There is only one possible way to solve this problem that has a shot at working and I'm developing/testing it as quickly as I can.
It requires a combination of crowdsourcing and machine learning (NLP is cool, but optional).
It is part rating, voting, tagging, reputation, recommendation and more.
It sounds complicated, it isnt for users (~2 clicks from ~10% of readers) but is for me to build (3rd week with rails, so relative).
It is mostly still theory at this point but the math checks out and user tests have exceeded expectations but are very early stage.
If you've read this and think I'm crazy, you're right (best to accept it).
If you think I'm naive, it's a possibility that's kept me up many nights over the past year.
If you are interested in learning more on the off chance I'm right, you can reach me at (Mat . Tyndall at gmail).
I am in San Francisco working on this full time after dropping out from grad school.
I will tell you in excruciating detail in person or via Skype how it works and why.
I have not slept very much in the past week so I have no idea if posting this is a good idea but I doubt I'll get even 5 votes or any takers but thanks for reading.
I must admit you've made me a bit curious, though I have a feeling that whatever it is you're proposing will probably just go straight over my head anyway.
Since you're in SF though, have you thought about holding a small group discussion about it over at noisebridge sometime? People there are always so insightful, friendly and willing to help, I'm thinking that holding a think-tank at a place like that might help you get the sort of 'back-and-forth' feedback it seems you're looking for.
As for the prototype going over your head, first order it looks pretty simple, second, third and fourth, ... Hey, even I'm not sure I've mapped out all the possible dynamics but the possibility is unquestionably there.
Noisebridge is cool, hung out there a few times. I'm super busy now but maybe in a week or two though I am a sucker for coffee chats (hint hint). Basically I suck at writing and would rather connect through any other medium.
Also to add a few details, I'm doing this as a startup (Tagbax, don't mind the landing page, I'm really not trying to drive traffic yet). Community comment moderation isn't necessarily my first choice to apply the tech $ wise (I really ought to drop by disqus one of these days), but it was designed specifically for user-generated content.
Before you lose any more nights losing sleep about this idea you have and its third or fourth iteration, maybe you should just release the idea in its original form and see how people react. I'm sure your idea is great, but I worry that it's a perfect solution to an imperfect userbase...meaning people value worthless, soundbite comments more than well thought out responses.
It's kinda like how people complain about how politics are only about soundbites. Sure, it's great to argue that among friends, but who really spends the time understanding the facts about every issue a candidate debates?
The solution is to align the interest of the individual with the group. I've run this past hundreds of people, it's pretty solid as theory but coefficients matter so it's gotta be tested.
Consider another, simpler way to solve the problem: http://news.ycombinator.com/item?id=3473753 . Self-organizing, no extra steps required, and mostly transparent to the users.
It requires a combination of crowdsourcing and machine learning (NLP is cool, but optional).
It is part rating, voting, tagging, reputation, recommendation and more.
It sounds complicated, it isnt for users (~2 clicks from ~10% of readers) but is for me to build (3rd week with rails, so relative).
It is mostly still theory at this point but the math checks out and user tests have exceeded expectations but are very early stage.
If you've read this and think I'm crazy, you're right (best to accept it).
If you think I'm naive, it's a possibility that's kept me up many nights over the past year.
If you are interested in learning more on the off chance I'm right, you can reach me at (Mat . Tyndall at gmail).
I am in San Francisco working on this full time after dropping out from grad school.
I will tell you in excruciating detail in person or via Skype how it works and why.
I have not slept very much in the past week so I have no idea if posting this is a good idea but I doubt I'll get even 5 votes or any takers but thanks for reading.