It's unusual to watch one of the world's most powerful editors in scientific publishing play with a marionette puppet.
But Dr. Fiona Godlee, editor of the BMJ, specializes in the unexpected.
The puppet she's holding is dressed as a doctor, complete with a stethoscope around its neck. Its strings represent the hidden hand of the pharmaceutical industry.
Godlee keeps it on her desk to remind her of the dark forces at work in science and medicine. And she is blunt about the results.
"I think we have to call it what it is. It is the corruption of the scientific process."
There are increasing concerns these days about scientific misconduct. Hundreds of papers are being pulled from the scientific record, for falsified data, for plagiarism, and for a variety of other reasons that are often never explained.
Sometimes it's an honest mistake. But it's estimated that 70 per cent of the retractions are based on some form of scientific misconduct.
"Medicine and science are run by human beings, so there will always be crooks," Godlee says.
There will be commercial pressures, academic pressures, and to pretend otherwise is absurd. So we have to have many more mechanisms, much more skepticism, and much more willingness to challenge."
As the editor of one of the oldest and most influential medical journals, Godlee is leading several campaigns to change the way science is reported, including opening up data for other scientists to review, and digging up data from old and abandoned trials for a second look.
She has strong words about the overuse of drugs, and the influence of industry on the types of questions that scientists ask, and the conclusions that are drawn from the evidence.
"It's not my job to be popular, I'm very clear about that," she says from her office in the historic British Medical Association building in central London.
"She's taken her licks, as it were, because other people don't like the level of transparency she is bringing to the process," says medical writer Dr. Ivan Oransky, who writes about flawed science on his blog Retraction Watch.
Based in New York City, Retraction Watch is fascinating reading for anyone interested in what goes on behind science's closed doors.
Every day there are one or two new examples of research that has been quietly withdrawn.
"People leak us things, people send us documents, we get reports from universities that aren't supposed to see the light of day," Oransky says.
"There does remain a really entrenched problem with institutions, when asked to investigate allegations of misconduct. They will tend to close down, will tend to prefer not to investigate, will tend to hide any evidence and see it as a damage to their own reputation if they were to take action," Godlee says.
So retractions are, paradoxically, a good thing.
"I think this trend toward journal retraction is a positive sign against what we've known to be going on for quite a long time," Godlee says.
Godlee admires Oransky's work, although they've never met.
"It's doing a good and important job," she said. "It's doing more than retractions, it's looking at misconduct in research."
In that sense, Godlee says, they are on the same page. But Godlee says the journals themselves are part of the problem.
It is up to the journals to decide what science gets published, and they usually choose positive findings. That means a study showing that a treatment or theory doesn't work rarely makes it into a high-profile journal.
It's called "publication bias" and it distorts the scientific record.
"All along the way, the system tends to encourage a sort of optimistic positive view of new drugs and drug treatments generally," Godlee says
Her solution? Transparency. Throw open the windows, let everyone see everything.
"I do have a belief in the fundamentality of science to correct itself. We can't do that under the blanket of secrecy," she says.
"We also need to have more independence in science, less commercial bias, less ability of academics to follow their own biases. All sorts of checks and balances of that sort. But in the end, transparency, to me, seems like the only correct route."
Her policy is already changing the scientific record.
Just last week, the BMJ published the results of a second look at a long abandoned clinical trial testing the hypothesis that a diet high in unsaturated oil would reduce heart disease and death.
The new conclusion? Not only did corn oil not improve health, the data also showed a higher risk in death from the high corn oil diet.
Two years earlier, the BMJ published an analysis of another lost trial, by the same team. After digging the data out of a box in an old garage, they came to a similar conclusion about the effect of a so-called "healthy" oil on health.
And there was the re-analysis of Study 329, a controversial clinical trial into the use of the antidepressant Paxil to treat teenage depression. The new findings contradicted the original industry-funded researchers, concluding that the drug wasn't safe and didn't work.
It took a court case to get access to the hidden Paxil data, which was protected by corporate secrecy. And that raises another controversial question about who should be testing drugs in the first place.
"It's led me and others to increasingly question the idea that the manufacturer of the drug could ever be considered the right people to evaluate its effectiveness and safety," Godlee says.
"That seems to me to be very mad idea which has grown up historically, and we have to start questioning it and we have to come up with alternatives, which would mean independent studies done by independent bodies."
And it matters, Godlee says, because bad science can be dangerous.
"Patients do get hurt. Drugs that shouldn't be available are available. Drugs with harms are used and patients are unaware of those harms. Devices that shouldn't be on the market are on the market. So yes, we do know that patients are harmed, and we know that the health systems are harmed as a result of poor science."
