Coordinating all the departments and agencies involved in federal drug control policy wouldn’t be possible without the Office of National Drug Control Policy (ONDCP), which plays a crucial role in keeping everyone on the same page. At the helm of the ONDCP is the “drug czar” — the director — who also commands an important bully pulpit in shaping the way people understand and respond to drug problems.

On Jan. 20, one of the challenges that President-elect Trump will face is steering the Office of National Drug Control. Here are three broad suggestions for how he should approach that challenge:

1. Manage the opioid crisis.

If the Office of National Drug Control Policy tackles nothing else in the next four years, tackling the opioid crisis should be priority number one. By a wide margin, the most pressing agenda for national drug control policy is to get the opioid crisis under control. A decade or so ago, it looked like the opioids — a class that includes poppy-based opiates like heroin as well as synthetics like oxycodone and fentanyl — were becoming a minor footnote in drug policy, as there were relatively few new initiates replacing the addicts of the Vietnam era’s heroin epidemic. Now, astonishingly, heroin-related deaths have just surpassed firearm deaths in America; all drug overdoses combined kill more Americans each year than AIDS did at its peak. We have a variety of treatment options for opioid dependence, but they are only effective if they are deployed where they are needed. Encouragingly, the recently enacted 21st Century Cures Act has earmarked significant new funding to combat the opioid crisis, Even so, if the Affordable Care Act is repealed, the treatment gap could grow larger, not smaller.

2. Provide federal leadership for marijuana regulation.

Even with a vastly larger user base, marijuana is less consequential than the opioids for public health. On the other hand, we are at a historically important moment in the emergence of a major new industry. The new administration has three paths available. First, it can try to roll back state changes in marijuana law. But with 1 in 5 Americans residing in a state that’s legalized marijuana, with over half of the states legalizing some form of medical marijuana, and with a majority of all Americans favoring legalization in various recent polls, that would be a deeply unpopular and distracting fight.

Second, the administration can continue the current laissez-faire policy. One advantage of that approach is that it allows us to learn from the variations in state policies (see below). Indeed, in the next few years we will learn a great deal that we didn’t know about marijuana consumption, marijuana laws, and their consequences.

Third, the federal government can play an active role in promoting responsible regulation of marijuana sales. Maintaining a stark dichotomy between prohibition and a free market once served the rhetorical purpose of making marijuana legalization look unthinkable. But now, that dichotomous thinking has left an aggressively profit-seeking liquor-style commercial market as the default choice for the great many Americans who want change. Activists and state legislators who want reform need to be encouraged to consider the full spectrum of possibilities for regulating marijuana without commercialization, including non-profit or for-benefit user co-ops or simply allowing home cultivation.

3. Encourage experimentation.

The word “experiment” has two different meanings:

1. To innovate; to try new things.
2. To “conduct an experiment” — a rigorous scientific test of a hypothesis, by controlling all extraneous factors in a laboratory and/or randomly assigning experimental participants to different treatment conditions.

We fail to experiment in drug policy in both senses of the word. We hardly ever innovate, and many of our innovations are minor variations on policies that would have been familiar to anyone in the Nixon or Ford administrations of my youth. And we almost never conduct experimental tests of drug policies; in fact, we tend to implement policies in ways that are almost evaluation-proof. We often implement them everywhere, all at once, e.g., rolling out a national mass media campaign rather than testing different messages in different locations. This is in stark contrast to modern business practice, where constant innovation and testing are essential to stay competitive. That is not to say that government (or drug policy) can or should be run like a business; rather, the point is that modern businesses have demonstrated how to speed progress by “learning by doing.”

The gold standard for testing is the controlled experiment with random assignment (e.g., double-blind clinical trials). There are good precedents in for randomized experimentation in drug policy and other areas of social policy, but they are far too rare.

To be sure, there are often strong ethical arguments against some forms of experimentation. Medical research shows that there are often ways to address such concerns, especially when the changes are incremental and there is uncertainty about what works best. Also, random assignment is a kind of lottery, and lotteries are often an ethically defensible allocation scheme for government interventions that are scarce or backlogged or slow to roll out. And there are also many missed opportunities to conduct good “quasi-experiments” that take statistical advantage of the existence of waiting lists, eligibility criteria and other exigencies of policy implementation to learn more about what works. We can also learn a great deal from comprehensive longitudinal studies of drug use through the life course, like the newly launched Adolescent Brain Cognitive Development (ABCD) Study. Such studies require sustained funding and commitment to fulfill their promise.

The truth of the matter is, no one really knows the right combination of carrots and sticks and nudges to control all problems associated with drug use, and our policy responses often look like a game of “whack-a-mole.” Encouraging an effective culture of “learning by doing” would be a powerful legacy for future generations.