In writing my first post in this series, I explained that ICT seems to be referenced in migration policy frameworks almost exclusively for its role in data collection, mostly associated with border management. You may – likely influenced by your political persuasions and where you live – see this as a logical part of global migration management, or worrying in terms of potential for abuse, or perhaps you fall somewhere in the middle of the broad spectrum of ideology when it comes to migration management. But what about uses for such data collection that could be removed from thorny political rhetoric about sovereignty, solidarity, borders and points systems? How can data on movements of people and technology be used to protect the vulnerable and support humanitarian action? The Danish Refugee Council (DRC) has one possible answer: Foresight.
The capital F is important: Foresight is software developed to predict how crises might impact refugee flows as a specific element of global migration. In an article that is part of Der Spiegel International’s humanitarian series funded by the Bill and Melinda Gates Foundation, DRC’s general secretary Charlotte Slente explains that Foresight “helps us predict more quickly what will happen so that we can plan better and intervene earlier in a humanitarian crisis.”
Algorithms and data analysis are not entirely new in the humanitarian sector, but their use seems to be finally receiving the attention that their potential deserves. Whilst other examples include correlating local weather changes with market fluctuations and social media posts to flag potential fluctuations in migratory flows at local level, DRC’s Foresight software analyses data from over 120 sources to allow forecasting to be applied worldwide. Foresight’s machine-learning system identifies patterns to predict how a particular variable (or combination) might affect migratory flows, the result of which, it is hoped, will assist governments and humanitarian organisations to prepare more effective and quicker responses to crises. Although attempts have been made in the past to use algorithms for such forecasting, Foresight is designed to be more precise and permit users to apply 15 variables (such as economic situations, human rights violations, unemployment, corruption, public services, food security) in a causality model for a more comprehensive forecast of the impact on people in an affected area. After an iterative development process over two and a half years, testing of the algorithm with historical examples now records an error range of 8-10% – an impressive level of accuracy to have in the humanitarian response toolkit.
Big Data: opportunities vs. challenges
Der Spiegel’s article focuses on the use of Big Data to develop Foresight but acknowledges the limitations of the datasets used as they do not always reflect the local reality. Should Foresight continue to use such datasets exclusively, it would not expect to improve its error range and the practical use of the software would be limited. However, DRC intends to make the software available to other humanitarian actors and allow them to manipulate the variables according to their needs, whilst feeding in additional data. This crowd-sourcing approach has the potential to ensure Foresight grows in accuracy and therefore usefulness. Whilst until now mining Big Data for humanitarian purposes may have been stilted by lack of long-term investment (the question of project-based funding rears its ugly head again) and technical expertise within the humanitarian sector, here we see a hopeful example of how ICT can be leveraged by the humanitarian community as a whole, if a long-term approach can be sustained.
If, as DRC hopes, Foresight can be used in the future to enhance the effectiveness of humanitarian response, by allowing responders to better plan resources and build local partnerships more quickly, the positive potential of tracking migratory flows for more than just national border control is evident. One might even predict coordinating and consolidating comprehensive and comparative data will become more of a humanitarian priority than a political one, with the priority to piece together the migratory flow jigsaw to enhance protection of vulnerable people worldwide. Bearing in mind this grandiose optimism for prioritising migratory data, I find myself asking whether the algorithm goes beyond statistics and how people can or have been involved in the development process to make it as human as possible.
As it turns out, Foresight is the latest iteration of what began as a pilot project via an Impact Grant from IBM’s Corporate Social Responsibility programme: Mixed Migration foresight (MM4Sight). Not only is MM4Sight therefore an interesting example of a public-private partnership for humanitarian objectives – with IBM supporting pro-bono with technology, model development and training – but the one-pager of the project on IBM’s website also explains that in addition to open-access datasets, it uses the results of over 15,000 in-depth questionnaires collated by DRC from people involved in mixed migration flows (including refugees, migrants, smugglers) around the world. The very design of Foresight therefore incorporates lived-experience of migration so as to include choices made by migrants in the analysis of the Big Data and aids identification of patterns to be applied to assist future scenarios.
Evidently, this assumes that the relationship between migration and push factors witnessed and recorded in the past will apply in the future, a limitation that is impossible to avoid but important to bear in mind – and perhaps one of the reasons that DRC categorically state that Foresight is designed as a supporting tool to be combined with the analysis of development experts, crisis workers and local specialists. Limitations must also be acknowledged in terms of the datasets used which – as mentioned above – may not reflect the local reality in the present day, as well as historical datasets being of mixed quality. Not least are the inaccuracies in data around irregular migration, the full extent of which is neither documentable, nor observable because of its nature. As the application is used by a broader spectrum of actors that add new data, these inaccuracies will hopefully be reduced and perhaps new uses will be developed for the forecasting software, diversifying the potential of such software for migration and development.
One final challenge, however, is what other uses may come out of such predictive software being open access? As I insinuated at the beginning of this post, if using technology to collate migration data can be leveraged for ‘ill’, the same is surely logical of forecasts on migration – fear mongering and anti-migration rhetoric and policies spring to mind. Pragmatically, Rana Novack, ‘solution owner’ of IBM’s Refugee & Migration Predictive Analytics Solution (the precursor to Foresight), responds to this challenge:
We need to make sure that the tools are available for the people who want to do the right thing. I think with any tool, any technology, there’s the opportunity for someone to use it with malintent, but that doesn’t mean that we shouldn’t create these tools so that we can be a force for good in the world. Ultimately, it’s up to us to choose how we want to use these tools. – Rana Novack
Rana’s response sums up succinctly why it is important to highlight the use of technology for ‘good’ – as in the case of Foresight – and not just technology as glorified information management. Foresight serves not only as a tool to enhance humanitarian response, but also as an example of a public-private partnership that makes the most of the expertise of each, with a long-term perspective on its potential. It is also a reminder that Big Data is about people and should therefore be leveraged to their benefit, rather than being consigned to commercial purposes. Humanitarian actors, policymakers and civil society should strive to be as forward thinking as the private sector in developing solutions whilst simultaneously advocating for people-centred design in technology solutions.
I look forward to seeing how Foresight is honed by DRC and disseminated for use among the humanitarian community and beyond. It may not be a crystal ball, but if identifying patterns in mixed migration flows can help governments, humanitarians, communities and individuals to better prepare and respond to crises that might trigger movements of people within or across borders, then it is a worthy tool to invest in collectively. The migration and development world, as well as ComDev observers, should seek to identify, amplify and emulate such innovation.
Related reading:
Predicting refugee movements? There’s an app for that [Der Spiegel International]
Where will future migrants come from? [Phys.org]
Mixed Migration Review 2019 [Mixed Migration Centre]
Assessing Immigration Scenarios for the EU in 2030: Relevant, realistic and reliable? [IOM]
