The great race for a COVID-19 vaccine has more than 130 medicines in development, with 40 being tested on humans, of which 10 are in large, phase 3 trials. The U.S. Government has invested about $11 billion in Operation Warp Speed, making advance purchases from Moderna, Johnson & Johnson, Pfizer/BioNTech, Sanofi/GlaxoSmithKline, Novavax and AstraZeneca — betting that at least some will be soon approved by regulators as “safe and effective” vaccines.

No matter your Twitter feed, “vaccines have been one of the greatest public health tools to prevent disease,” as The New York Times explained in January. But as late as May 9, the Times reported that “American officials and pharmaceutical executives have said that a (COVID-19) vaccine remains at least 12 to 18 months away.”

That forecast may prove unduly pessimistic: Good Judgment, a respected forecasting firm, places odds of a vaccine (approved and distributed to at least 25 million Americans by March 31, 2021), at 46%.

(And on Friday, Pfizer said it would not apply for emergency authorization of its vaccine before mid-November.)

Too fast? Many are terrified that the Food and Drug Administration may hastily authorize injections into hundreds of millions. The FDA and drugmakers are trying to assuage such concerns with enhanced commitments to safety.

But the demand that new vaccines must be safe at all costs is itself dangerous, and the strange bedfellow of anti-vaxxer protesters.

Pulitzer Prize-winning journalist Laurie Garrett inadvertently quantifies the problem. In a Sept. 3 article in Foreign Policy, she cited the H1N1 (swine flu) episode in 2009 as “the last mad rush to vaccinate.” Warning that those shots “caused Guillain-Barr (GBS) paralysis in ... 6.2 per 10 million patients who received the vaccine,” she argues that phase 3 trials for COVID-19 vaccines, typically involving just 30,000 people, provide little protection. “There’s no way ... we can spot a safety hazard that’s in 1 out of a million, much less 1 out of 10 million, vaccine recipients.” The “safety side,” she told a TV interviewer, “looks insane.”

But, in fact, the “insanity” here is not found in the push for speed or in Garrett’s skepticism about Operation Warp Speed. It lies in a lack of balance between the two. An insufficiently vetted vaccine may cost innocent lives, but so will delaying a vaccine that, on net, saves them.

COVID-19 now costs over 700 lives a day in the U.S. — 30 per hour. Would reducing the toll even by just one-third compensate for the possibility that as many as 205 people might be paralyzed (if every Americans were vaccinated and GBS spread at 6.2 per 10 million)? Presumably, yes — in less than a day.

Insane to move forward rapidly — or insane not to?

Indeed, the 2009 H1N1 inoculation episode offered by Garrett is pro-vaccine. In the article she cites (a 2010 paper in Neurology Reviews), it is noted that GBS was more associated with the seasonal flu shot (10.6 per 10 million), and H1N1 vaccines were still recommended. “James J. Sejvar, M.D., a neurologist and epidemiologist at the CDC, emphasized, ‘The potential risks associated with the H1N1 vaccine are far outweighed by the benefits provided by the vaccine.’”

Now, limiting the lethality of COVID-19 is of crucial importance for society’s health and welfare. Widespread unemployment, school and small business closings, and social isolation are inflicting vast damage, slamming (in particular) the old and the poor. Beyond America’s borders more than 4,000 people are dying each day. The World Economic Forum gauges global coronavirus costs at $8 trillion to $16 trillion.

When promising therapies appear, reducing time to market is often worth the risk — as reflected in a raft of pre-COVID-19 policies, including the FDA’s “emergency use authorizations,” “fast track” drug approvals and “compassionate use” permissions for experimental drugs. In phase 3 trials, independent monitors observe results, and trials may be terminated when pre-specified benefits appear. Patients in the control group become eligible for the treatment instead of the placebo. Larger samples would enhance scientific knowledge, but as probabilities shift regulators act on the reality that the ideal can become the enemy of the good.

Thomas Hazlett is the Hugh H. Macaulay Endowed Professor of Economics at Clemson University in South Carolina.