The New York Times reports on a remarkable trial of a new gene-based therapy for leukemia, which reprograms the patient's own immune system to fight the rogue cells. The results have been nothing short of remarkable:

Three adults with chronic leukemia treated at the University of Pennsylvania have also had complete remissions, with no signs of disease; two of them have been well for more than two years, said Dr. David Porter. Four adults improved but did not have full remissions, and one was treated too recently to evaluate. A child improved and then relapsed. In two adults, the treatment did not work at all. The Pennsylvania researchers were presenting their results on Sunday and Monday in Atlanta at a meeting of the American Society of Hematology.

"Despite the mixed results," the story goes on, "cancer experts not involved with the research say it has tremendous promise, because even in this early phase of testing it has worked in seemingly hopeless cases.

"Mixed results?" These are fantastic results. Twelve people were treated for end-stage, multi-drug resistant leukemia; one third are in full remission. More than one third have at least slowed the progress of their disease. These are people who had one foot, and the toes on the other foot, in the grave. Even if this is only a temporary respite, it's a major success.

It strikes me that medical research is haunted by the memory of penicillin, and the other antibiotics that immediately followed. For a period of ten or fifteen years, "miracle cures" were the stuff of everyday life rather than television movies and late night infomercials: you took a pill, and something that had previously been fatal, like pneumonia or tuberculosis, simply went away. Often, you went from death's door to the picture of health in hours or days.

As a consequence, people think that this is how medical discoveries are supposed to work: you find a cure for a fatal or crippling disease, everyone gets better instantly, and the world is a better place. I suspect that this is Big Pharma critic Marcia Angell's mental model. The reason she's decided that drug companies are about as useful as a third buttock is that they're no longer delivering the miracle pills on schedule.

But this is not how innovation works most of the time. Innovation is not "Problem solved forever!" Instead it's mostly "We've reduced the incidence of this problem by 11.3%." But if you do that over and over again, eventually you can produce a pretty good fascimile of the miracle cure you thought you were supposed to get in the first place.

Unfortunately, Penicillin Envy can make it difficult to achieve those lesser innovations. A while back, Michael Mandel pointed out that the FDA frequently makes the perfect the enemy of the "needs improvement"--and thereby kills promising technologies. A company that developed a computer system for diagnosing melanoma was turned down on the grounds that it didn't do as good a job as the best dermatologists. Of course, not everyone has access to the best dermatologists--and more importantly, there's no reason to think that the first-generation system will do as good a job as the eighth generation. But if you kill the first generation, you also kill the eighth.

The FDA eventually reversed itself, but the mentality remains. We expect to many miracle cures, and as a result, we get fewer medical miracles.