you're proposing here is exactly the opposite of what a properly engineered decisionmaking system would include. The best decisions (most efficient, most fair, most stable, etc.) are those made by unbiased participants, but all your proposal does is ensure that decisions will be made by the most biased individuals. It's a recipe for "tyranny of the minority" where a very small subset of the population, and unfortunately those with the most extreme positions and indeed only the extremists on one side of the issue who have even slightly more power than their opposition, end up getting their way to the detriment of everyone else (i.e., it'll devolve into exactly the same kind of oligarchies we've had all along). Similarly, and although you didn't use the term, it should a red flag any time you hear someone proposes getting the "stakeholders" together to participate in a making a decision for exactly this reason: The quality of the decision is inversely related to the overall level of bias in the participants. The best way to improve the quality, therefore, is to facilitate the participation of moderates.

Proxyfor.me (https://www.proxyfor.me/) handles this problem by automatically matching people to delegates/proxies using general personality characteristics rather than by bias on any particular issue. It's also been proposed to correct for this kind of bias by randomly assigning people to issues (sortition) as in "citizen juries", but unfortunately that just introduces a different bias problem: Anyone who knows what they're doing can easily get out of this kind of jury duty which unfortunately means that the juries end up being composed of the least intelligent and most authoritarian individuals. There may be other ways to matching people to issues or delegates, but the bottom line is that including a "recommendation engine" that is specifically designed to detect and then utilize bias, whether it's for debate or delegate/proxy selection, would just be a design flaw in a public policy decisionmaking system. If anything the system should probably do the opposite of what you propose. That is, it might detect bias and then use this information to exclude those individuals from being participants or being selected as delegates...