How should governments build community around their datasets?

Question

The likes of data.gov.uk and data.gov are keen to build a community around their data. How should that best be done?

The main aim is to create a feedback loop, encouraging use of the data and release of the data (updates/more/better). Specifically, these sorts of activities are what users and government might want:

• Users to share:
  • tips for wrangling the data
  • known issues with the metadata and data
  • apps, visualisations & articles related to the dataset
• Government to:
  • respond to questions about the data
  • get feedback about what data is useful and where to improve

Assuming there was budget to try and achieve these things, what would be the most effective way for government to do something? Comments on each dataset page? Host a forum? Twitter? Independent forums e.g. this one? All of these or just one?

Disclaimer: I'm involved in data.gov.uk

Answer 1

Some general advice:

• Invest in community managers and evangelists. The worst thing you can do is create a forum or invite feedback and then not have anyone with a mandate to respond to it in an official capacity; I'm seeing this happen right now with the newly launched project-open-data in which they've invited contributions and have no one to actually respond or accept those contributions

• Create time-boxed campaigns. Governments and policy mandates come and go, so rather than creating an open-ended "chat with us about anything" that may trail off, create 3 or 6 month challenges that you can budget for a high degree of feedback and engagement. This is also helpful for testing your own ability to engage long-term because you can explain your outcomes to superiors/oversight in the short-term. A platform like Challenge.gov can be helpful for this... assuming you actively engage with participants during the challenge periods (many agencies on Challenge.gov are really bad at this and it turns into a black box of disappointment when the winners are announced).

• Write publicly about your experience in managing the platform and performing community engagement. Have a blog that you post to (at least) weekly about how you are helping actual people do stuff. Use first-person pronouns and first names; don't make it a series of press-releases. The Github blog is a good example of this kind of engagement. If you can't regularly write about your own experiences running the platform (whether because of interest, mandate or oversight), you probably won't be able to relate to the experiences of people using it.

• Create in-person "networking" events. Embrace opportunities to meet with the public, on their level. Which means either attending the meetings of existing community/developer groups, or hosting your own (though I'd recommend starting with the former). Chicago's OpenGov Meetup group often has government employees attending (even when they aren't making a presentation!) who can put a public face on things and help further relationships that may otherwise seem transactional or distant when their interactions only take place online.

The advice above is tool and platform agnostic. Specifically, I'd recommend:

1. Put Disqus (or equivalent) on everything; never let a comment go unresponded to (no page should ever have only 1 comment, ever!)
2. Start a development blog (as described above)
3. Continue engaging in spaces like this one

Answer 2

This is a complex opportunity to which many of us in open data are trying to respond. (Disclaimer: I'm the Evangelist for Data.gov and this answer addresses examples from my work there and at NASA.)

There are two broad kinds of communities to engage in around open data. One is the technical community, like many of you here, of users and consumers of the data, developers, analysts, academics, and researchers who are data- and tech-savvy. A second type of community is represented by citizens and businesses who want to learn from or act upon the information gathered from the data.

In addressing the first community, the responder above is correct in that the best way to have conversations with people is to go where they are congregating. Hence, members of the Project Open Data and Data.gov team helping to initiate this Stack Exchange, and many of those people showing up a networking events around the country and the world to share, listen, and learn from others. We do host forums on the Data.gov site but the effort to lift up those to bring people there for conversation isn't as useful as having broader conversations here.

For the second type of community, we have a focused effort building on a rich history of work in communities of practice. These show up as communities on Data.gov organized around areas such as Energy http://energy.data.gov, Research and Development http://research.data.gov, and Public Safety http://safety.data.gov, and local data as well such as Cities http://cities.data.gov (see the listing of all 18 communities at http://www.data.gov/communities/). The communities on Data.gov are public-facing spaces that present data, information, and subject matter knowledge about a single topic from many agencies in one place.  The topics for these communities are chosen based on priorities from the public, agencies based on their mission, or issues of national importance. 

The Data.gov team provides services to help each community determine the following:

1. Vision: What will the community connection and collaboration look like in the future?
2. Leaders:  What core group will help to lead the community?
3. Participants:  Who will participate?  This could include application developers, scientists, analysts, consumers or providers of services, and members of the broader community.  
4. Outcome:  What are the expected outcomes, metrics, and measurements that will show success?  Examples might include challenges around a tough problem in the field, a set of applications or guidelines that help people make better decisions in this area, an API or data library, smart phone apps, and visualizations and mashups.  How will this community work to improve the lives of American citizens?
5. Functionality:  What types of activities will be conducted on the site (forums, blogs, wikis, ranking, rating, challenges, or apps)?
6. Content:   What content should be displayed?  For example, this could be the conduit to other resources within or outside of the government.
7. Interactivity:  What ways will the community interact with the leaders, with each other, and with the public?

I've found the most effective communities look at the entire open data ecosystem, from getting agencies to release useful and usable data to putting it in the hands of citizens and businesses to make decisions. Active leaders (inside and outside of government), diverse participants, online and offline interaction, integration with existing initiatives and businesses, and focused goals help drive these communities forward.

Answer 3

In my mind, the biggest move is to ensure a nearby, front-facing feedback mechanism for the public and to have agency staff actively respond and engage with those who post there.

In the Federal Government, some great examples are the work of the Census Bureau forum, comments on National Renewable Energy Lab's API pages, the feedback section on the Energy Information Administration, and the Developer Feedback & Discussion section on the Federal Communications Commission's developer hub.

(Disclaimer - I am the Sr. API Strategist for GSA)

Answer 4

I work for NASA, and I've wanted to set up some sort of a wiki-like structure for our (existing) community to contribute documentation, code samples, value-added datasets and other comments. ... but I've run into way too many issues to do it:

• You can't take submission from the general public w/out moderation (if someone puts something inappropriate up, security might shut us down ... hard)
• Giving people the ability to log in requires a slew of paperwork and re-approval every year. (so much so that my boss opted to remove every website that required authentication)
• If we're moderating, is that then considered 'government censorship' if we reject a comment or undo an edit?

... my suggestion for folks who have to deal with similar situations is to find someone who's not the government to host it.

One thing that we need to do, however, is establish identifiers such that people can reference or cite the data that we're authoritative for, and so if someone wants to say 'this is software for dealing with (x) data', they have something unabmiguous to reference. (see Laura Wynhold's Linking to Scientific Data: Identity Problems of Unruly and Poorly Bounded Digital Objects).

I participated in the BRDI meeting on Data Attribution, so I'm probably biased, but I think the "Landing Pages" concept is something that we need more communities to do. (I'm still not there myself, as our "archive" isn't considered stable enough for minting DOIs ... I'm trying to find a library to host the content for us. (and outside of government, so I can avoid all of those issues I mentioned).