By Dennis D. McDonald

Intro

Waldo Jaquith’s presentation at last week’s excellent  Socrata Customer Summit in Arlington, Virginia was a refreshing look at the challenges of publishing open data. Here are a few quotes I jotted down during his presentation:

• “You have no idea who your users are.”
• “A lot of you are mediocre.”
• “Governments talk a good game about open data but they provide little or no data.”
• “Easy is a trap.”
• “Let experts be experts.”
• “You need to get to someone who really controls the data.”

His statements, admittedly, were intended to be conversation starters. Yet I’ve seen the challenge firsthand of initializing data programs from the ground up and have written about them on my blog and in the Balefire Global blog; for example, see Three things about open data programs that make them special and Open data program managers need both analytical and structural data skills.

Easy is a trap

I definitely agree that “easy is a trap” and “let the experts be experts.” One aspect of the challenge is that open data programs are sometimes assigned to the IT department to develop and run. A major determinant of success will then be how well the IT department interacts with the various business units it serves. The more connected IT is with the organization’s business strategy, the better aligned the open data program will be with the organizations goals, objectives, and programs.

If the IT department doesn’t have a seat at the management table, on the other hand, it will be less likely that the open data program will be tied in to what the government agency of department is trying to do.

Approach the hackathon carefully

The same goes for government-sponsored “hackathons” which Jaquith also had some harsh words for. As I wrote in How to make datathon effort sustainable, what happens before and after a hackathon is just as important as what happens during. Also important is to have data experts and business experts involved, not just enthusiastic number crunchers.

An open data program, to be successful, must involve more than just putting datafiles out on a well-organized and indexed portal site. Open data should be viewed as part of the various operations of the government agency that sources the data and should have both an internal and external component.

Internal and external dimensions

The internal component should be focused on running, managing, and analyzing the government agency’s performance. The external component should support and extend the government’s programs through agency controlled service delivery mechanisms as well as the creative use of agency data by third parties.

This doesn’t mean that the core elements of data quality and completeness should be ignored. The more critical data are to managing and delivering government services, the more attention data quality needs to receive.

Will open data programs go away?

One might argue that, as data-focussed concepts and ideology become more embedded in how government programs operate, the more likely it will be that today’s open data catalogs and data portals will be viewed as things of the past.

I’m not so sure about that. My reason has to do with variations in data literacy levels and variations in how comfortable — or uncomfortable — government managers are with using data as a flexible management and service delivery resource.  I come from the school where the data you gather, organize, and analyze is gathered, organized, and analyzed for a reason. That reason might be many different things: to help do a job, to support a decision, to measure some operating cost or performance, or help make a prediction. Different uses require different data.

Users and data change over time

As uses change over time the data required to support those uses will have to change as well. How well data will support new or unanticipated uses may become an issue if the costs to modify or adapt the data to new uses are viewed as prohibitive.

Data as a resource

In other words, open data need to be managed as resources, not as end products.

It is likely that the manner in which government open data programs are managed and supported over time will need to change as the programs and the people running those programs change. My vision is that the data associated with a program or service will have a variety of different audiences. These audiences — both internal and external to the organization — may have different needs requiring different services delivered with varying degrees of openness and packaging.

Different services for different audiences

When it comes to the “open” data associated with the program, some users will want raw data to do their own thing, some will be satisfied with self-service tools that allow them to interact with the data in various structured or defined ways, and others will be more comfortable relying on the services of intermediaries that understand the data, the tools, and are qualified to interpret the information requirements of those they serve.

Open data programs will need to support all audiences or risk being redundant.

Copyright © 2014 by Dennis D. McDonald