Gregory Maciag

Standards: Infinite ROI

New legislation which reuses procedures from a previous Act will deliver the USA's international treaty obligations on child support recovery. I mention this because the legislation is said to reuse “no-cost data standardization provisions” from three other social programs.

As well as making sure attempts to recover support payments don't stop at national borders, the move will save up to $500,000 over 10 years by improving efficiency.  Cost zero, return one half million. That's an infinite Return on Investment. Crazy not to.  Dave Camp

The Mosaic Effect

Back in the day, data was a topic for geeks – and these were not even alpha geeks (sorry guys). Now data is everybody's concern, because data about everybody is getting everywhere.

The British commentator Simon Jenkins recently devoted a column to the dangers of government data getting into the wrong hands. As well as likening the situation to the early days of atomic power, he said: “Data protection is a blazing contradiction in terms.” Now, data protection is not a contradiction in terms, let alone a blazing one. Jenkins probably means he doesn't trust claims made by technologists that data protection is possible.

Though I think his tone is a little hysterical, it's true that not enough has been done to guarantee the anonymity of personal data released in the public domain. And yet this issue has been flagged up by government itself. The so-called mosaic effect is when you combine two or more data sources to reveal a pattern that doesn't exist in any single source.

It was the Health and Human Services Department that discovered this, when it combined data about Medicare payments with Census data.  This inadvertently led to some health information about people in low population areas being potentially identifiable.

The “mosaic” terminology comes from database marketing, the field that discovered long ago how combining different data sets could lead to increasingly accurate predictions about people's income or propensity to buy certain products and services.

You could argue that data standards only help people to create mosaic views from multiple data sources. You might conclude that the more disparity we have in public data sets, the safer the public will be.

I don't think that's right. We actually need data standards in order to map potential mosaics before the hackers get to them. Then we can restrict the relevant data sets accordingly. Standards will let us see actual business relationships – such as the ones revealed by the Medicare/Census mashup – ahead of data release. In this way, standards can help us safeguard privacy.  Digital Revolutiong   Open Data Policy

And That's an Order!

The White House has said government agencies must collect and publish new information in open, machine-readable and, whenever possible, non-proprietary formats.

President Obama wants more and more government data published online, so that new businesses can grow around its exploitation. The model given for this is the release of GPS data, which led to the creation of navigation and location based services.

So, non-proprietary standards where possible. This is kind of fuzzy. But the White House is also promising a reboot of its data.gov site, which should include better data handling tools and APIs. Maybe we'll see some de facto standards for open data emerge from this source.  NextGov