It turns out, the "search for kml" feature in Google/Google Earth had been around before, but it’s now much improved and it’s worth blogging about the consequences. In lay terms, this is one big step to the metaverse so many have speculated about.
Why? For example, right now, all I’d need to do to get on the 3D map is put out an AviBarZeev.kml file with an icon of me, or even a 3D model or me and/or my house. If you search for me, or perhaps even for keywords for which I rank highly, you might just find me on the earth. It’s not what the avatar-seeking crowd wants just yet, but mainly because the frequency of update is low and non-interactive. However, if that KML file contains a network refresh that updates my position, then you can begin to see the next step in that particular direction.
There are limitations and downsides, of course. For one thing, this feature, while powerful, has the same practical issues as generic web search — semantics. And I’m hoping as Google tackles that overall (as it must to survive), it gets integrated here too.
But the more interesting angle is the question of what content will people find. Though Michael Jones refutes the idea that Google would support fictional (anything other than mirror-world) data planted on real Earth, Chikai’s example using Lord of the Rings is telling of what people will actually choose to do with this. Here, it seems someone plotted the shooting locations for fictional Middle Earth in actual New Zealand. But what’s to stop someone from plopping a version of Manhattan that’s straight out of "Under Seige" or "Escape from New York?" or even try to use GE as a platform for some game based there? It may not be part of the official GE experience, but to an end user, what’s the difference if it shows up on the Earth? In other words, if you search for "bridges out of Manhattan" and find them all barricaded with razor wire, that might affect your travel plans.
As long as it’s so obvious or at least labeled properly, it’s not a big concern. The bigger problem comes in when data is presented as factual and it is not. This is a problem of the web in general (and of areas like the West Bank in Google Earth in particular). I’m hoping the wiki-style mashups will help address some of that. And Google reportedly promotes "good" web pages implicitly, due to how well linked they are. It’d be interesting to see how well KML files get promoted or demoted in the search results based on their prominence, as well as other "quality" factors.
Now, I’m not aware of any sort of cross-linking among KML files yet, similar to how web pages can point to or reference each other without forcing each other to load. What would be fascinating to see is whether KML files will evolve as web pages have — my excellent Eiffel tower model could reference your fine Champs-Élysées and someone else’s Arc de Triumph. We would thereby cross-promote each other (in context: Paris Landmarks)… We’ll have to wait and see. But if it’s not already there, a new field for KML which offers the idea of an optional (e.g., clickable) link to other KML files, coupled with a sort of page-rank algorithm using these links, might just do the trick to creating this sort of "web of goodness" we enjoy.