This week's reading consisted of information about social and implicit knowledge. I was pleased to see these addressed - Weinberger has been talking about tagging all along, but he hadn't yet gotten to the aspects of it that really make it a big deal - the inherent limitations, and how we're going to get around them.
In essence, sites like Flickr and del.icio.us are all about crowdsourcing. That's a buzzword making the rounds these days that means you're taking a task and throwing it out to the world to work on, a little bit at a time. It's especially functional when you're set up with tags - each user is going to bring their own particular perspective to any given item, and the words they use to reference it will reflect that. Once a few million people have tagged a few million items, a computer starts to dig through the cloud, and comes up with some interesting ideas about how to put things together. I can't escape the notion that this is what a reference library is going to look like in 20 years.
Of course, there's still a few bugs in the system. Take the Markov chainer. I don't pretend to understand Markov's laws, but this implementation is fascinating - you feed it a text, and it analyzes each pair of words in order, learning how to link things together. I've done some experimentation with these over the years, and I've seen some grow so big they can carry on conversations in perfect English. The only problem is the words only have relative context, so these sentences, despite being grammatically sound, are complete gibberish. The biggest challenge in turning a mountain of tags into an index is going to be making the results useful.
Fortunately, if we find a way around that, I think we'll reach a point where we don't have to worry about implied knowledge. Certainly picking up a jar of strawberry jam gives us more information than looking at a picture of it, but there's a certain question of how much of that knowledge is actually useful to us. It is possible that as this experiment proceeds, we'll learn that we work better when our minds aren't filled with useless trivia. On the other hand, it's equally possible that between our previously posited millions, we'll manage to capture every relevant bit of information, and it will all be presented contextually when we reference the aforementioned jar. Time will tell.
Speaking of useless trivia, during the chapter about implicit knowledge, did anyone else flash on the episode of Scrubs when Dr. Cox is carefully explaining each character's comedic role to Carla? That's always been one of my favorite bits of meta-commentary.
Subscribe to:
Post Comments (Atom)
That markov chainer is cool, I'll have to check it out when I get the chance.
ReplyDeleteHow do we go about teaching a computer how to interpret text in the correct way? As far as I am concerned, we program the computer how to do everything. From reading your post I feel as if we need to ask computers to do something that is just not feasible, to come up with something (as in a coherent sentence, from your example) on its own. Am I interpreting this correctly?
You make a point about the usefulness of information. Sure you can create a huge data monster, but it is USEFUL to anyone? The question is to whom is it useful. And the nice thing is that it doesn't matter - there are so many people cloudsourcing that a lot of work is being done, surely some of it will be relevant. But what if it isn't?! What if it's a bunch of irrelevant garbage that seems true because so many people have referenced it that it has undue credibility.
ReplyDeleteLike your example - so what that some computer program linked words, they can't make a coherent sentence. It's is a pointless exercise - unless it leads to something better. I like that you criticize the value of things and see thru the shiny, fake 'look at me!' accomplishment that really isn't contributing to the idea of increasing knowledge. It's more like self-congratulatory narcissism.
Because I forgot to comment on the rest of your posts...
ReplyDeleteI had never used blogging tools prior to this class, and never had too much knowledge of tagging, or whatever else certain sites call it. I know that you weren't the original creator, but when reading through your posts your tagging helps to offer some great examples to every thing you are talking about. I know that if I was reading through your post and I came across "The Markov Chainer" example, I most likely wouldn't type it into my Google search to find out more about. But since the tag is right there, I do. I think I will probably start tagging more in my posts.
Chris - I don't think it's impossible, I just think it's going to take a lot of work to get to that point. Anything is possible - it's just a question of how to do it.
ReplyDeleteAs to the other, are you talking about linking (so you can click on the chainer and go to its page) or tagging (so you can click on the post label and see other posts by me with that label)? Either way I'm glad I could help. :)
Celena - What if it does all turn out to be irrelevant? The advantage to that is then we'll know. "We tried this, it didn't work." Maybe we can try it another way next time. Or maybe we move on to something else. Either way, it's not a waste - it's experience that our society can build on. But I have hope. (Don't tell anyone I said that, though! They'll kick me out of the Pessimist's Club, then complain that they always knew I didn't have what it takes.)