We thank the commentators for helping to move forward the discussion of specific policy options rather than broad general concepts. All three discussants offer important insights and perspectives. Specifically, Dale Gieringer suggests there are lessons to learn from the costs of production and prices in 19th-century India and of medical marijuana in Israel and the Netherlands [1]; Jan van Ours points to the importance of cannabis use dynamics, which are still poorly understood [2]; and Ethan Nadelmann and his colleagues observe that legalization comes in many forms and that some initiatives are ‘far more tightly drafted’ than others [3]. This is precisely the kind of more detailed policy discussion we hoped this paper might stimulate.

Most discussion and even analysis to date has compared the status quo with a nebulous and inadequately specified equilibrium post-legalization. However, the initial policy choices matter, transitory effects matter and the long-term equilibrium may not necessarily reflect the starting point due to mid-course changes and market dynamics. For example, prices may not fall to their final levels for some years because it will take time for the legal industry to expand. Similarly, it could take a generation or more to see the full effects on consumption; birth cohorts that are now over the age of 25 may remain primarily alcohol consumers, even if younger cohorts who grow up with legalized marijuana sustain higher rates of cannabis consumption throughout their lives. These are the kinds of dynamics we can only speculate about today. We concur with van Ours when he says: ‘the most important design choice of legalization is the flexibility to adjustment, allowing for learning by doing’[2].

There is little to disagree with in these comments. We do, however, take issue with two points. First, van Ours asserts that legalization in the United States would not take prices much below levels seen in the mid-1990s [2]. However, like Gierenger, our conclusion is that production costs post-legalization can drop far below current wholesale prices, unless increased artificially by extremely stringent regulations. Hence, while most people might agree with van Ours in principle that ‘taxes should be sufficiently high to discourage cannabis use and sufficiently low to drive out illegal supply’[2], we are skeptical that such a level can be achieved, at least not without designing the entire legalization regime around that objective.

Secondly, while Nadelmann et al. [3] note that we did not discuss the benefits of marijuana use, we also did not address the costs; our essay focused explicitly on design choices for implementing legalization rather than an assessment of the pros and cons of legalization versus prohibition. Our analyses of the latter appear elsewhere [4–6]. While such an assessment might seem, logically, to precede the design task, we think progress on the design front could actually facilitate progress on the assessment front.

Thus, we appreciate Nadelmann et al.'s useful discussion of differences between California Assemblyman Ammiano's bills and California's Proposition 19, and similarly the differences between Proposition 19 and the initiatives likely to appear on the ballots in Colorado and Washington in 2012. Because of these differences, we hope partisans on both sides will stop referring to legalization as if it were a well-defined entity—something about which sweeping statements can sensibly be made. Instead, we hope the literature and public debate will make statements along the lines of: ‘in our estimation, the benefits of legalization along the lines of Proposition 19 would be . . .’ or ‘if marijuana was taxed and advertised like tobacco, the effects would be . . .’. This would promote a more productive debate about marijuana policy.

Finally, drug policy analysts could draw profitably on expertise and experience from related fields. The Kettil Bruun Society has been discussing the nuances of alcohol control for 25 years, suggesting just how difficult it is to get this kind of regulation right. Studying gambling and prostitution markets and policies may also yield useful insights [4]. Coming up with a good design for the regulation of a legal marijuana market is a scientific, as well as political, challenge.

Declarations of interest

JC has consulted for a company whose clients include law firms that provide services to tobacco companies.