Last month the Administration posted perhaps the first github pull request to change federal policy. One WIRED writer was quick to call success, writing here, “By opening up the revisions and the discussions behind them, the White House is making its thinking clear.” But no.
It’s easy to be fooled into believing that a new medium also signals new substance.
No substantive policy change in this pull request
The pull request is a proposed change to a federal memorandum on github regarding open data. The change clarifies when agencies should openly license their data. The memorandum originally said that federal agencies should always use open licensing. But as I pointed out when the memorandum was issued a year ago, that’s not legally possible. Most federal data is not subject to copyright in the first place, and works that are in the public domain can’t be licensed.
The proposed update to the memorandum fixes the Administration’s mistake by adding at the top:
“In instances where government data . . . does not fall squarely within the public domain . . .”
clarifying that open licensing should only be used where copyright applies. Mainly that means when the data was produced by a government contractor. There is no substantive change made in this pull request though. It clarifies the only sensible meaning the original memorandum actually had.
Omits discussion of the substantive issues
If this were the only issue in the paragraph being edited, then I too would call it success. But late last year 14 organizations backed a statement supporting the public domain for government data — not open licensing — and several of us who wrote the letter met with the Administration about the issue. The absence of any mention of the substantive issue in that paragraph should be a red flag for thinking the pull request represents open dialog.
The substantive issue is that the policy condones the copyrighting of any government data, much of which might be used to create or enforce government policy. That’s a serious First Amendment concern. It means that even if journalists can get a hold of some data, they might only be able to share it on terms set by a government agency or even a government contractor. As a broad government policy, the notion of copyrighting government data is ridiculous and flies in the face of our country’s traditions and values. (Note: Forget national security, privacy, etc. This could be data about any mundane policy.)
The pull request omits discussion of this issue, as well as other issues that I and others have discussed with the Administration (as I noted in my reply to the pull request).
Where was the dialog?
There was dialog on these issues, but it wasn’t on github. It was in private in-person meetings, as these things usually are. I and others met with Administration staff in private meetings in August 2013, April 2014, and May 2014. Our discussions each time were thoughtful and productive.
There was plenty of good dialog, but it wasn’t online. I first raised the licensing issue on github a year ago in issues #5 and #64, to which the Administration replied only that they would look into it. The issue was picked up against in issue #257, but again there was no participation in the github issue by the Administration. (There is a lot of dialog in that github repository, but it is about data standards and not policy, and most of the participants in those discussions are government employees or contractors (including myself, in those conversations) — which is a good thing, but not the subject of the WIRED article.)
The pull request posted last month represents the end of a year-long process in which discussions were taking place off-line, and proof that even with github most dialog will still continue to take place off-line.
Lest journalists get confused let’s just be clear that there wasn’t any discussion of substance on github. It was elsewhere, off-line, like normal.
Now I’m just going to be a jerk and red-line the WIRED article because it got a lot of details wrong:
This White House GitHub Experiment Could Help Fix Government
BY ROBERT MCMILLAN
While many of our nation’s problems are quite clear, the way our government addresses them is too often a black box—opaque and closed to all but insiders and lobbyists.
But the White House has taken a remarkable–if small–step toward bringing greater transparency to the legislative process. (“legislative” refers to the legislative branch of government, i.e. Congress. This is an executive-branch memo and thus not related to the legislative process.) For the first time, it has used the GitHub social coding website as a forum for discussing and ultimately changing government policy. With one GitHub “pull request,” it modified (The document has not yet been modified.) theProject Open Data policy document, which spells out how government agencies are supposed to open up access to their data. This represents the fusion of open source software and government policy that open-government advocates have long predicted (#notalladvocates predict this). And it might be a sign of things to come as others—the city of San Francisco, and the New York state senate, to name a couple—bring collaborative government into the light.‘We’re taking a well-known page from the open source playbook: that developing policy in an open and iterative way will create a stronger, more effective product.’
Late last week, Haley Van Dyck at the Office of Management and Budget submitted a pull request that suggested small changes to Project Open data that clarify how agencies think about open source and public domain software (The memo does not cover software. It is about data.). Pull requests are a Silicon Valley innovation. They’re typically used by software developers on GitHub to suggest and discuss changes to code. But they’re also a good tool for tracking changes to complex legal documents, even government regulations.
While Van Dyck’s changes weren’t big, it’s important that these issues were raised and addressed in a public forum where anyone can suggest language for the policy document.(Anyone can, but no one did. The pull request was submitted by the Administration to the Administration’s own document. Let’s wait until they accept a pull request submitted by the public to a policy document.) “We’re taking a well-known page from the open source playbook: that developing policy in an open and iterative way will create a stronger, more effective product. The more we can involve the community, the better that product will be,” said Van Dyck—a senior adviser to the U.S. Chief Information Officer—in an email to WIRED.
The White House will wait a few weeks to review comments to the pull requests, but then Van Dyck’s changes become official government policy with the push of a button. This is open source government: The tonic that could cure the back-room deal. (Most government policy-making involves public comments, review periods, and pushing a button to upload the final policy to the Internet. There is absolutely nothing more open-source about this than the usual agency rule-making process.)
By opening up the revisions (there is no policy-making in our government that doesn’t involve posting revisions) and the discussions behind them (as I mentioned, there was no discussion on github), the White House is making its thinking clear, and there’s an added bonus: The changes are easier to read and understand. Compare Van Dyck’s revisions here, to Rep. Lou Barletta’s proposed changes to existing law in his Emergency Unemployment Compensation Extension Act of 2014. In the GitHub document, you can see the old text struck-through in red and the new additions in green. Congressional bills like Barletta’s, on the other hand, read like uncompiled source code, detailing all the changes to be made but giving the reader no idea what the finished product will look like.(That’s not what uncompiled source code looks like. And ‘compiled’ source code certainly looks no better.)
That makes some bills unreadable, as far as the average citizen is concerned. (This isn’t an apples-to-apples comparison. Modifying 200-year-old statutory law is going to be harder for the “average citizen” to read than modifying a memo written last year.) “The thing that is actually voted on is the edits,” says Ben Balter, GitHub’s government evangelist. He has been working with the feds for years, convincing them to use more open-source software and adopt more of an open-source attitude. “The open government community has been talking about doing stuff like this, but it’s never reached fruition because there weren’t enough stakeholders in government.”
That’s begun to change, Balter says. He says he’s spending more time explaining to federal employees how they can use open source tools and methods. Two years ago, he was still convincing them to give open-source a shot. Now he’s watching the White House merge pull requests.