When you grow up reading and listening to one side in a political discussion, you tend to adopt it.
Then, when you later encounter persuasive individuals from the other side, the temptation is to assume that there is a case for that side, and to think that the best thing is to split the difference and adopt a compromise between the goals of the two sides.
In constrast, I'd say both sides (in fact almost any political position which has survived and retains supporters) have rich and sophisticated models of the world. And the strengths and weaknesses of these models need to be understood. Then a subtle synthesis is needed of the important insights from each. Such a synthesis may be very different from a naive horse-trading between the different groups.
I've been thinking along these lines for a while, too, although I think in order to use comparison and integration of models to come to some agreement, far more work would need to be done by the proponents to shift their approach from goals to logic, as it were.
What I'd like to look into, when having the chance, is how to model people's arguments in a formal manner. I'm of the impression that, in a more abstract manner, people share the same domain of facts (or reality) within a particular discussion but, either due to ignorance or polemic blindness, each party follows a particular subset of this domain in order to reach, or support their goal.
So, the difference between the two parties is a). their view of the model, which could be expanded (leading to more civilised, progressive discussions?) through a formal model, and b). their choice of goal, which is also partly influenced by factors outside of the model of discussion, i.e. by past and personal experiences, although it may be that, as these experiences are a factor in the decision, they could/should be worked into the formal model.
Is there anything like this out there that you know of? So far, I'm thinking it would involve a linked hierarchical categorisation of arguments, as it were. Each point of discussion depends on one or many other points, or may be thought of as a collection of points. (Thinking a cross between XML-style nesting, and RDB-style pointering.) It might not solve the problem of people choosing different points of view from each other - after all, that's what makes it so interesting ;) - but it'd produce some pretty pictures...
: There is something called "Six ThinkingHats" - a method by E. De Bono for categorising arguments. I don't think it is very practical, but I feel it is worth trying. I have not explored yet this link - but it seems promising: -- http://c2.com/cgi/wiki?SixThinkingHats -- ZbigniewLukasiak
: Trying to capture the large scale structure of arguments is one of the motivations of TypedThreadedDiscussion. Links there to PhilosophicalMarkup and similar attempts to formally capture argument structure. -- PhilJones