Rondinelli suggests that the policy process can be seen as a sort of social experiment. The policy process is less a matter of prediction, and more a matter of trial and error. Actors need to be able to readjust and adapt to unforeseen circumstances as the process develops. To address complexity and uncertainty of development policies, Rondinelli argues that 'success requires flexibility in planning and design, opportunity to adjust plans as projects progress, and continuous redesign during implementation'. This requires that formulation and implementation are more closely integrated in order to reduce the impact of uncertainties and unknowns. Policymakers can try different types of projects, including:

• Experimental projects: that investigate possible courses of action are useful when uncertainty exists about the problems, feasible solutions, and the effects of different interventions).
  
• Pilot projects: 'that test alternatives and identify conditions under which interventions are more or less effective' are useful 'when the problem or objective of a policy is well defined or when much is already known about the effects of small-scale experiments'. Pilot projects are often used to test results of experimental projects in other contexts. They often require strong evaluation and feedback mechanisms.
  
• Demonstration projects: can be useful 'to exhibit the effectiveness and to increase the acceptability of new methods, techniques and forms of social interaction on a broader scale.'
  
• Replication, diffusion or production projects: involve widespread replication after sufficient is knowledge obtained in order to 'test full-scale production technology and to organise an effective delivery system for dissemination results or distributing outputs.' Some experimental and pilot projects should not lead to full-scale replication if they are found to be useful only in specific conditions or ineffective.

Back to EBP bibliography index