What makes a street beautiful? The folks over at OpenPlans are trying to figure that out. The project asks website visitors to choose from pairs of randomly selected streets in Philadelphia to decide which is more beautiful. It’s a neat idea, but has some major flaws:
Firstly, the project uses images from Google Street View. Anyone who has used street view knows that images are distorted, and quality varies greatly. When clicking through the “beautiful streets” survey, I found myself clicking the better quality image almost every time…even when I was trying not to. Something that was fish-eyed and hazy was hard to identify as “beautiful.”
Secondly, the angles of the photographs vary and have a big impact on the viewer. You can rotate the images (as in google street view), but will people actually? I didn’t much. Thus, if the first view is that of a concrete wall, it’s not going to look beautiful, though the street may in fact be.
Improving the project would take a lot of additional work, and I’m not sure it’d be worth it. But perhaps the organizers could go out and take their own 200 pictures. How to avoid bias by the photographers is an immediate issue here. There’d also be a loss of “3D” capabilities, but I’m not sure that’s an issue.
You can decide for yourself, though. Check out the project HERE. And let me know what you think. Is the project going to get any useful data? Are there any easy fixes the organizers could have taken to improve on the project?
And for good measure, one of my favourite beautiful streets, Newberry Blvd in Milwaukee, WI. Google Street View doesn’t really do it justice.