Numbers and algorithms at society's service

Mortality rates, prevalence indices and models for predicting deaths from COVID-19 are some of the numerical indicators that have become ingrained in our everyday lives during the pandemic. This is yet another example of the importance given to numbers today, a trend that ranges from finance and economics to university rankings and the algorithms that direct online searches. A member of the Open Evidence research group at the Universitat Oberta de Catalunya (UOC) is part of a group of international researchers who are calling for the need to promote an ethics of quantification to investigate the responsible production and use of numbers, both the tangible ones and the invisible ones that support algorithms. With this objective, the researchers also propose the creation of an observatory to improve people's understanding of this situation and to monitor the risks associated with the inappropriate use of numbers, algorithms and mathematical models.

"We live in a world that is significantly structured by numbers, where they convey the truth and construct our reality. Numbers are so deeply entrenched in our existence that we barely reflect on them critically any more. An appeal to return to a society without numbers is unrealistic, but a defence against its excesses is necessary, and possible," explained Andrea Saltelli, the first author of a working paper written by a group of experts from the universities of York (UK), Leiden (Netherlands), Sussex (UK) and Essex (UK) and published by the UCL Institute for Innovation and Public Purpose.

New standards of 'normality'

This exposure to different forms of quantification in our daily lives has increased with the development of technology and digital literacy. This means that the potential to generate a negative impact is also multiplied: a poorly-designed statistical analysis to evaluate medical treatments, for example, could cost lives and waste millions of euros; using a biased algorithm can lead to inequality, such as the crime prediction programme of the Chicago police force which hit poorer neighbourhoods the hardest. Also, a poor predictive model can lead to wrong or simply unjustified political decisions.

These are just some of the consequences of the implementation of erroneous or ethically questionable models and algorithms. However, according to the researchers, the implications of the uncritical adoption of this technology can even lead to a new understanding of what is good and bad, as well as new standards of what is normal or acceptable: "It now seems normal to us that after visiting a shoe store online we are bombarded with advertising about shoes for several weeks. By the same token, once we accept that universities are measured by international rankings, every institution will start studying how they can boost the variables that these rankings measure, with devastating effects on the costs and nature of higher education," noted Andrea Saltelli.

Unequal access to the numbers and calculations on which models and algorithms are based is another aspect highlighted by the authors. At present, powerful governments and large companies have access to all sorts of numbers about the general public, yet the people are denied access to the numbers that could accurately orient governmental decisions or guide corporate behaviour. "Both access to the numbers that matter and the ability to use them or misuse them reflect and reinforce power imbalances in society and the economy," argued Saltelli.

The 'humble' example of meteorological models

On the other side of the coin, the researchers point to weather forecasting as a role model, as it makes useful, short-term predictions that are constantly updated with new information, whereas  any uncertainties are reported with great care. "Despite their huge size and the enormous amount of data they use, the predictions of these models are humble; they warn, for example, that there is a 20 percent chance of rain. However, people trust them and accept the fact that the information they offer might not be entirely accurate," said the researcher.

In this respect, the authors highlight the importance of adopting what they call "humble technologies", which implies considering the sources of ambiguity, indeterminacy and complexity to highlight the ethical dimensions of problems and identify the winners and losers in the distribution of costs and benefits, taking the most vulnerable people into account as well.

An observatory to improve people's lives

In view of this situation, the researchers propose the creation of an observatory that would act as a mediator between science and society in the different fields that inform the public through numbers, such as public health, technology and economics. They said: "The idea would be to select specific cases in which numbers have proved to be incorrect or have been used wrongly, and show this to make people aware of these issues and help them to be more cautious without entirely losing their faith in numbers."

In particular, they underline the fact that the observatory could be very important in creating a much-needed link between the worlds of science and policy, firstly by conveying the "built-in limitations of scientific data" to those who have to make decisions on their basis and, secondly, by channelling the data that is meaningful and can be used safely in public deliberation.

In addition to this proposal, the authors have also undertaken other initiatives such as a seminar on the ethics of quantification with scholars Wendy Espeland and Andy Stirling organized by the University of Leiden which was attended by over 160 delegates.

Related article

Saltelli, A., Andreoni, A., Drechsler W., Ghosh, J., Kattel, R., Kvangraven, I. H., Rafols, I., Reinert, E. S., Stirling, A. and Xu, T. (2021). Why ethics of quantification is needed now. UCL Institute for Innovation and Public Purpose, Working Paper Series (IIPP WP 2021/05). Available at: https://www.ucl.ac.uk/bartlett/public-purpose/wp2021-05

UOC R&I

The UOC's research and innovation (R&I) contribute to solving the challenges facing the global societies of the 21st century by studying how technology and the human and social sciences interact, with a specific focus on the network society, e-learning and e-health.

Over 500 researchers and 51 research groups work among the University's seven faculties and two research centres: the Internet Interdisciplinary Institute (IN3) and the eHealth Center (eHC).

The United Nations' 2030 Agenda for Sustainable Development and open knowledge serve as strategic pillars for the UOC's teaching, research and innovation.

More information: research.uoc.edu. #25yearsUOC