Development needs collective local action. More science and technical advisors should be supporting that, says research fellow Harry Jones.
Much of the debate on science's role in promoting development concentrates on how it contributes to the process of making policies, focusing on high-level inputs that drive central government decisions, usually on newsworthy topics such as climate change or infectious diseases.
The human and financial resources involved mirror this focus, directed at central government departments, ministries of science and technology, ministries of health, or ministries of the environment.
However, this tendency ignores a great need to link science with development at a sub-national level, and misses opportunities for well-brokered scientific and technical inputs to make genuine and pivotal contributions.
NGOs may be the users and beneficiaries of that knowledge as local partners, but they could also play the brokering role (as could many other organisations such as think tanks or international agencies).
In a large number of the countries I've worked in there is a major gap between policy and implementation — between what is developed in central ministries and what the reality is on the ground. Policies that look good on paper are often not implemented for a host of reasons, including insufficient guidance or budget.
The limitations are compounded by poor technical knowledge outside capital cities. Local government and ministry offices will often have low human resources and limited ability to call in assistance from elsewhere.
But there is reason to believe that filling some of these gaps might make a tangible contribution to better development outcomes.
My colleague David Booth recently argued that many development challenges may be best thought of as 'collective action problems' that require institutions, agreements and actions to be brokered by a variety of groups with a stake in an issue. 
It is well-known that contributing to collective action is often a subnational enterprise.
A great example from Nepal, where I am based, is the Annapurna Conservation Area Project (ACAP). ACAP management is built from the bottom up, on community committees, in a country where governance institutions are not generally strong. It has been cited as an example of a protected area being successfully managed to preserve its ecosystem and benefit the inhabitants. 
Science has a role to play here. It can help to make coalitions and collective action more effective.
Collective agreement and action
The idea that science can help policymakers and practitioners be more effective is nothing new. But viewing the issue as part of a collective action problem adds an important dimension.
Take the example of forest use in Nepal. People have a choice: to harvest as much wood (or other forest products) as they can, or to adhere to common agreements that limit harvesting in order to ensure that stocks can replenish. Because working with others may bring fewer immediate benefits than following one's own self-interest, collective agreements need firm support from local stakeholders if they are to succeed.
This is where scientific and technical guidance can help. It can make collective action more effective. Explaining different options and their potential consequences increases the legitimacy of collective agreements — and the chances that people will stick to them.
Input from scientific research can also help contribute to a shared understanding of problems, leading to common goals.
Work by the late political economist Elinor Ostrom highlights how having a common 'image' of the challenge helps.  She showed that research into the state of underground water reserves and assessments of ocean fisheries were instrumental in promoting agreements (at the subnational level) that preserved these resources.
There are other examples. In the project I manage for the Nepal Centre for Inclusive Growth, we are working to catalyse better delivery of local services and management of tourism in Pokhara and Annapurna through partnerships between local business and government.
Brokering science and technical knowledge on issues such as renewable energy, environmental protection and waste management has been a key element. The initiative has not directly funded partnership projects, but has instead brought in technical expertise, as needed, to support the priorities chosen by local stakeholders.
It has helped to bring divided groups onto the 'same page' about development issues — for example, giving a clearer and more neutral understanding of the causes and consequences of waste management practices. And it has strengthened commitments to collective action by making sure that businesses' hard-earned contributions to local public goods and services are well spent.
Changing donor focus
But how can scientific and technical capacity be promoted at a sub-national level?
Decentralisation has an important role to play in building a culture of evidence-informed decision making at a subnational level. However, experience in sectors such as health and education shows that retaining skilled technical staff outside economic hubs is a major challenge for governments with weak institutions and low public sector pay. It is hard to ensure skilled health professionals work in remote rural areas.
Our work in Nepal shows that aid programmes that broker scientific and technical knowledge at a sub-national level can help catalyse improvements in local development. It's not the only way — governments and partners could do more to trial different models and explore options for institutionalising such services.
Donors' assistance will be needed to do this, but it needs a departure from the typical focus. Instead of agency technical advisors spending all their time based in capital cities, they would be more effective brokers if they were embedded with local partners.
Although this doesn't guarantee impact, with careful targeting,  scientific and technical knowledge can be deployed in areas where the chance of making a tangible impact will be highest.
Harry Jones is a research fellow at ODI's Research and Policy in Development programme. He is currently based in Nepal and can be contacted at [email protected].