Tool to be an informed voter

Commenting on The Future of Voting from Dilbert Blog, I'd like to talk about Scott's suggestion for making voters more informed. He suggested a website that summarized arguments of issues. The summarization is structured so that points and counterpoints can be added easily with links to proper sources or evidences. He also suggested a ranking system so that the better arguments are more visible.

I have thought about creating a website implementing such tool ever since I have difficulty following arguments in a mailing list about evolution. Too bad I don't have that much leisure time anymore to play with various projects. But I did find some things I'd like to share:

1. Many people already thought about such tool, one even show a prototype on issue of abortion. But I cannot find it anymore, sorry..

2. One major problem with the existing tools is how to assimilate all existing information in a glance, or even assimilating only relevant information in a glance.

3. Presenting relevant information is difficult. Because stating/coding the request of we consider relevant, is complicated. It's complicated because either:
a. we must know the structure of relevancy in the issue, which automatically means that we already know a great deal about relationship between arguments on this issue, which defeat the purpose of having this tool in the first place.
b. we give relevant keyword or state the relationship between relevant keyword and pieces of arguments we're trying to retrieve. But this result might not be what the coder intended.

4. The problem of point 2 is also related to difficulty in coding arguments into the tool. Relationship between various points and arguments must be coded well to increase tool performance in displaying relevant arguments, and it's difficult to code it well. Additionaly, increasing complexity of arguments will greatly increase coding effort required, because we need to verify that the new argument:
a. has not been coded yet.
b. does not create inconsistency.

5. The study of knowledge representation is useful, but not that useful for this project. It's not that useful because many study of practical use in knowledge representation are on implementation of computer's knowledge representation only. Study on human's knowledge representation - which might give hints on how to make coding information into tool easy or how to make retrieving information from tool easy - is still limited.

btw, for point 3b, I think it's plausible to use google search result to generate list of related relevancy from relevant keyword we entered, where the list will be used to find the relevant arguments instead of using the keyword we entered. Or we can just google the tool's database (I'm serious).

Or we can just google the internet since google try to optimize it's search engine to analyze websites in the world (read: the aggregate of all human knowledge) to generate relevant result. (I'm also not joking here)

In anyway, I bet people working for google know a lot about this things. It's my dream to work for google. :-)