<p>Performance-based regulation establishes mandatory goals rather than enforcing prescriptive standards. Performance-based regulation has become popular over the past two decades as an alternative to prescriptive regulation, as it holds out the promise of simultaneously achieving health, safety and environmental outcomes while facilitating innovation and reducing regulatory costs. In the early 1990s New Zealand adopted a performance-based building control regime. This demonstrably failed and was replaced in 2004 with a new regime, still performance-based but more conservative. Using legal determinations, adjudications and court cases, and reviews of the failures, contributing factors have been identified. An assessment has been made of the extent to which these factors can be attributed to the performance philosophy and features of the regime. Strategies to addresses the weaknesses of performance-based regulation have been explored. The change from a standards-based regulatory regime, where technology shifts are on the margin and occur through a process of incremental trial-and-error, to a performance-based regime, displaced traditional institutions for aggregating knowledge required for risk-based decision-making. At the same time, the new performance-based regime was permissive of greater technology shifts, which demands more of decision-makers who are operating in an environment of inevitable uncertainty. The significance of the regime change was not well understood and new institutions did not evolve. Reverting to traditional institutions is not an option as they are inherently conservative and therefore innovation as one of the normative benefits of performance-based regulation is likely to be constrained. New institutions are required to aggregate knowledge, but also permit decisions that enable the technology threshold to be pushed out in situations where it is not possible to measure accurately how a novel technology will perform in all of the circumstances of its use, and failure in the field is a possibility. This requires knowledge that is both technical and evaluative. Technical knowledge is more than science, but increasingly knowledge in other domains such as psychology, economics, and both domestic and international law. Evaluative knowledge is that which is required to assess risks and consequences. This study explores two strategies for resolving the challenges of decision-making in a permissive performance-based regulatory environment: improving the predicative capability of decision-making systems through the better application of the intuitive judgment associated with expertise and wisdom, and treating novel technologies as explicit experiments. Both strategies show promise, but may be difficult to implement. If the conditions for materially pushing out the thresholds of technology while managing the risks cannot be met, it may be necessary to revert to incremental trial-and-error in high-risk areas. This does not preclude innovation, but it will be at a slower rate.</p>
Generalized reinforcement learning for building control using Behavioral Cloning
