We are sidetracked on whether points alone or some complex feedback measure is better (i.e., huge variety of badges in SO; complex way of getting and losing karma).
There's something more crucial.
I've been thinking about this for a while.
It's the distinction between fact and opinion.
On the web, it'd be mighty useful to be able to tell them apart. And assign a credibilty value to *anything*. Otherwise, we are at risk of being manipulated.
Back on topic of So vs. DC...
When you have fact-like info, I think SO interface is better.
Example: "how do i change font size in eclipse?"
This can be easily tested to see if it works.
Now we all know and love that not all the content here is facts, but opinions. Same on stackOverflow! and it was designed for factual stuff.
Example: "what is the best [foo]?"
Here we are in opinion territory.
Q1: Can humans distinguish between fact and opinion easily? Do they? My gut feeling if that we take advice -even for big decisions- on things that have very little relation to facts or empirical evidence _most of the time_.
Q2: Could machines be trained to make this distinction?
Q3: How about a mix of human and machine?
Q4: Could this be done at web scale?
Q5: Would the system be trustable and reliable enough to have reference value? i.e., can it be gamed? Example: Running shoes. There's a XX billion industry on top of them. However, nobody has shown any effect of expensive shoes getting less injuries to people.
What I'm thinking is really not practical because the marketing depts of the world, who implicitly run the media
would tag together and stop this system from being successful.
Game1: next time you see a sentence that is likely to influence your behavior, ask yourself: "How do I know what this is saying is true?".
Game2: Try to assign a trust value to each sentence on any random paragraph. Prepare to be amazed.