People, Not Data. Again.

Well, the Presidential election turned out a little differently than most people expected. All that data, all that polling, all those models. Almost all wrong.

Data can be a powerful agent of change. But only when we have some understanding of what the data describe. We were adrift in a sea of numbers over the last six months. We missed the people behind the numbers.

Please, go (re)read Jake Solomon’s post People, Not Data from almost three years ago. Three years ago many people had just had their first taste of FiveThirtyEight and it was addicting. But despite working to apply lessons from technology to improve government services, Jake barely mentions data, or any other technology, in his post. Instead he spends his time sharing his first hand observations of people and how the obliviousness to real people that’s been inadvertently built into so many government services affect them. Here’s a taste of it:

This is San Francisco’s main Food Stamps office. People call it twelve-thirty-five, as in 1235 Mission Street. The first time I went was on Thursday, February 7th, 2013. I walked past the concrete pillars, through the metal detector with two security guards, past a table of scattered paper forms, and into the main waiting room. It was loud. Voices echoed from speakers on the ceiling and off the linoleum floors. There was one big line leading to one big countertop, called Service Counter B.
A tall black man stepped to the front of the line. He hunched forward and leaned his hands on the countertop. A thick, clouded sheet of bulletproof glass separated him from the worker (short for social worker in the world of human services); they spoke through two skinny, conference-style microphones mounted on the glass. He was having trouble hearing. He leaned in and grabbed the mic to flex it upwards, but it wouldn’t go any further. He hunched a bit lower, putting his ear to the glass. Then, after a few more moments, he lowered his knees to the ground, pulled the mic down to his face, and rested his arms along the countertop. He finished the conversation on his knees.
So there I was: In San Francisco — one of the greatest and most prosperous cities in our country — watching a man on his knees, struggling to hear through bulletproof glass, trying to access nutrition assistance from our Federal government.

I am often asked to serve on panels at conferences about big data and cities, big data for social impact, big data in the public sector. They’re not my favorite topic, partly because I’m not a practitioner, and those who work with data every day are better suited to speak to it. And it’s partly because the big data label is largely misapplied in a local government context. But it’s also that if we don’t understand the users about whom we are collecting data, that data doesn’t mean much. We’ve seen it in the ways that government often turns to marketing campaigns for programs when the data show low enrollment, not realizing that it’s not so much low awareness suppressing enrollment as poor customer experience and high rates of abandonment throughout an overly complex enrollment process. They look at data about the program, not about how users experience it. And not about how people feel about it, as Jake starts to get at in his post. Something went wrong with our data about the election too. I don’t know what it was, but I will echo the thousands of voices who are saying that the media (at least) obviously don’t understand the electorate. Lots of people are guilty of that at the moment.

[Update: danah boyd offers valuable thoughts on this topic: “I believe in data, but data itself has become spectacle.” Well worth a read.]

Jake’s observations about the users of our social safety net start to get at the understanding we need to make a difference with data. And let me point you to the words of another Code for America alum, Dan Hon, who wrote the following earlier this week.

Advertising, focus groups and quantitative polling don’t work anymore.
1) Understand your users’ needs
2) Show — don’t tell — that you understand those needs
3) Meet your users’ needs
The President-elect won by doing a bit of 1 and a lot of 2. He’s eminently unqualified to do 3. The extent to which the “system” or the establishment is broken is that it’s not clearly doing, 1, 2 and 3.
Government is all about understanding user needs and delivering on them. And now, demonstrably and especially, at the political level.

Understanding users and user needs — people and people’s needs — should be the dominant framework. Data is one important tool to do that. But without deeper understanding, it can be terribly misleading. We will hopefully learn many lessons from 2016 election, but let one of them be the value of small data. Sometimes very small. Like a conversation. Or a visit to twelve-thirty-five.