Building trust in personal data ecosystems

Editor’s Note: The World Economic Forum released three reports today as part of its Rethinking Personal Data initiative. Microsoft is on the Steering Board of this multi-year effort to explore frameworks for the use of personal data in order to improve transparency, accountability and empowerment for people worldwide.

A glance at the big data stories that now appear daily reveals a mix of optimism and pessimism about how this technology may affect every aspect of our lives. Ambivalence toward new technology is common. One finding from our recent research on user behavior and attitudes regarding the use of personal data is that while technology enables people to do more, it also creates a dependency on something which they do not fully understand, making them feel powerless. The potential of big data, and the hyper connectedness and element of unpredictability of a data-driven world, exacerbates this.

This perceived asymmetry of power is very real, driven by the information differential that exists between institutions and individuals. Together with recent revelations about government use of data and frequent, extensive data breaches at large retail chains, these elements have combined to deepen the existing concerns about trust.

In 2002, Microsoft created the Trustworthy Computing initiative in recognition of the evolving role of computing in society and the critical need to establish trustworthy computing systems. In 2014, the World Economic Forum’s Rethinking Personal Data initiative has identified this same need to establish trustworthiness for personal data ecosystems. Its latest report on A New Lens for Strengthening Trust is focused on creating sustainable data-driven ecosystems in the current environment through three primary principles: meaningful transparency, reinforced accountability and empowered individuals.

If data equates to power, then transparency is a means to redistribute this power and restore trust in the ecosystem. Transparency for users entails enabling meaningful interactions and decisions on how data related to them can be used. Microsoft has been a leading voice in the industry on institutional transparency, especially as it relates to government requests for customer data. Since early 2013, we’ve been publishing information about the requests we get from governments around the world, and this summer we took legal action against the U.S. government to secure permission to publish details of the orders we receive under U.S. national security authorities. This is a good example of the importance of transparency, since it ensures both a level of accountability for governments around the world and provides some reassurance for customers about the likelihood that their data has been subject to legal demand.

Reinforcing accountability will require verifiable evidence on how data is used throughout personal data ecosystems. In a complex, hyper connected world, such capabilities can only be supported effectively if mechanisms such as the interoperable “metadata tags” discussed in a recent blog are built into the data infrastructure. When implemented as part of an overall trust framework, such techniques can start to address both accountability and auditability requirements to ensure appropriate data usage.

Individuals’ sense of powerlessness and a lack of trust can stem from perceived use of data in ways that are inconsistent with their preferences – and the absence of meaningful interaction between users and service providers. Data use that is contextually consistent with user preferences and provides a delightful user experience are not only welcome, but also provide value-added services. Cortana, Microsoft’s recently announced personal digital assistant, is one such example, delivering context-aware experiences by analyzing personal preferences. The World Economic Forum report on “Trust and Context in User-Centered Data Ecosystems” summarizes the findings from a joint inter-disciplinary project with Microsoft to identify the relative importance of factors that define data context in different countries around the world and how to leverage these insights to develop context-aware systems. Findings from the report are further discussed here.

Realizing these principles requires additional research and technology innovation. A data taxonomy would provide a common language on data usage. A risk framework that can evolve to accommodate the changing definition of risks over time would form a basis for considering acceptable uses of data. Innovations on context-aware systems and interoperable metadata-based infrastructure would provide the underlying foundation for facilitating enforcements of data-use policies throughout the data ecosystems. A personal data-management reference model would offer a structure for discussing the elements necessary for building trustworthy personal data ecosystems.

There remain many unanswered questions and unknowns. For example, machine-learning techniques and algorithms are being applied more broadly to develop insights and predictions that can increasingly be acted upon without human intervention. Although such algorithms can provide great improvements in efficiency, speed and response times, a sole emphasis on technical objectives without consideration of the social consequences will not only lead to untenable systems, but also re-kindle the trust crisis. There are also more fundamental questions about the balance that will need to be struck between personal agency and communal good, how that balance might be determined and by what guiding principles.

Addressing both the short- and long-term challenges will require an inter-disciplinary discourse that can integrate the perspectives and needs of individuals, society, business and policymakers. In the short term, such dialogues should focus on developing mechanisms and evidence that can enhance transparency, accountability and individual empowerment. Microsoft will continue to partner with researchers, industry partners and other organizations to help formulate the issues and drive the discussion forward.

Tags: ,