A few minutes into the movie "Traffic," in a Washington, D.C., cocktail party scene, an amiable red-haired man offers some wisdom about the nation's drug problem: "You'll never solve this on the supply side."

The speaker is William Weld, the former governor of Massachusetts. I was taken aback by his comment--not because Weld was acting in a movie but because I had never heard him reject a supply-side solution to anything.

But if you've seen "Traffic," you know that Weld was merely expressing the idea that is the movie's central theme: that the U.S. war on drugs is a failure, and that it has failed mostly because the supply can never be effectively cut off, no matter how much money and blood go into the effort. Weld says it; so do the film's fictional characters. There's Eddie Ruiz, the mid-level dealer who agrees to become an informer after he is caught by federal agents. "It's an unbeatable market force," he warns the agents.

There's also Caroline Wakefield, the 16-year-old addict whose father happens to be the nation's newly chosen drug czar. "For someone my age," Caroline says, "it's a lot easier to get drugs than alcohol."

And finally there's Steven Soderbergh, the movie's director. He insists he didn't make the movie in order to change drug policy, but he also thinks that the time is ripe for change, and that the movie will help the process along. "I feel absolutely that it's in the air right now," Soderbergh told a reporter recently, as "Traffic" neared the $60 million box-office mark after just six weeks.

He's right about the timing. Strange things are happening in the drug-policy realm that seemed politically impossible just a short time ago. Nine states have now passed ballot measures legalizing the use of marijuana for medical purposes. New York's Republican governor, George Pataki, has proposed lightening up on the state's draconian drug- sentencing laws. The recently departed national drug czar, General Barry McCaffrey, urges that we stop calling the anti-drug campaign a "war," suggesting the remark of the fictional czar in "Traffic," played by Michael Douglas: "I don't know how you wage war on your own family."

But more important than any of these is Proposition 36, which was passed overwhelmingly by California voters in November. Proposition 36, which goes into effect this July, essentially abolishes prison sentences for non-violent drug users. It authorizes $120 million to be spent across the state on treatment programs. Some estimates are that it may divert as many as 30,000 prisoners a year from the state's correctional system. This new law has not received an enormous amount of publicity, even within the state--California has other problems to worry about these days--but it represents a major experiment, sure to be repeated elsewhere in the country if it appears to produce real results.

Is the war on drugs the abject failure that its many critics claim it to be? I've never been sure. A while ago I sat down with a stack of reports from the Drug Enforcement Administration to see if they served up any truths glaring enough to be evident to an untrained (and on this issue, pretty much agnostic) reader.

Actually, I did learn a few things pretty quickly. One is that it's hard to make a plausible case that the war on drugs has done anything to cut off the supply of cheap product on the streets of American cities. The street price of heroin, according to DEA figures, is barely one-fifth of what it was 20 years ago, and the potency of what's available is several times greater. When the teenager in "Traffic" says she can find drugs more easily than liquor, she is almost certainly telling the truth. Former FBI director William Webster, after hearing hours of testimony on drug traffic a couple of years ago before his Commission on Federal Law Enforcement, reported that he was "not aware of any evidence that the flow of narcotics into the United States has been reduced."

That's not to say production hasn't been affected. We're taking out huge chunks of productive capacity all over Latin America, as DEA figures make clear. In 1998, U.S. agents seized 742 metric tons of marijuana coming in from Mexico alone. That's compared with 102 tons just seven years earlier. All told, the federal government now spends about $18 billion a year on the drug war, and two-thirds of that money goes to interdiction of supply. You can't say they aren't finding stuff.

What you can say is that it hardly matters how much they find-- there's more coming in from somewhere else. Critics call this the balloon effect: destroy the marijuana fields in Mexico, or the cocaine crop in Bolivia, and Colombia picks up the slack. Clear the opium poppies out of Afghanistan, and they sprout up overnight in Pakistan. The end user never notices any difference--not even an increase in price.

I don't know what this says to you. What it says to me is that we aren't so much losing the war on drugs as we are fighting it on the wrong front. For years the experts have been arguing back and forth about whether the proper strategy is to go after supply or demand. I look at the numbers and conclude the question is at least partially settled. Weld and Soderbergh would seem to be right. Attacking supply doesn't work.

Of course, that's not the official position of the U.S. government. Just last summer, Congress approved $1.3 billion in new aid to Colombia to help eradicate the cocaine crop. Among other things, this money will pay for 63 new helicopter gunships to use against the drug lords. Not many people realize it, but Colombia is now the third- leading recipient of U.S. foreign aid, after Israel and Egypt. On the evidence, that does not look like money well spent.

But if the failed war against supply suggests a quick verdict that the entire anti-drug effort is a fiasco, other facts point in a different direction. It's undeniable that there is less use of illegal substances in this country now than before the war on drugs began. In 1979, 25.4 million Americans reported that they were current users-- "current" meaning sometime within the previous 30 days. Last year, that number was 14.8 million. Among teenagers between 12 and 17, the percentage of current users was 16.3 percent in 1979. Now it's 10.9 percent. The DEA says there are 3.7 million cocaine users in this country. That's disturbing. But in 1982 there were 10.4 million.

Some of this, I realize, is just a function of demographics. Due to the ebb and flow of the American birth rate, there are far fewer people in the high-risk 18- to 25-year-old age cohort today than there were 20 years ago. And it's true that use among teenagers, after declining steadily into the mid-1990s, has started going back up again. But even taking all this into account, it still seems reasonable to suggest that, over the life of the anti-drug effort, education and treatment programs have generated a significant improvement. Demand control, unlike supply control, does not look hopeless.

That's the point that the current crop of reformers keeps trying to make. At the end of Traffic, Caroline Wakefield is enrolled in a treatment program with her parents there to support her, vowing that she will make it and showing every sign of a rapid return to mainstream life. That scene is no accident. "Education and treatment," director Soderbergh insists, "pay off like gangbusters."

But I'm a little skeptical. Let's say we accept the DEA figure of 3.7 million cocaine users. That group by its nature includes individuals at all stages of the addiction cycle: new users just experimenting with the drug; intense addicts with few goals other than finding money to pay for their habit; and late-stage users looking desperately for a way out. It's reasonable to suppose that better treatment programs will capture a larger percentage of those at the end of the road who are ready for help. But it's hard to see why they would have much effect on the earlier-stage users who don't yet want to be treated. This was a powerful argument when James Q. Wilson first made it more than 20 years ago; I haven't heard anybody answer it effectively.

Of course, there's another item lurking near the top of the drug- reform agenda, and that's legalization (or at least decriminalization) of marijuana. Two governors support this, several big-city mayors have suggested it, and the number seems certain to grow.

It's an idea that has to be taken seriously. The long-term effect of marijuana on human health remains an unanswered question. But while the issue is being debated, the government is spending billions of dollars destroying crops and chasing down and prosecuting distributors. Of the roughly 400,000 people in prison on drug-related charges in the United States right now, a significant proportion got there through their involvement in the marijuana trade. Decriminalizing marijuana would also decriminalize tens of thousands of non-violent offenders every year and save the entire criminal justice system billions of dollars and untold grief.

Still, it's pretty clear that, in the current legal and social environment, marijuana does lead to the use of more dangerous drugs. Kids start out smoking marijuana, as Caroline Wakefield does in "Traffic," and then escalate their use upward. Now this could be mostly because marijuana is illegal. Repeal the prohibition, it can be argued, and the "forbidden fruit" connection to cocaine and heroin might vanish. Marijuana might turn out to have no more of a link to hard drugs than beer does.

Then again, maybe that's not true. Maybe there's something in the chemical composition of marijuana that would make it a pre-cursor to hard-drug addiction for a substantial percentage of users, even if it were as legal as Budweiser. We really don't know. It would be comforting to have more data on this before we make any bold leaps toward legalization.

Sometimes I think that marijuana could be the ideal subject for public policy devolution. Legalize marijuana in one state, track the use of cocaine and heroin in that state over five or 10 years, and then use the data to make an informed decision about just how dangerous a substance marijuana is.

The state that volunteered for this experiment would, in all likelihood, be making a major contribution to the long-term social health of the nation. I have to confess something, though: I'd just as soon raise my family somewhere else.