How Your Online Burrito Review Could Help Standardize Municipal Data
For the July/August issue of The Atlantic, I wrote a piece discussing the intriguing new initiative between Yelp and several municipalities, starting with San Francisco, to publish restaurant health inspection scores right alongside reviews on the consumer platform. Technically, these scores are already public information, sitting in the data portals of many cities. But a lot of work has gone into converting inspection results into a format that Yelp, or any other review site, can automatically ingest. And the potential implications are pretty compelling: As I write in the magazine, researchers and city officials are hoping that this idea could make the data more useful, the restaurants cleaner and safer, and the people who dine in them less likely to vomit.
The idea is about both open data and public health (oh, and the awesome power of algorithms!). But there's also some greater potential significance here that I did not dive into much in the story.
We write frequently at Atlantic Cities about tools cities are building (or enabling citizens to build) with open data. But a building code violation dataset in one city is not necessarily compatible with a building code violation dataset in another city. And this is true of all kinds of things. Cities measure miles of bike lanes differently. They track snowplows differently. They respond to and count potholes differently. In fact, until now, we've really only had one major example of a common data specification used across multiple municipalities to feed information into national consumer platforms: That's the General Transit Feed Specification, which Google helped develop with cities so that their transit schedules and station locations might usefully be imported into platforms like Google Maps.
The Yelp health inspection page for a restaurant in San Francisco with quite a lot of problems.
This new health-inspection format – Yelp is calling it the Local Inspector Value-Entry Specification, or LIVES – is just the second time cities have attempted to communicate this kind of valuable data in a common language. But they're going to have to learn to do this with more than just restaurant scores and transit schedules if we're going to realize the full potential of open data. A local developer can build an app for Louisville with data from and about Louisville. But if we want municipal information on national platforms (and we want entrepreneurs to develop new national platforms to serve cities and residents), then 50 cities can't release 20 incompatible spreadsheets variously identifying potholes with five different traffic engineering euphemisms.
In trying to unify all their data, cities will eventually realize their processes aren't unified either. This has been a central challenge with the health-inspection standard. San Francisco measures restaurants from 0 to 100. New York and L.A. do it with letter grades. Chicago uses pass/fail. We can only assume they count rat droppings differently, too. For now, the LIVES standard leaves room for these local variations. But its existence makes it easy to imagine how cities could feel compelled down the road to start inspecting their restaurants, and not just publishing their data, in a common way, too.
One city, for instance, may appear to consumers on Yelp as if its restaurants all have higher grades, when in fact its inspection process is simply more forgiving.
“These grades will mean different things, and I don’t think the public’s going to want them to mean different things,” Rajiv Bhatia, the director of occupational and environmental health for San Francisco’s Department of Public Health, told me. “And some things about process are also going to get uncovered.”
This conversation started with the idea of putting health inspection scores next to reviews for pizzerias and BBQ joints. But what if by sharing data standards of all kinds like this, cities were forced to rethink the very processes that data describes? Open-data standards could change not only how cities carry out some of their most basic services (inspecting restaurants, repairing potholes, responding to 3-1-1 calls). It could also create a whole new ecosystem of applications built on that data.