Would you put your head in the jaws of a lion? Not keen? What if the lion was old, tame, most of its fangs had long-ago fallen out – and by doing so you could help save the world?
This, essentially, is the conundrum faced by Oxford scientists pursuing a Covid-19 vaccine. With just 1 in every 1900 people (at latest ONS estimates) infected, it's impossible to be sure if those who have been given a trial vaccine are not getting the disease because they are protected, or because they just haven't run into the virus.
You can try to get around this with trials in countries, like Brazil, where infection rates are higher. But the way to be really sure is with a human challenge, which is a nice way of saying: "Deliberately infect people".
Adrian Hill, director of the Jenner Institute at Oxford University, wants to go down this route with fit, youthful volunteers. Sarah Gilbert, leading the vaccine hunt at the same institute, reportedly doesn't. Understandably so.
There is still a huge amount we don't know about this coronavirus and, given that it only recently emerged, its long-term effects top that list.
Problems may not be obvious either: scans have revealed lung damage in some who don't report grave symptoms. A powerful residual threat may linger even in a disease that, for the young at least, appears to be more pussycat than maneater.
The truth is that we have known for months that such a situation might arise. In early June I spoke to Adam French, lab director at hVivo, a spin-out from Queen Mary University which, in Whitechapel, runs the biggest quarantine clinic in Europe, precisely for the challenge trials Hill is talking about.
Even back then, according to French, it was clear that low infection rates meant a challenge trial would be the best way to test any vaccine's efficacy. He said then that his team were "working with partners and government to see if it could be offered for Covid" – either with a weakened version of the virus or the strongest volunteers.
But there was one caveat: they would need "true data that we have a treatment we can give people".
In other words, if the Covid lion unexpectedly began to close its jaws, the ringmasters behind the challenge trial wanted a whip to crack to stave off disaster. Even at the sharpest end of medical research, where risk/benefit is etched into the ethics code, it is nice to have a safety net.
But what if there isn't one? What if there is no other practical way round this or another impediment in the vaccine development process? What margins of error, or failure, or risk, are we prepared to accept in limited numbers undergoing trials, given the global need?
The development of Gilbert's own vaccine, after all, has moved through various stages. It was tested on mice first, but many humans in trials afterwards developed headaches. Each stage in learning that the vaccine is safe to administer faces unknowns, risks. The same applies to learning whether it actually works.
This is not about being impetuous or cavalier. Quite the reverse. Think of the astonishing feat of sending men to the moon in 1969. Think of the innumerable checklists, fail-safes, and back-up systems that made it possible. In the end, though, they had to put it all to the test and blast off.
Scientific advance has been a victim of its own success. We turn to scientists as we turn to Google – for clear-cut "truths", perhaps where none exist.
You can see it on the TV news: "For the answer to this difficult question, we now turn to XX man or woman in a white coat".
But we forget that science is not about answers. It is the difficult, tedious, foggy and, yes, sometimes dangerous process of asking questions, hoping against hope to edge closer to resolving them.
Twelve of us have walked on the Moon, a triumph not just of human technological progress but also of the courage and adventure in human nature.
If 24 people today, choosing to face the uncertainty, the danger, the risk, but also the glory, ventured into a lab in Whitechapel to deliver us from Covid, would that not also be something to celebrate?