I think the biggest challenge is UI and user education (most people don't know what a fallacy is, let alone how many there are and how often they likely make them).
Then you further need to relate those back to the original statement by clarifying how different subjective target functions will have you come to a different final conclusion. (Not to mention that the "GPL is freer than BSD" or vice versa discussion is itself merely a sub-part of the larger "GPL is better than BSD (for my purpose)" or vice versa discussion.)
So my point is that it is possible in principle to provide a fully qualified link to a non-subjective presentation of arguments on that particular issue (and any other issue). I agree with you that the real challenge is UI, and probably moderation more than user education.
I've played around with UI ideas in the past. I think you need a nested structure, which can ideally be generated by the users (but would need to conform to the basic rules of logic [to whatever extent it has been defined]). Any argument will be in some knowledge domain, until you get to some "ultimate high level domain" (which probably doesn't exist -- not sure how you'd represent this in the UI; probably just the superset of all your as-yet-undomained arguments), so the nesting of domains should be reflected in the interface.
I think the UI problem is related to the data modeling problem. In my view, it seems like the UI should directly expose the data model. If that is the case, then the question is, "What is the right data model?".
For that, I think it's primarily directed graphs showing the relationship between the arguments. Of course there is more needed, as you'd need to annotate the arguments with various meta-arguments (eg. argument's conclusion is assumed as premise).
I do agree that one can probably come up with solutions for handling subjective arguments relatively well, I just think it might be difficult to come to a "this will handle all cases" solution. If someone says a particular argument is fallacious, why should we believe them? What happens when your database gets spammed with tons of bogus info? How do you make sure it stays out (or at least, can somehow be "excluded" from what you normally see)?
For "For software engineers, open-plan offices lead to greater productivity than individual offices" we don't need arguments but experiments and science.
If you had 5 studies regarding open-plan offices, you still need to check which ones are the most trustworthy, find out if and why they disagree, decide which outliers you ignore, etc.
I just wrote a comment about this, responding to global warming debate. Perhaps what this platform needs to address that is an ability to separate facts (that is, sources) and opinions (that is, synthesis or analysis of the sources).
IOW, I suggest such a system simply moves the problem from answering virtually the same question over and over to the problem of which level to ask a question at. Perhaps, in the name of progress, a step forward, or perhaps not.
None. For any area, there will exist numerous levels of perspectives such that people will argue over which level matters.
"This is basically about framing issues"
People will move to arguing which frame is relevant.
For instance, nothing would make me happier than to be able to reply to the nth discussion regarding the idea that "the GPL is freer than the BSD licence" with a universal fully-qualified link to every argument for and against the idea. Or "For software engineers, open-plan offices lead to greater productivity than individual offices".
While it may appear that this would lead to some sort of _Futurological Congress_-esque situation where we respond to people in paragraph numbers, it has many advantages:
* No longer will people be misled by a correct statement poorly argued for.
* No longer will message boards be polluted by the nth iteration of the same argument.
* Undiscovered lines of argument will be universally available.
Of course there's the disadvantage that you'll get less participation, and there's value in just having some number of comments even if they're repetitive: at the least, the desire to respond to that may bring people who later on make novel arguments.
This seems like a fine UI to do that. Deep link to the relevant sub-graph, and let the collective intelligence of thousands do your arguing for you. I like it.