My Visit to Socrata, and Data Analysis About Data Analysis

August 30, 2013 8:00 am PST | Data as a Service

By Thomas Levine


A couple months ago, I downloaded the metadata files for all of the datasets on 60 Socrata open data portals. Interesting things happen when you do craziness like this.

Last Friday, I visited Socrata in Seattle. The most important part of my visit was of course the teriyaki lunch; it was so important that Developer Evangelist Chris Metcalf sent me a Google Calendar invitation for it. They’ve been doing this for about 19 months and have only missed two weeks!

Of secondary importance was my discovery of a glowing ampersand.


(I like ampersands.)

I also met everyone, talked about my findings, and learned a lot about the implementation of open data portals.


After my talk, VP of Product Saf Rabah asked me why I had done all this study of open data. That was a good question; I regularly forget that I tend to do strange things.

I think most things do better when knowledge is shared. I run and write free software, I post most of my work quite publicly, and I pay attention to media licensing. I read copyright law when I was in high school, and I started a Free Culture Foundation chapter in college. Naturally, I also care about open data. But that still doesn’t quite explain it.

Hackathon Apps

“I think most things do better when knowledge is shared.”
Thomas Levine, Data Scientist

At the first hackathon for the NYC Big Apps competition, Ashley Williams and I started talking about the sort of apps that come out of contemporary hackathons. We quickly decided to algorithmically generate random hackathon apps.

We noticed that many apps follow a search-for-things-on-a-map paradigm. That is, you fill out a search form and get results on a map. We automated the creation of the cliche hackathon apps in AppGen (being cliche hackathon apps, AppGen-generated apps also broke a month after being created.)

Dataset Dataset

In building AppGen, I wound up downloading all of the datasets on the New York City data portal. I generalized it to run on any Socrata Open Data Portal, and then I downloaded all of the metadata about all of the datasets on 60 Socrata portals.

I started treating datasets as data points and doing analysis across datasets. I think this is the main novelty of all my open data studies: I’ve just been quantifying things that other people hadn’t thought to quantify.


Someone recently suggested that I’m generally interested in understanding how information gets created and shared.

Fun related story: this one time, I posted “EMERGENCY EXIT ONLY ALARM WILL SOUND” signs on doors that weren’t emergency exits only. Those emergency signs are scary! Even I was afraid of opening the doors afterwards! I continue to be intrigued by how signs are seen as authoritative sources of information even though anyone can put up a sign.

When so much information passes through a central and open resource like a data portal, we acquire rich data about how different people produce and interpret information. This ecosystem of open data doesn’t just give us more data to analyze; it also allows us to analyze data analysis.

Previous Article
Data Apps & Visualization, Data as a Service, Effective Governing
CIO Jon Walton on San Mateo Open Data

September 3, 2013

Next Article
Data as a Service, Effective Governing
Raleigh, NC Wins National Award for Open Data Portal

August 28, 2013