17th level Hacker

The World Outside

I went over to the BayCHI presentation on Mobile Media Metadata last week. Fantastic stuff. Marc Davis from Garage Cinema Research spoke about the research they have going on. The presentation he gave focused mostly on mobile metadata. For now lets just talk about images. One of the points he made was that image classification systems tended to focus completely on the image itself. They look at a stream of bits and either attempt to determine the content of the image or attempt to fufill queries based entirely on those bits. But with the source of photo content increasingly being connected mobile devices, there’s potentially a much richer source of info than the bits themselves. The temporal and spatial information can potentially tell the system a lot more about an image than can be inferred from the image itself. Add in some social aspects, such as “what have others who have taken pictures in approximately this location used as the subject of their photos”, and the system can infer quite a bit.

This is one of the aspects that’s missing from the wider practice of blogging. Efforts like GeoURL started getting me looking in this direction. And most recently the buzz around Plazes got me looking back. There really should be a general distributed system for looking at information in the blogosphere based on time and location. Right now our main methods of organization are social (or something like it). Most people read someones blog, they’re picking people who have interests that mirror what they want to hear about and subscribing to those feeds. There are plenty of topical blogs, which helps to refine that subscription list somewhat. But that’s decomposition based on taxonomy (or folksonomy). That works great for getting news, but that’s only one potential pattern for information flow. The whole flashmob thing popped up and died down pretty quick, but there were a few nuggets in there that we probably shouldn’t ignore. Such as information flow around a particular event at a particular time in a particular location. How do you arrange information around an event like this? Seems like it would be a pretty common problem. One that’s likely to get more and more common with an increase in mobile usage. But right now there’s really not a great answer to it. Shared blogs are setup for events that get planned well in advance. But even there the support doesn’t seem to fit the pattern that I think most users would like.

When I first heard about the trackback mechanism I was hoping that this would provide at least a partial answer. Someone could setup a trackback location for the event and everyone who has something to say would post on their own and register the link with the event trackback. In practice however there are a bunch of issues with that. First off, of course, is that you need to know that trackback location in advance. Then there are technical hurdles with the trackback mechanism not really being understood or tailored for the novice user. It’s a good idea, but maybe not the best. Ideally we would want everyone to just post their info independently, and have a system that allows us to find all those posts after the fact due to knowing the location and time of the event. In the end, this is really what you probably want for news creation also. Lots of times the news is made because something unexpected happened. One would want those news capturers to be able to just click away and record media, which we consumers could go back and find. I think this is one of the basic building blocks missing right now from blogging. Being able to say something like “what did people say at the protest last night in Palo Alto” shouldn’t have to be based on knowing who the people are who are most likely to have been at the protest. Right now we’re substituting social information as a proxy for some forms of metadata that we could have direct access to. Social information is certainly a great source, and a fantastic way to organize a system so that it connects people to what they want in the absence of other factors. But before we go too far down that path I think we need to pay a little attention to the other factors that can be used in organizing these information flows. A lot of this information is coming in from the world outside, away from our desks. Just like image analysis, paying attention only to who said it and what the content is ignores some very rich info. There is a world outside, apparently. Lets try to acknowledge it a bit more.