Oddity in Google Street View

I was virtually re-visiting the neighborhood where I grew up, and noticed something puzzling. If you look at the intersection of W. 71 and Franklin, Cleveland, you see that Cogswell Hall, the big stone building on Franklin, has some construction going on. OK, no mystery there; I knew from my mom that they were expanding about a year ago. But if you move one step away from the intersection on either street, you see the pre-renovation view of the hall. So apparently, Google already had coverage of the area before, but then later came through again and took a new picture just at the intersection. Do they do this for all intersections? Why? Why not get new pictures of the rest of the neighborhood, while they’re there?

It is odd. Perhaps the amount of work required to update means they sometimes only splice in one new picture where they thought something has significantly changed so they don’t have to redo the entire street.

I was going to suggest that the intersection was shot as they came down the cross street (71st), but you can tell from the lighting that 71st was shot on the same day as Franklin - only the intersection itself was a different day. And if you go up or down Franklin, same thing at the other intersections. The intersections were all shot one day, and the streets a different day.

So I’m stumped. This might not be the only neighborhood that was shot this way, but I’ve never seen it anywhere else.

ETA: and the fact that other intersections in the neighborhood were shot the same way suggests it wasn’t just one bad photo or some random naked person in a single shot.

The view of my street contains two different cars I owned, but not both at the same time (there were several months between them). The way the images are stitched together makes it look like both are parked there happily on opposite sides of the street.

I remember seeing that happening when Google first published Street View coverage of a town where I spent some time in growing up. But now the imagery seems to have been cleaned up so that all the “mid-intersection” shots are in series with one of the streets.

My guess at the time (and a tech-savvy friend who lives in the same town thought it seemed plausible) was that perhaps the “mid-intersection” shots were taken first to lay out the basis of grid of the town in their database, and eventually these shots were replaced with images taken in a later photo shooting excursions.

Came to mind for me too, but here it looks like the intersections were taken after the rest of the streets, based on the construction status at this one intersection.

Disclaimer: I work in digital mapping, but not for Google.

My guess is that the new intersection pictures were taken as part of their ongoing initiatives to increase quality and make their map more navigable. In residential neighborhoods on 2-lane roads, the intersections are probably the most complex road features you’ll see. You have signage, speed limit changes, turn lanes, and some roads do not intersect in a place you can see from satellite (trees, etc.) Cities also change intersections (adding/removing stop signs, lights, etc) pretty regularly, compared to road links that are much more static. Road links themselves are just not that interesting or complicated to justify re-gathering all that data unless there is a known real-world change to document.

There are many more complex intersections out there of course, and Google is probably more interested in those. But there is no reason to send a truck out for 5 intersections when you could spend a little more time and get them all. Since Google’s data provider is TeleAtlas (who does not drive the roads), Google has to drive the roads themselves to increase data fidelity. Luckily data aggregation is what they do best.

Do you suppose as part of this entering addresses on my road might someday show more than a distant view of trees and a message that says “we are sorry, but we don’t have imagery at this zoom level for this region”? In a way, though, it’s sort of comforting that the closest view of my neighborhood looks like a green nerf ball with roads…

There are two incorrect assertions here. Google terminated their U.S. agreement with Tele Atlas last year (Tele Atlas still provides them with data outside of the U.S., as far as I know), and Tele Atlas sends their fancible trucks out just like NAVTEQ and Google do.

I hope it wasn’t for a random naked person… That building is an assisted living facility for the elderly.

And I can see why they’d especially want up-to-date information at the intersections, but it seems to me that if you’re going to send the truck out anyway, you might as well take shots of the street along the way, too. It’s not like they can just teleport from one intersection to the next. Or at least, I think they can’t: This is Google, after all.

Ah my mistake, thanks.

The main road through the village where I live seems to have been surveyed twice - once on each side/direction of the street - so the view switches between winter and summer depending on which side of the road you position yourself.

Google Street View is sometimes capable of recording interdimensional hand tools. Be forewarned.

It’s also possible the original picture of the intersection was requested removed for some reason. Somebody was giving the finger, or was getting a BJ from a hooker or something.

This makes sense given that the intersection image is of a significantly better quality - although their intent was thwarted by having a big freaking truck blocking the entire intersection!