One of the problems with the current ways we go about solving social problems is the uniform way in which we approach them. This is largely because we have handed over to government the responsibility for dealing with an ever larger number of them. As scientific research funding becomes increasingly dominated by government there is a growing danger that it too will suffer from the same approach, wherein there's little room for dissidence, for questioning the consensus.
But science - real science - isn't a consensual process. Government is inherently uniform, top-down and one-size-fits-all in its approach. It stifles diversity. In the private sector, diversity and creative destruction are essential for generating wealth. Similarly in science: failed hypotheses must be seen to fail and allowed to die, not kept on life support with funding supplied by politicians who depend on vested interests, the status quo, for their tax revenue or campaign funding.
Perhaps research funding needs to be determined by, and integrated into, our broad, long-term, social goals. So instead of funding research into, say, cutting back carbon dioxide emissions with the aim of preventing climate change, we should fund the prevention of climate change. If we issued Climate Stability Bonds, then people would have incentives to find the most cost-effective ways of stabilizing the climate. That could mean funding scientific research into cutting back carbon emissions, but that decision would be made by people motivated to find the most efficient ways of achieving the overall goal. Not, as at present, by bureaucracies, academic, national and international, whose interests appear to range from, at best, securing more funding for themselves to, at worst and in the time-honoured way of most global aid, transferring funds from the poor in the rich countries to the rich in the poor countries.
Under a Social Policy Bond regime, where activities are entirely subordinate to society's desired outcomes, people would have incentives to investigate hypotheses that at first sight seem far-fetched (see here for one example), even if they would prove unpopular if proven. And once they are proven they would have incentives to act on that knowledge ... rather than conceal it.