Dopass.com

It has been an interesting day. I actually started to write this once before but an errant key press wiped the whole thing out. Who knew you could still do that?

Anyway, let's try this again. This morning, after learning that I could not watch the Tournament of Roses Parade here in Colorado, I decided I'd just bum around; No games, no work, no nothing. So, I queued up Cosmos on the Tivo, grabbed a Diet Rockstar and wrapped up in a blanket...

I was watching the Encyclopedia Galactica episode. In this episode, Carl Sagan explores our search for life amongst the stars and contemplates what a truly inter-stellar culture would be like. He delights at the idea of a 'Book of Worlds' - an encyclopedia containing the knowledge, not of a single world, but of a billion worlds. As he considers this, we are shown entries from this encyclopedia. The vital statistics of some alien species is displayed on the screen. Despite the fact that they are labeled, I find them meaningless. It's as if they are just flat lifeless facts with no connection to anything else. Don't get me wrong, I know what they are, notes about culture type, descriptions of the species, etc., but they just don't mean anything to me. A few minutes later, as Carl brings the focus back to us on Earth, we get to see the entry for humanity. Unlike the alien entry, this one makes me smile. In it, I can see Carl's optimism and even a hint at the cynicism that accompanies it.

I had the context of being human, and knowing Carl Sagan's work, to understand how each piece of the data, each vital statistic, related to each other. As the show went on, I took a sip of my rock star and thought to myself:

The only good way to keep a grasp on a database like that - a database of a knowledge of a billion Earths - is to somehow include the context of the user into the search. Simply put, a scientist's idea of the 'vital' statistics of a species is going to vastly differ from my own. If I search for 'species descriptors', I don't want to know that they're monospecific monochromatic technological hemogenists. That's great information, but I'm a writer, a storyteller. I want to know if they're known for being greedy, temperamental or philosophical or if they have rich mythology. I want to know what kind of people they are, not what kind of organism they are. My perspective requires a different set of data to complete the search for 'vital statistics' successfully.

So, then I got to thinking about 'context sensitivity' and 'context dependant views' as they exist today. Why doesn't Google ask me what kind of person I am or what perspective I'm searching from when I do a search? It seems like you could get much better searches if the database knew what kind of information you considered 'good.' Sure, it's harder than it sounds, but is it really outside of our capability, right now?

I mean, go to Google and do an image search for "ass" - try it with strict safesearch on first. Then try it with safesearch off (don't do this if you're underage or not prepared to see something a bathing suit covers, folks). Do you see how you get very different images for each search? Did you notice there isn't a single donkey on the first page of either search?

What if I were a vet or a farmer? That search would fail for me, even though it doesn't have to. I'm sure that if I did a search for "donkey" (TURN SAFESEARCH ON), I'd find a picture of the animal I was looking for, but an ass is a donkey, so why should I have to?

- Snipehunter