I need to write more, so I’m going to try to change things around for kaddar dot net. I’ll try to write one short thing a day, no matter how random. Hey, if this lasts more than a week I’ll be happy.

The website digg.com is really great now that its user base has grown.

I’ve been thinking about the model that exists for it, and how we could re-apply it to other websites. What sorts of websites could benefit from the Digg model? Note, I am generally defining the Digg model as user-contributed non-editable content that other users can give feedback on.

Forums
I love the commenting system for Digg.   What if it existed for forums? Threadjacking is reduced by their half-threaded system.   I keep reading forums and feel like I want to digg down a user’s comment. It’s the internet bitchslap.

Webcomics
Another site I’d like to see the Digg Model exist for is webcomics.   There’d be a comic repository where the best webcomics, added through rss feeds, would be dugg up by users. Also, it would include a comment system so we could talk about how funny it was (we all know that just makes a comic funnier).

Review sites
No, I don’t think games / movies should be reviewed by diggs-per say. I think reviews themself should be reviewed by a digg system. I used to use crankycritic.com for a lot of my movie reviews. Metacritic is pretty awesome, but lacks enough user interaction on the meta side.

There’s always the problem with diggs though, they reflect the opinion of the userbase.   If the userbase only liked a certain type of movie, they might not digg up perfectly good reviews that argue valid points against that movie.   I imagine reviews of reviews snakes on a plane would be motherfuckin’ skewed.

Random Link: I really like this team special olympics comics
Tomorrow!1one: wikiforum idea? Or something else!

I found a website recently, called likebetter. Basically, you are given two photos, and you pick the one you like the best. It then guesses what kind of person you are.

In general, it fails pretty hard. This is fundamentally similar to 20q.net. In 20q.net we play a game where we use comparison to describing properties of an object and then tell the system what object we are looking for, thereby making the system more intelligent. In likebetter, we are describing the properties of ourselves and telling it who we are, thereby boosting those properties for those photos. The correlation should provide accurate guesses to who we are.

But the problem is that there is no true implied correlation between minor photo preference and personality, since photo preference is not based on personality. It is ambiguous to users as to whether we should be selecting which “item” we like better or which photo we like better (for example, spiked hair versus a screenshot in a video game). Viscerally, I select the spiked hair because I find the photo to be more interesting, even though I may like playing the game better.

Another problem is degree of how much I like something better. To say I like spiked hair more than video games, I wonder if the server takes this to mean I do like spiked hair and don’t like video games. If that’s true, then there is no measure for the degree between the two. Without offering a “neither” selection and “both” selection, the likebetter system also allows for more common errors.

There’s probably a better standard with which to decide what kind of person we are through simple comparison and correlation, but this isn’t it.