April 18th, 2021
OpenStreetMap
#2,308
Cover to cover
It was homesickness combined with an interest in roads that got me into OpenStreetMap originally, but I keep getting more deeply involved in the monumental task of completing the map in every imaginable way because it’s for a good cause and because, paradoxically, it’s unfinishable. Even a pandemic, for all its horrors, presented an opportunity to make a difference through mapping.
Tragedy of the commons
For several years after I moved to Silicon Valley, I kept contributing to OSM’s coverage of Ohio and never paid much attention to its coverage of my new surroundings. I assumed the capital of high tech would have an open-source map fully figured out by now. But in fact, when I finally got around to zooming into my part of San José in 2015, it had basically been untouched since the initial nationwide import of the Census Bureau’s low-quality road data, with far less detail than one would find in many Midwestern cities. It was reminiscent of how Google Maps and other proprietary maps had long neglected my hometown in Ohio. I guess everyone around here made the same assumption I did.
OSM is a far cry from proprietary maps. For one thing, we don’t have product managers deciding which cities should get a dedicated surveying team based on market potential or good PR. Perhaps that is how OSM’s corporate mapping teams prioritize, but they don’t have boots on the ground, so they can only help in so many ways. Setting priorities for the volunteer grassroots mappers, the ones who can really transform a map into something eye-catching and unexpected, would be harder than herding cats.
To my chagrin, Silicon Valley has an unearned status as the book cover of every world map out there. All too often, a product manager making purchasing decisions for a tech company will judge a world map solely on how it performs as they search and run errands in their neighborhood – which more often than not is somewhere in the Bay Area, right in Google’s backyard. A poor first impression here means OSM never gets a chance regardless of its coverage elsewhere. If only they knew about our gems, our labors of love, in random other parts of the world!
Mapping on a shoestring
By now, OSM’s map of Silicon Valley is also a labor of love. As soon as I started mapping in San José, I wanted to get it up to par with Cincinnati, but this time I didn’t have the advantage of knowing my way around town. I was never as bad a driver as when I struggled to jot down notes at stoplights or dictate to my phone while training my eyes anywhere but the road. It didn’t help that I kept saying the shop was on the right when I meant left – or was it right? As soon as I could get through my backlog of voice recordings, I’d have to resurvey the same places because new restaurants had replaced the old ones.
A few of us grassroots mappers found each other by chance and set up shop within Code for San José, even though mapping doesn’t really involve coding. We started scheming to build out the kind of superb coverage one would expect from an OSM community ten times our size. We copied some 3,500 miles of sidewalks, 250,000 buildings, and 200,000 addresses from local government databases, instead of mapping them by hand as larger communities often do. We’re like a scrappy startup (with no hope of a business model).
Blind spots
With streets, sidewalks, and buildings, San José now has a respectable level of coverage on OSM compared to other major U.S. cities. But these days, an ordinary user would consider a street map lacking businesses and other points of interest to be akin to chartjunk. Ask anyone who works on map and navigation software and they’ll tell you that most user feedback has nothing to do with the color scheme, turn instructions, or estimated travel times – it’s about the relevance and accuracy of search results.
Until this year, our approach to mapping businesses largely entailed collecting street-level imagery, then waiting for people to enter the businesses manually using the collected imagery. Occasionally, one of us would drive around taking notes, or a business owner would spam the map with an advertisement. (We’ll take what we can get.)
This organic approach helped us cover roughly 20% of the businesses in Silicon Valley. This shortcoming is by no means unique to this part of the country. I think it’s fair to say a lot of mappers have come to view comprehensive POI coverage as a Hard Problem and unattainable luxury.
Worse than being incomplete, our POI coverage was extremely uneven. You could generally rely on OSM for prominent, car-culture-friendly businesses like gas stations, but not for professional services or healthcare. Downtown areas and some suburbs had dense POI coverage, but working-class, minority neighborhoods and new suburban developments had sparse coverage.
Catching up
When the pandemic hit, the county’s first-in-the-nation lockdown threatened to shutter many businesses and render all the years of ad-hoc POI mapping useless, especially in the minority neighborhoods where we were already falling behind. How would we catch up to the proprietary maps now?
Luckily, the county public health department required every business to submit a site-specific protection plan online as a condition for reopening. At a brainstorming session in September, we discovered that the county was publishing these forms on a website, including the name and address of every kind of business, many in the very places we were finding difficult to cover well. It’s as if the county has dropped a phone book in our lap, but this phone book is machine-readable and updated daily with businesses that are sure to be open during the pandemic.
We turned the county’s dataset of forms into yet another massive import. Eventually, this import will result in 20,000–30,000 new POIs, making it the third largest POI import in OSM’s history, larger than most countrywide imports.
Unlike our previous imports, this import is especially labor-intensive because each entry needs to be hand-reviewed for accuracy. So we simplified the workflow so that even the most novice mapper can contribute from the comfort of a Web browser. It’s convenient enough that even I can comfortably contribute despite shying away from the usual import tooling. Now the challenge is to recruit more participants before the pandemic ends and this dataset starts getting stale. If it were just up to the few of us longtime mappers, it would take us years to complete the import, so we really need help.
Though we’re still mired in the 15% range, this import is already making an impact. Whenever an iOS developer tests their location-aware application in a simulator, the simulated routes around Cupertino feature many more POIs than they used to, including childcare centers and offices we never thought to map all this time. The map looks full like never before. It’s a baby step toward changing outside perceptions about OSM. Meanwhile, we seem to be filling in gaps in places like Milpitas and East San José that had gotten neglected all these years. I wouldn’t be surprised if we end up covering this county’s many ethnic enclaves better than any other map.
I’m hopeful we can build up an active mapping community that can not only fill in these gaps but maintain our momentum for years to come. In time, the businesses that survive this pandemic will benefit from being included in OSM. Maybe we’ll even serve as a model to other cities’ mapping communities, so that our coverage of Silicon Valley will be more than a mere book cover.
Please help
First, if you aren’t familiar with contributing to OSM, go to openstreetmap.org and create a free user account. Once you’ve logged in, zoom into your neighborhood and click the “Edit” button at the top of the page. Spend a few minutes completing the interactive tutorial to get a feel for editing, then dive in and add missing details to your neighborhood.
Once you’ve gotten the hang of basic mapping, go to this MapRoulette project and follow the instructions to get started. It’s a safely socially distanced activity and a small way to help the local community come out stronger after the pandemic.
TrackBacks
TrackBack URL: <http://panel.1ec5.org/mt/mt-ping.fcgi/1710>
Comments and Concerns