lots of equipment in the car. We had to put the generator on the roof. But we made a nice enough demo for the execs. Then we got approval to make this a real project and hire a few engineers. That was late 2005.
X: So then what did you learn?
LV: What’s difficult is to go everywhere. The dominant cost is operations—we had lots of cameras, lasers in all directions, so much equipment that we had to have a mini data center in the back of the van. The system was so overdesigned that in fact there was always something going wrong with this. So instead we decided to focus on collecting only imagery, at large scale. We went to a system with an off-the-shelf camera and computer. That we were able to scale quickly. (We spent 2006 to make the product, and built the front end. In early spring 2007, we moved to off-the-shelf.)
The first five cities launched in May 2007—San Francisco, New York, Las Vegas, Miami, and Denver. We had Google security guys drive the vans, and we drove some ourselves.
X: How did the trike project start (see photo below), and what’s the significance to the company?
LV: Google being Google, we have great ambition to keep expanding and doing new things. After we covered roads, people started asking us about going off-road or inside. We’re slowly moving in this direction. We start with the trike—we put the same equipment as in the car, in something human powered—for campuses, pedestrian areas, bike trails. We don’t want to leave it at this. We are experimenting with other systems to go indoors. We have more portable systems, to put in a backpack. We’ve gone in a snowmobile, for the Olympic Village in Whistler.
We want to capture areas that matter for historical or shopping reasons. We have collaborations with locations like Faneuil Hall, Boston University, U. Penn, and San Diego State. A number of universities see this could help students and parents [with virtual campus tours], and help keep alumni better connected. [Trike images from San Diego State University and San Diego Zoo are currently viewable, and upcoming sites include Rochester Institute of Technology, Boulder Creek Path in Colorado, and the Detroit Zoo—Eds.]
X: How do you deal with consumer privacy issues?
LV: Google takes this very seriously. We provide very easy ways for people to report problems—you complete a form and submit it. A lot of stuff [images] we’ve removed because of users. We also do large-scale [automatic] detection of faces and license plates in the images. It’s never been done at this scale. It’s automatic 99.9 percent of the time. The rest is manual.
X: Where is this all going for Google? What’s the impact on the business side?
LV: What we would love is to be able to take you anywhere. Anywhere you navigate using maps or a phone or Google Earth, you’re able to see high-quality photos and navigate through them. We want to give you the best possible view of what you want to see. Whether it’s areas, street-level, whether in a city or countryside or a park.
We collect all this imagery, and we might not collect it all ourselves—there are user photos we can match to street-view imagery, in key touristy areas. We want to find more of those photos. (We work closely with Steve Seitz, who’s at Google and University of Washington, on this.)
For now, we don’t have plans to monetize this. We are still expanding. Our goal is to make our maps products more compelling for our users. Down the road, who knows?
