What is Known From a Network? Privacy & pandemics in the digital-age
Kelsie Nabben, June 2020
As I’m sure you are aware by now, you are not just a person. You too are a rich data set of digital exhaust. So you may as well think about how you would like that to go as a society.
Today, I attended a research group on ‘Privacy and Pandemics’, coordinated by the University of Melbourne. One of the most interesting talks, by Mark Andrejevic, was on technology platforms as “inferential infrastructure”. Mark was speaking in reference to commercial technology platforms and health data, in terms of the fate of data.
The idea of ‘inferential infrastructure’ links to some recent research of mine on data analysis of human networks in the context of disaster and digital infrastructures. A modified excerpt of the full-length journal article which is being submitted for publications has been shared below, as it pertains to current discussions;
- — — — — — — — -
In the (very real) context of COVID-19, including the power of digitisation and ‘data for good’, we thought it would be worthwhile to seek answers to the question, what can be known about people from network data?
Among the most interesting observations is the ability to model human behaviour, such as mobility, both now and well into the future, based on snapshot of data.
Data sets reveal incredible amounts of information. For example, mobile phone applications collect background information when downloaded, such as phone model, device name, other apps running, WiFi routers accessed, etc. WiFi information can be easily converted into precise geographical position. If router information is known at a single point in time, it can produce high resolution mobility data which makes it straightforward to predict persons mobility six months later. Even low-resolution data allows accurate inference of human mobility patterns. De-anonymising network data has been possible for years. When multiple points of information are correlated, a lot of information can be ascertained about an individual.
What must be considered is the implications of data networks at scale, and thus the trade-off between the value of creating this data, versus the risks.
Revealing network connections between as little 1% of individuals in a network, or 36% of one person’s communications, can lead to observation of 46% of all communications in a network. This means that a motivated attacker only needs to compromise 1% of smartphones across a population, to monitor the location of more than half of the population of a major city. This technique doesn’t even rely on compromising data that has been collected and stored on central data-servers. In other words, your smart-phone is very smart .
At scale, the lack of privacy and subsequent vulnerability needs to be a significant concern to the state and the broader population at large.
Privacy is structural, meaning that the power of inference is amplified at scale. Although past behaviour is not always an accurate predictor indicator of the future and humans can be unpredictable (thanks Kobi Leins for making this point during the call), the emergent space of data analytics and modelling requires much deeper inspection before we rely on it to infer human behaviour and act on it.
COVID-19 is the first global pandemic of the digital-age. The trend in digital-political responses to COVID-19 is transitioning from contact-tracing to risk assessment and prediction, such as private instantiations of contact-tracing by employers and digital immunity passports.
What are the intended and unintended consequences inferential infrastructure systems, that are designed to inter-link data to infer people’s behaviour?
In this context, critical analysis of the technical and governance design of digital infrastructures is essential. We need better solutions, through considered, deliberate, interdisciplinary design of digital infrastructures.
Otherwise, we risk building the next generation of digital infrastructure that lacks transparency and accountability to both creators or attackers.
— — — — — — — -
I’m doing my PhD and writing stuff helps. If these ideas are of interest, feel free to get in touch.
