Comments on: Developing Big Data Analysis for Public Benefit

By: Lynn Eakin

Lynn Eakin — Fri, 23 Sep 2016 16:05:50 +0000

In reply to Ted Richmond. Love the discussion. These are excellent points. Data use is not simple or benign. That is why it needs to be publicly owned and not proprietary.

By: Ted Richmond

Ted Richmond — Tue, 20 Sep 2016 15:07:02 +0000

Lynn has started an important discussion concerning the use of Big Data for the public good. There are undoubtedly numerous current and potential benefits in using Big Data, ranging from tracking municipal water bills to plan services and promote conservation, to analyzing patterns of HIV infection in Africa in order to better plan medical interventions.
Diane has introduced a note of caution for various practical reasons, and I would agree with her observations. But we also need to consider the nature of Big Data more broadly.
My first issue concerns definition. Lynn has used the widely-accepted one that identifies the nature of Big Data in its size, and in the complexity of analysis that is required. But in an age where our smartphones have more computing power than a supercomputer of the 1970s, does this definition tell us anything useful?
I find it more useful to think of Big Data as large data sets that are collected for one purpose, and then used (actually or potentially) for another. That is why one US General could say that ‘We use Big Data to kill people’, referring to the targeting of suspected insurgents with GIS data from their cell phones (eSet security newsletter, 2016). Another example: researchers from the US and China are studying data from transit smart cards used in Beijing to locate pickpockets and purse snatchers (The Economist, August 20 2016, ‘Cutpurse Capers’). Although the results of this project are ambiguous, the researchers are already planning to extend their studies to ‘asocial groups’ such as ‘alcoholics, drug-users, homeless people, and drug dealers’. We should not assume that the use of Big Data will be benign.
My second concern is with the difference between Big Data and Open Data. There is a natural public interest in the potential knowledge to be gained from access to currently-restricted data sets, particularly the vast holdings of administrative data by provincial and federal governments. But the issues around this Open Data are distinct. Most government data is locked away, and will remain so, for various reasons including legacy computer systems, and privacy legislation. Governments like to talk about Open Data and Open Government, but in Canada little is happening on this front. At the federal level the main users of Big Data / Open Data are the police and security services (Redden, ‘Big Data as a System of Knowledge, Investigating Canadian Governance’, 2015).
From a public policy perspective, perhaps the most useful initiative for community organizations would be to ally with the research community and establish some formal liaison with governments (all levels) to discuss the potential of Open Data. What data sets are potentially available (given legal and technical restrictions), what could be learned, and what costs would be involved? Some answers to these questions would move us along the path towards public benefit from Big Data.

By: Diane Dyson

Diane Dyson — Wed, 03 Aug 2016 15:49:49 +0000

I have been part of conversations like this over the years. Data quality and openness are indeed important principles, and as a researcher I delight in thinking what we could learn. But I also know that big data fails not simply on a technical level of issues like data quality, but because it runs smack dab into some other public and conflicting protections, such as privacy and population/group vulnerabilities. Legislation and issues of consent stand in the way as a barrier. Over-surveillance is also a risk. Our sector will need to be able to demonstrate to the public that the use of their data does no harm and is for a good greater than our own narrower interests. We need to be able to stand any resultant critique.
I say all this as someone from an education background where testing and evaluating (a.k.a. grading) are key system drivers – and are fairly critiqued for being that. I say this having heard a cabinet minister urge the community sector to behave more like the health sector, to build a strong evidence base, to describe exposures and dosages, as though this is the way we worked.
We must also be quite sure what is it we would be trying to measure and to what end. Will some of us be weighing demographic populations against each other (too old, too far gone, etc.) as though it reflected solely on those populations? For an example of this, see how low-income schools are ranked and rated as though it was a level playing field. Or how about the the time a donor wanted to know what age group should receive his donation, for the biggest impact, you understand. All others be damned.
Or, more properly, will we use big data analysis as a test of the system, its performance and its failings? Will we be able to do this without creating a further burden on those already at the edge? Will we have both the skill and the will to do these things?
If we can do these things, then we will be able to make a case for an investment in big data.

By: Andrew Benson

Andrew Benson — Wed, 06 Jul 2016 18:52:00 +0000

Terrific article, Lynn. Congrats to ONN and its non-profit members in pushing the open data conversation in the right direction.