Opening up environmental footprint calculators for transparency and user privacy

Life cycle thinking, in the form of environmental "footprints" such as water or carbon footprints, has become an essential concept in society's response to growing ecological crises. An online footprint calculator is a web service that accepts information about a user’s consumption habits and uses it to generate an indicator score, such as a carbon footprint. While these services are increasingly common across the Internet, their technical design causes a number of significant problems.

A footprint calculation is a form of life cycle assessment (LCA), a complex scientific practice for quantifying environmental impact. Generating a meaningful LCA study is tricky, and the simplified computations that go into a footprint calculator mean that the result is often unreliable or inaccurate. Calculations that are valid under one set of circumstances may be invalid in another. Validating and interpreting the result is difficult because calculators often operate like a “black box” - the methods used to compute it are poorly documented and difficult to update. Finally, the way these calculators are designed almost always requires users to abandon control of their private information, sharing the full details of their consumption with the remote server in order to use the tools.

In this paper I envision an alternative structure for footprint calculators and other LCA computations, one that inverts the traditional model of the web-based service. At the heart of every footprint calculation is a “product model” – a particular type of algorithm that describes how user information is linked to data about the environmental impacts of activities. The design of this product model determines how the indicator score is computed, yet it is often entirely hidden from view. In the proposed approach, these product models are elevated from the secret inner workings of black-box calculator tools to data products in their own right, capable of being inspected, adjusted, and improved over time. Documented product models can aid users in interpreting their footprint scores and can even enable users to customize their own models to improve their accuracy.

Most of all, by making use of shared and published product models, users can retain control of their private data. Instead of submitting their secrets to a service in the cloud, they can “check out” the models and run them from the privacy of their own devices. The creators of footprinting services will no longer need to manage users’ private data, and instead focus on improving the tools’ technical capabilities and interpretive support. And tool users can find new ways to collaborate—by reviewing and improving product models and by using novel communication technologies to perform cooperative calculations using secure multiparty computation.

This paper describes a vision for transforming the way footprint calculators and other LCA computations are done using an open network of distributed, private data sources. Readers will learn of some of the drawbacks of traditional footprint calculators and gain a high-level understanding of their inner workings. A graphical walk-through of a “footprint model library” called CarbonEq is presented in the appendix and online to illustrate the interpretive support that a product model can provide to a footprint tool. The paper ends with a list of principles and key concepts for opening up footprint calculators.