Co-founded by Paul Duan in 2014, Bayes Impact is an international NGO with offices in the United States and France. Bayes Impact is developing a methodology for the creation of citizen-led and citizen - owned social services that is financed by government institutions.
A key element in Paul’s plan to create citizen-led services is to tackle what he calls the “giants” of societal issues. Issues such as healthcare, social justice and unemployment which are universally acknowledged as priorities, yet impossible to solve through generalized programs which if created, often fail to help those most vulnerable and in need. However, addressing these “giants” - as Paul calls them – provides Bayes Impact with some key advantages. For instance, governments often collect massive amounts of data on such social issues, making them the best opportunities to showcase the ability of data science engineering to re-organize a reservoir of scattered and incoherent information into highly tailored and useful information. Paul believes that access and navigation of this data is best done through partnership with the governments that collect it, rather than independent of them. By partnering with government agencies to access the data and co-design a service using that data, Paul argues that he is leveraging the government’s ability to collect highly detailed information from its citizens that would be unmatched through independent or commercial avenues. Bayes Impact has already piloted partnerships with the Sutter Health system (one of the largest hospital networks in California), the U.S Department of Veteran Affairs Services and France’s national employment agency – Pole Emploi.
An additional advantage to addressing the “giants” of social issues is that it allows for Bayes Impact to position itself as an advocate of digital social services among the citizen sector and to gain significant attention within the media. Paul believes this can stimulate “grassroots” movements at the local and national level around the concept of citizen-led social services while forcing government legislators, policy makers and influencers to take notice and meet the demands of the public. To date, Paul and his work with Bayes Impact has been featured by: Forbes.com - 30 under 30 social entrepreneurs for 2016, The Wall Street Journal, and Le Monde (one of the top national media outlets in France). With this type of media exposure Paul Duan has been able to share his vision of citizen-led services directly with the President of France - François Hollande through special invitation, to discuss unemployment in France. These opportunities are not just an attempt at media coverage, but an opportunity to grow the Bayes impact vision while leveraging the power of the media to influence political and administrative players to become more sensitized to a shifting role in how social services can and should be provided.
Just as important as being able to influence governments, is the strategic opportunity to co-design a social service product with current governments and institutions. Paul argues that working alongside government officials helps him and his team to better understand and navigate the institutional challenges and barriers that might bar a “great” product from being successfully implemented or launched to the citizen base it is meant to support. Paul identifies that this type of intimacy of a project based team, allows him and his Bayes colleagues to uncover the landscape of political dynamics and institutional relationships not always available through more public channels. This also helps to build a foundation of trust and influence between his team and their institutional counterparts. Such access to the inner-workings of these types of bureaucracies is not available when if his team is simply “commissioned” to deliver a product. These experiences have evolved the Bayes Impact partnership contract to include two critical and currently non-negotiable clauses. The first, is that all partnerships must be fully funded and finite, on a project by project basis. This is different from more traditional funding models, where funding is often distributed and renewed on an annual basis, over renewable contracts. This, Paul believes leaves valuable services vulnerable to administration changes and “power-plays”. It is important that a full commitment up front to “finish what is started” is understood by all partners. The second clause, is that sole ownership and all proprietary information from the final product is owned exclusively by Bayes Impact - which the NGO then promptly open sources and distributes freely upon completion. Through open sourcing all algorithms, formulas and the entire project methodology, Paul affirms that the product and the service becomes the ownership of everyone, nothing is proprietary, and the product can never be “pulled back” by a leadership change.
Once Paul has these strategic partnerships and contracts with agencies in place, Bayes Impact works to co-design and co-develop a product (often an online platform tied to a specialized algorithm) which best utilizes the data available. The key is to be able to produce a nuanced and tailored solution for any person that uses the product. For example, through their partnership with the Sutter Health system, Bayes Impact designed a program that predicted the re-admission risk rates of individual patient’s in order to improve Sutter’s care management procedures. The partnership with the U.S Department of Veteran Affairs, enabled Bayes Impact to co-design algorithms that helped the office identify mental health and unemployment issues for more than 21 million US veterans. Yet, Bayes’ biggest accomplishment to date is through its partnership with the government of France to create an unemployment service tool which not only provides a personalized guide for citizens through their employment journey, but also adapts its service provision through a simplified form of artificial intelligence (AI) to improve its recommendations and support to subsequent users based on the outcomes and success rates of previous users. This unprecedented partnership with the French government, gives Bayes Impact simultaneous access to multiple employment administrations and branches within France which are currently siloed. Communication and leadership between the branches is fragmented, which leads to frustration from citizens attempting to navigate employment services in France. The platform, which launched in November 2016, has already had more than 85,000 citizens registered, with positive qualitative feedback already being collected by restraints that have been able to find a job, Yet, Paul is much more excited about creating a system that can be continuously improved by anyone and everyone who uses it.
Looking to the future, Bayes Impact is launching a campaign to get one million users in France to benefit from the new employment platform, and is partnering with the Abul Latif Jameel Poverty Action Lab (J-PAL) to conduct an independent review of the impact of a citizen-led service, with results expected to be reported in February 2018. Already built, the platform is the first to fall under Bayes new partnership structure which means all algorithms are open sourced and the platform is built to resist any changes in national leadership, which is timely given the arrival of French elections in May 2017. Paul sees France a “key” leveraging country in stimulating citizen-led services across Europe. His agreement with France has already led to talks with key influencers in Belgium, Switzerland and Luxemburg. Paul hope to replicate and adapt the success to economic and employment issues there, ultimately positioning Bayes Impact as the UN of data sciences and citizen led/owned services.