This story appears as a part of Open Data PGH, a joint reporting project by Technical.ly and PublicSource on open data trends in Pittsburgh, underwritten by Heinz Endowments. Learn more here and get updates here.
When the City of Pittsburgh launched its snowplow tracker in 2015, the goal was to show where city plows were during a snowstorm. The well-intentioned app led to some confusion, however, since not all plow trucks being tracked were actively plowing, despite what the map seemed to indicate.
It’s a somewhat benign example of what can happen when civic data is presented without context, leading to conclusions that may be unintentionally misleading. A map of vacant properties in a neighborhood, with the lots color-coded red, for instance, can create a negative impression; if the lots are presented in relation to other inhabited properties, perhaps they become side-yard opportunities rather than vacant lots.
“There are a lot of different dimensions to consider,” said Bob Gradeck, program manager of the Western Pennsylvania Regional Data Center (WPRDC), which provides infrastructure for distributing data sets. “Just because you can do something with data doesn’t always mean you should do it.”
"We should be encouraging publishers to make data available disaggregated by race and ethnicity wherever possible."
Gradeck said he and the team at WPRDC are mindful of the way they map city 311 information, for instance; it would be problematic to map complaints by precise address if one neighbor is complaining about another.
“When you start getting into things like medical data and 911 calls, how do you release that data in a responsible way?” Gradeck said. He added that it’s important to stick to the facts when presenting data, but that presenting information totally out of context isn’t a great approach either.
Gradeck said a key takeaway of the National Neighborhood Indicators Partnership conference in May was that there should be more focus among civic technologists on highlighting structural racism and inequity: “We should be encouraging publishers to make data available disaggregated by race and ethnicity wherever possible.”
Being able to access data from closed systems such as gunfire detection technology ShotSpotter (which isn’t required to share its raw data) and having the ability to understand how sensors, technologies and algorithms are used and why — and how well they work or don’t work — are all important aspects to consider, Gradeck said.
He frequently refers to a set of guidelines put together at the Responsible Data Forum in 2014, that address how to handle data from marginalized communities. The “QAF” or Questions to Ask Frequently, suggests that data scientists need to consider whose data they’re using, what privacy concerns they may have and how the data could be manipulated. A few of the QAF suggested:
- Do you understand the implications around data ownership? Are you taking power away from a community by being in control of the data?
- Who is actually making the decisions about the data and what are the implications?
- Are your activities disempowering the community?
- Who should analyze the data? Is this an important skill for them to learn for the long run?
“It’s about developing personal relationships and trust with the communities whose data you’re working with,” Gradeck said.
Sean Luther, executive director of public-private partnership InnovatePGH, said the city snow plow tracker is a good example of an almost journalistic drive among technologists to attempt to present data in a neutral way.
“The idea was not to say the city is really fantastic at snow removal, it was about creating transparency,” he said of the plow tracker. That push to be neutral can create the opportunity for someone to develop a counter-narrative.
"I think technologists and data scientists need to be actively thinking: If I put this map out, what is the Dr. Evil use for it?"
An example: using data to map deed-restricted housing could show a high density of affordable housing in certain neighborhoods, which could lead to an unscrupulous developer claiming there’s no more need for affordable or publicly subsidized housing.
“There’s a level of professionalism happening in the civic data space right now that is not necessarily matched by the universal understanding of what the data we’re trying to use is saying,” Luther said.
His organization’s mission is to implement the recommendations from the 2017 Brookings Institution report “Capturing the next economy: Pittsburgh’s rise as a global innovation city.” Part of that work is identifying opportunities for technology use across the city.
Allowing users to dynamically change what they’re looking at may take away some of the potential for manipulation of data, Luther added; rather than publishing a static map, create a map that users could adjust by filters that create a clearer picture.
Ultimately, the snow plow tracker led city officials to reconsider plow routes and other potential improvements to its snow removal efforts, so the tracker’s unintended effect may turn into a positive outcome.
Luther said he’d like to see civic data technologists approach data presentations like chess players.
“What chess forces you to do is to think not just about what your move forward is, but to constantly think about what your opponent’s strategy is going to be for the next three moves,” he said. “I think technologists and data scientists need to be actively thinking: ‘If I put this map out, what is the Dr. Evil use for it?’”-30-
3 technologists’ hopes for a more open Pittsburgh
How to build a civic tech community, according to Pittsburgh leaders
5 ways Pittsburgh’s public servants are using human-centered design
Pittsburgh’s 412 Food Rescue: ‘Born digital’ and growing
Sign-up for daily news updates from Technical.ly Pittsburgh