For centuries, scientists have been confounded by one of the messiest aspects of the human experience - diarrhoea, and why exactly we have to endure it.
While it makes sense that its purpose would be to flush out infections as quickly as possible, there's been a surprising lack of biological evidence to back up this assumption.
In 2017, researchers waded into the science of diarrhoea to find that we might be doing more harm than good by trying to prevent it.
"The hypothesis that diarrhoea clears intestinal pathogens has been debated for centuries," says one of the team, Jerrold Turner from the Brigham and Women's Hospital in Boston.
"Its impact on the progression of intestinal infections remains poorly understood. We sought to define the role of diarrhoea, and to see if preventing it might actually delay pathogen clearance and prolong disease."
Turner and his team were interested in one key aspect of the diarrhoea experience - how does all that water get past the intestinal wall and into the final product?
Working with mouse models of 'traveller's diarrhoea' - the type caused by bacterial infections - the researchers examined the animals' intestinal lining to figure out how large amounts of water are able to pass through to facilitate diarrhoea.
The mice were infected with Citrobacter rodentium, the mouse equivalent of an E. coli infection.
The team discovered that in response to the infection, immune cells started gravitating towards the intestinal wall, and triggered the production of a protein called interleukin-22.
Interleukin-22 fused with cells in the intestinal lining, and prompted them to produce a second protein, called claudin-2.
This protein's job is to coordinate with the cells to form large openings in the intestinal wall, so more water than usual can start flushing through.
This all happened within just two days of the bacterial infection - well before inflammation and tissue damage start to make the infection more obvious.
The activity of claudin-2 has been noted in humans before, but these are the first signs of the interplay between it and interleukin-22 in the event of a bacterial infection.
And it suggests that while having diarrhoea might be a nightmare, not having it could be an even worse fate.
Below you can see the interaction between interleukin-22 (blue sphere) and claudin-2 to fight the C rodentium bacteria:
It's not just the mechanism by which diarrhoea is more complicated than we've assumed - we've been misjudging how it interacts with the bacteria to fight the infection too.
The team figured this out by observing three types of mice - one that was engineered to overproduce claudin-2, one whose ability to produce claudin-2 was blocked (called claudin-2 knockout mice), and a control group.
As expected, the control group got diarrhoea in response to the infection, and the claudin-2 over-producer had diarrhoea all the time.
But none suffered as much as the claudin-2 knockout mice.
Despite the fact that they were the only ones to avoid a messy bout of runny poop, they experienced symptoms of the infection far more severely than the others, and it took much longer for their immune systems to clear out the bacteria.
"Their immune system went crazy trying to clear this infection [but] it couldn't," Turner told Ryan F. Mandelbaum at Gizmodo.
In the early stages of the infection, the claudin-2 knockout mice experienced far greater tissue damage than the other two groups, and greater proliferation of the harmful bacteria.
And here's the kicker - they ended up getting terrible diarrhoea anyway.
By day 11 of the infection, the diarrhoea finally set in, and remained "significantly greater" than the other two groups up to day 21.
The researchers suggest that if the body isn't able to produce diarrhoea to flush the infection out - facilitated by the activity of both interleukin-22 and claudin-2 - it will start breaking down its own intestinal lining to force the water through and bring on the flow itself.
It's important to note that the experiment has so far only been carried out in mice, so until we observe it in humans, it's too soon to say that the same process occurs in our bodies.
But humans do produce the protein interleukin-22, so it might well be responsible for triggering the production of diarrhoea in us too.
And what this research shows is that any medications designed to block its or claudin-2's activity could be doing more harm than good - we've been underestimating just how important it could be in preventing more serious and prolonged infections.
"People are working on drugs to prevent diarrhoea. But this tells us that ... you have to be careful about it," Turner told Gizmodo.
"In the case of this pathogen, if you block it, you make the infection much worse."
The research has been published in Cell Host and Microbe.
A version of this article was originally published in June 2017.